<%BANNER%>

Nonlinear Modeling, Identification, and Compensation for Frictional Disturbances

xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101119_AAAAAG INGEST_TIME 2010-11-19T06:06:32Z PACKAGE UFE0014262_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 76874 DFID F20101119_AAACWD ORIGIN DEPOSITOR PATH makkar_c_Page_049.jp2 GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
6eb99f96042095f8b98aead2c7607412
SHA-1
1cb5ec8d6ca0951705a7bb68576c849b8fcac49b
1053954 F20101119_AAADAV makkar_c_Page_072.tif
d714cee2e715f0f40b06b7f604380102
39ca750c0c50a94b1f098205044fdce6d9884dfb
555895 F20101119_AAACVP makkar_c_Page_035.jp2
1163585dd743272a9405b2f09b1b31c5
d9b657d799f89e0fa75f5f434d23087b37bef4fb
F20101119_AAADBK makkar_c_Page_087.tif
90130349fa7966c00290b131c28b3a9c
2dd18976c9a811ad89a652c8a24885e2efe27b75
783445 F20101119_AAACWE makkar_c_Page_050.jp2
6da87a0c5283fd81ff15379f34478823
ac5feefede49a2d29665fea3b2fdb14746b4edd1
F20101119_AAADAW makkar_c_Page_073.tif
a4302dcca228a9fd2625b28cdc5d5678
28917c8f58fde3f29e6a44c5e9d9d8b6e4332002
105827 F20101119_AAACVQ makkar_c_Page_036.jp2
d2dd8a7881a6ef3a9e53e140ab52ee1e
3ee5944597c18337eabc9f30ea92e4285522332d
F20101119_AAADBL makkar_c_Page_088.tif
6c99e029fa1711c8d306613bd86ef143
9c654ba2b86f7bc855db4e4221c63338e67e9aaf
67781 F20101119_AAACWF makkar_c_Page_051.jp2
8e7b746595bf9fa47ae5d3b71dd04594
f8cbbd741eaf24379666edc059e93b0b633ff3a6
F20101119_AAADAX makkar_c_Page_074.tif
b97ee13b2487a77d18ee419984460a53
2f3672bc0d22acca47ad6f3aae9f208828081848
1051986 F20101119_AAACVR makkar_c_Page_037.jp2
3015961d3e8b6bfd602120cab634df51
0210251af26a237198abc2e50f22ba8f62aeb40b
F20101119_AAADCA makkar_c_Page_105.tif
84ab8d94ab60cfe50a046ce243f4b323
a7e48e973823c84a2f1fb5c287f7972a2ec35b9c
F20101119_AAADBM makkar_c_Page_089.tif
794fffbcca0713d0a6ac23fe4ef75f05
3e9f6ccd886423c53831031cb127512671f83ddf
603308 F20101119_AAACWG makkar_c_Page_052.jp2
f0143265c42456db3e91f4cb47c602f1
b0efeb06455539423a6304d8bcdb0cb377bca0f2
F20101119_AAADAY makkar_c_Page_075.tif
b2ac61b2bccc9f631404cf4ae1a691ba
7b8175edd376ff2fafee60ab621cb9f46b9ef444
737496 F20101119_AAACVS makkar_c_Page_038.jp2
a8093686a979b764e1b1aec755d3c269
3f73b64d13d48c9abc7dded344fcc9c5fed4ac14
F20101119_AAADCB makkar_c_Page_107.tif
7fabccac8102f9db6ceb8aefe24efa25
6e499d4d3dfe74509cfc49f136da7c63a3b0d695
676739 F20101119_AAACWH makkar_c_Page_053.jp2
492a5b45abe949b589b1ba930d6d4f24
3343910b3304476ecfd2a1486ab64434fadf3bdb
F20101119_AAADAZ makkar_c_Page_076.tif
c5a255200865cd0e5b91e35ca6b88493
5ea09c43a2341d8650bbaedc4e4cfe46033839d9
20101 F20101119_AAACVT makkar_c_Page_039.jp2
227509e664a248753253224f7ec4805e
5adbe7bded34b74e905a4895ed5784d2f8ecf420
8004 F20101119_AAADCC makkar_c_Page_001.pro
4256453a29b3128257e1cb8f0f49ab71
4b32b6536986e764ec4fb5e6aede9b55a16a3b07
F20101119_AAADBN makkar_c_Page_090.tif
22f350c9a82f31a31b55da3eb83aa9e1
05425494a4f53f12020a369b8051f279b646e15a
676610 F20101119_AAACWI makkar_c_Page_054.jp2
916813fc2912d8e63e209cad7c28312d
0a363e87f6cb7936b7bc5a463ec56b7abc2f2ffa
104690 F20101119_AAACVU makkar_c_Page_040.jp2
eb0190db44ed7c0c4a4300bf48f58a7d
482f73b756d37de0ccf88eeb0150c433d411e8fa
1097 F20101119_AAADCD makkar_c_Page_002.pro
8d4630bebe77ad20e44b646218755f02
ce2a8f6ff8ffccc0fe61d98390c409f4ff70ba9d
F20101119_AAADBO makkar_c_Page_092.tif
b70a0a08f99a5530194b8123244e266f
a26794e7ad7df5ecfe56d2c9975e8e0475951391
572740 F20101119_AAACWJ makkar_c_Page_055.jp2
765d83b3b919a2e353f3e0c5b46e8d7f
10f417b4bdfbf7b64635d5206dc2d1d0d1520c21
100104 F20101119_AAACVV makkar_c_Page_041.jp2
ca2dc21b31d8fa7ecaf047278037a627
87d6bb7514dfdc067a80fd061a249bf9a482ad61
3266 F20101119_AAADCE makkar_c_Page_003.pro
e554e0abecb2186cb356b64a836d7b6e
d5e48e91ef7bcc8d2d6f458a3b68185858b02d8e
F20101119_AAADBP makkar_c_Page_093.tif
ea9b65c8fb151a9610080361871af35f
12fe015a9ae995ef72426816cd628d7e95c3e8a2
618235 F20101119_AAACWK makkar_c_Page_056.jp2
713e0ba26326560058bb85bae2310c8f
f0d022ca9812ec9228f3301b03909645836ac250
80524 F20101119_AAACVW makkar_c_Page_042.jp2
fd4cc955f0fdb11f9653b2292d2c1590
caea5168d7ed93bbaacdb8f0b6a1c12deb1e5ad7
33043 F20101119_AAADCF makkar_c_Page_004.pro
8f38243f7df2eb0b490eba44ba3045d5
c82e708d5d5fa9fb2bc513d55e982eb8257c2a22
F20101119_AAADBQ makkar_c_Page_094.tif
761b7e9c40975f37538fcca9b2e0ca30
a8efdbd541f9dd48f164a6299e0feea140f6be8c
949320 F20101119_AAACWL makkar_c_Page_057.jp2
5464da63ba466161d160568d684d5730
0ea59d855daf6253ee1f8577ce3fd2fe146a3b08
77547 F20101119_AAACVX makkar_c_Page_043.jp2
cdfc8cdbcdfbcb4548c47e144d1e2216
71a0d2408464c0c25666b8781cfee78d78afe809
43851 F20101119_AAADCG makkar_c_Page_005.pro
d0b824c431d0bdc5f67858ea6ac29281
2274f73aa5b6b177a500831738c18cbc19c94a53
84317 F20101119_AAACXA makkar_c_Page_072.jp2
2e65c0fb0c6d6edc62fdd2a08962d6ef
3f42f516c2100a82e6a70dbfc808339a03f0e1b6
F20101119_AAADBR makkar_c_Page_095.tif
0b6e850f1181fb6176f9fe6b67402ba2
bc22c6cc06b1fc9698efcc4482655d0a53684d00
638576 F20101119_AAACWM makkar_c_Page_058.jp2
c0949e51c23efb05459137a9d8c1b983
6025d45b670ca1d03c2f5fab3a452a49c0d73214
69704 F20101119_AAACVY makkar_c_Page_044.jp2
fb1600567948bfaa1188b486a0ca24d0
cfc2a07fdf722a4fdb44d82e8c8d3470095b8c1a
18657 F20101119_AAADCH makkar_c_Page_006.pro
f834376d5c8a0596869eefba672c5776
ede02e08ad3a43167fdfc1b04dfc04b9b0a14511
F20101119_AAADBS makkar_c_Page_096.tif
b8a14969fac90a9a95f1e31a84444729
4d08af96e9d92c15a7ee64dd0ff6dedd87188871
797551 F20101119_AAACWN makkar_c_Page_059.jp2
b46a2c09602d7376f2c170c7b218ba17
c4026bcc6193979cc624f6d6bf1e2a69627f089b
64496 F20101119_AAACVZ makkar_c_Page_045.jp2
b53c27137f8a7726deb2cdf5c3e3459d
5c2769fecd11366d28f9f6fc7cc0be30e0dfaeed
9822 F20101119_AAADCI makkar_c_Page_007.pro
02814bfa45b5e9335d7ae4b16da25d26
bc57cfdedb8f4d5d6c992608cb3c5c56d109af0f
86858 F20101119_AAACXB makkar_c_Page_073.jp2
cfabb03323d4b0f3733ed61d502fc6ad
37431eb8deb4bb6df35ea66c0b0ce24b021d5c2f
F20101119_AAADBT makkar_c_Page_097.tif
87f51ad815ce7a8f70ef1d60d7976542
5a3feb97632d7f91f09b2dda2e5cd069072f76e2
625663 F20101119_AAACWO makkar_c_Page_060.jp2
da55a669645bf2f0440c553e4b83517c
265853b5112a3e9d6f2b11a291beb5a75e984a4b
52606 F20101119_AAADCJ makkar_c_Page_008.pro
02475eaf33a6e4b9fb7cb6dacad83f0f
1b6a5969a82ce27b0b5d3885a3beccc73105d545
85048 F20101119_AAACXC makkar_c_Page_074.jp2
60be2eaba2242605a7706bdf9e3939e0
bc4267351139e6af165adc8f3158a0ed388860c2
F20101119_AAADBU makkar_c_Page_099.tif
32bac6432a445bf31e78615d1a92eeb6
dd79c6615d3f8f4ebe772c92c07c5a8b2818d5c2
566036 F20101119_AAACWP makkar_c_Page_061.jp2
bc12fc1b10a51731bbed729acdbbf5a0
15a0de48b27382d04a8d30adb12c90f13fb5932c
62939 F20101119_AAADCK makkar_c_Page_009.pro
f12056ca1f4342d8d9552aa674d2b9c4
40c36f14bb9f67a4de6f947048d0d7c92c06d735
75222 F20101119_AAACXD makkar_c_Page_075.jp2
3f48ced702a08fd70409c8a52dc9b6fc
915f4787969230225ba0639dda1a948dae4b8cc0
F20101119_AAADBV makkar_c_Page_100.tif
1aefd4dd3f536e4165470106d68ce3d9
0c9916ce079617c71c4eb0d5d4f691bf92a59992
634162 F20101119_AAACWQ makkar_c_Page_062.jp2
8092d3a86822f60599ed4981808bea07
1eab9d3cfe8e21970e1b47ca71381bedc949312e
68491 F20101119_AAACXE makkar_c_Page_076.jp2
ba19df0e8b777974370f58ee80686ca1
775e9c72a39cc387311081a615cb6d6a520f31dd
F20101119_AAADBW makkar_c_Page_101.tif
b594e71aa2dc68ab78b23a25bb60f7e1
dbeefc79c958c3199431ecedb0c5d500d583b915
975966 F20101119_AAACWR makkar_c_Page_063.jp2
caf00709f5db1c1139d15e1606749c3d
dee2795b68454f1a49e3bdda77b9a3f7e7b9f9f7
37371 F20101119_AAADDA makkar_c_Page_027.pro
c7fca4ecc7a678fa9b60e12d6a07c650
bb5fc22ec7360290edd993e2d014d74d328156a9
61364 F20101119_AAADCL makkar_c_Page_010.pro
91621009cd7010056f4b91e229223f8e
673c45c953dd51585cb47bf74a6fd4056976c43a
67530 F20101119_AAACXF makkar_c_Page_077.jp2
7daba8e729e0c1bb97473c87080dec0a
50f29d0ac3c95343576e7cfc7e366f4492a1377c
F20101119_AAADBX makkar_c_Page_102.tif
87174af81c90a7e9ece8acacfa08a7df
7653555ac180c688ada23eff17cb30f61f359d50
773624 F20101119_AAACWS makkar_c_Page_064.jp2
9331be720fc7b14002b60e1e1361800a
077d12f183d15c53c607929c1edef4eb3bdc7717
7713 F20101119_AAADDB makkar_c_Page_028.pro
819c7dbbd2f79f4877c99ce6715a37ee
c4244b280ad9d18e88c9c26d438e2c4d76318cd7
43974 F20101119_AAADCM makkar_c_Page_011.pro
e1e3df29ca40086cff7cc57065b76497
553ed4c141b207c6c942bbd1abcbd83020cb950d
73864 F20101119_AAACXG makkar_c_Page_078.jp2
362666402eaface971e89610f47c9142
10dcae9c5f1f5f44cf20beb60e99114aa1732a11
F20101119_AAADBY makkar_c_Page_103.tif
9db26ba7c037dca29e47c899fe86bb76
fb796780c45460e3ce32fd13953b6d6196a26eee
859336 F20101119_AAACWT makkar_c_Page_065.jp2
771e79de28b25080c27145aba3484043
034c0f887e693aeabb086864cc12a43143b1fdbd
6198 F20101119_AAADDC makkar_c_Page_029.pro
3e4f785d8b6a0c8ea7d32913dc5142d1
dacdf781a9ac652da54ad801df8fa1999de8a391
41289 F20101119_AAADCN makkar_c_Page_012.pro
7fcee294a89a465f0aeb63e661baa087
ec8ef7b3f71152e08dac0ec406e3b1a1504e350d
94612 F20101119_AAACXH makkar_c_Page_079.jp2
d4d8b38e5616c28f731f9183fd4d8aad
5b5e2b4a84c47abbc68070c2355bacdfd05c5f05
F20101119_AAADBZ makkar_c_Page_104.tif
c5580f6770c5705cf4cf48752b4b30f4
17a4db81c93d083828fcd009bb675fcf928e52d1
108977 F20101119_AAACWU makkar_c_Page_066.jp2
a2aff83504a866011b9c58a4b91d5633
f457b0d3d65b4ff431043fdcb7497d708147b0a4
7906 F20101119_AAADDD makkar_c_Page_030.pro
842e3087a9e0faca597129862b23ee4c
cef7a3044b1b3019c370bbb65ade67bea5e1673b
42214 F20101119_AAADCO makkar_c_Page_013.pro
839ece0ef45930496c885a8b1e1e4593
e67eaa0390e053ca2c32cd9f73eaab173ec1d6be
1051973 F20101119_AAACXI makkar_c_Page_080.jp2
dc1ad0e32e456063eedaaa4e15657699
4b1ca4b3caec6f887f164878bf19b9f787e2e171
115349 F20101119_AAACWV makkar_c_Page_067.jp2
24f78b2d563c26b55e2fcb9dbf1e333f
08d84d081e75b4975714e533668194e18f774e2b
5800 F20101119_AAADDE makkar_c_Page_032.pro
7e183f1439349d2255f02d08dad634ce
252e342ecfcf3152f5aada4192631b8845081298
57949 F20101119_AAADCP makkar_c_Page_016.pro
8057b170e181fede0235fda5a30fa7ad
78a15ccdd0d29f72cf90586ff637cdc9014e81c8
71172 F20101119_AAACXJ makkar_c_Page_081.jp2
902069470976efcb9416689fc5933abe
51c5d19abb5285f4f0db450ef6db06e646fda07a
1051945 F20101119_AAACWW makkar_c_Page_068.jp2
7c72835f1bc1f4caad07eac6922e5799
d4a5265b6d39e7c11556cf001fe7fd1087b4642b
5913 F20101119_AAADDF makkar_c_Page_033.pro
346a353de66e89fd323dc633afd6d6f6
09b060881ee28efd5a95757c1385dafc65c18480
57497 F20101119_AAADCQ makkar_c_Page_017.pro
89f6689f5aa970f614e969168f01db3d
f916735ed5c31d654e5d12bd7a7c7ceab27e5b47
786990 F20101119_AAACXK makkar_c_Page_082.jp2
a4e110e053fe78738abc88bd7b09a5ed
bc8c8d18b60f5ab08fed7496bee5d2c134bbbe34
376573 F20101119_AAACWX makkar_c_Page_069.jp2
9ee5a46ee3aebebf84a58c227e65583a
11cc0e07ab9e6a6543cc70d5dba562a7e3a82de1
7130 F20101119_AAADDG makkar_c_Page_034.pro
3651e3a2a766b147196d9aafc24cb517
abda8dbcc90ce9e247c06b71afc5adb529dc6380
42773 F20101119_AAACYA makkar_c_Page_100.jp2
7786fbdf17eff0e94b2733e4db8d5ec8
e8e02d7c2faa7c3477b718a0297504ebb32522b5
55697 F20101119_AAADCR makkar_c_Page_018.pro
0280e17408ecd6b716d029f6586724a5
e4c173abe20d5dc917f9d332541a992c5605d8a8
666038 F20101119_AAACXL makkar_c_Page_083.jp2
81d6ac7167f4c1b0e45dfd7e2b58679f
63f6255723225bd9c6b5e7bc22c8c64483fcb920
100798 F20101119_AAACWY makkar_c_Page_070.jp2
099d0396e741f7c9f4d28d08f5696f55
4873068db2ff142f3a218b18a8b842e3f45a60cc
5809 F20101119_AAADDH makkar_c_Page_035.pro
f870dacbff83f3834a715d881a38165e
75b7cc964bf5727de3f22e9785ad2884afbcd5fc
26451 F20101119_AAACYB makkar_c_Page_101.jp2
05694aead0938bb732ed7ed2c252de37
d974adae5a1a65f95fdc684c0c5e2d048aa0de5b
55603 F20101119_AAADCS makkar_c_Page_019.pro
79cdd5f5997ea82f6a382317604d28d7
e505bcfb4686fe242300929a7ae475febf1d9a00
637315 F20101119_AAACXM makkar_c_Page_085.jp2
10f1a2d9ab6f43dc9b75ede98580778c
d5d8fd5d89dc42f04ac0583d0df0f3a6090e7627
96982 F20101119_AAACWZ makkar_c_Page_071.jp2
2ac91c2f8503a78a70321a7f2b2e6637
633a183d0cd96b16cd61d3e808b20f213fe028f5
49922 F20101119_AAADDI makkar_c_Page_036.pro
4185a798e49b3257670c321fc4d74e2d
32c58e2ed786bd434f26222aeea52239fa2ca3e6
9873 F20101119_AAADCT makkar_c_Page_020.pro
feaae151a194b88282c34b8e0aa260c0
3e5c53102e3338b6b19a06248e5353e43d6e9e5d
54401 F20101119_AAACXN makkar_c_Page_086.jp2
22a7ec39963ef2e54d784e835ffd3337
aafa42867b35f97630e7876db2ad4bd7778a3edc
32171 F20101119_AAADDJ makkar_c_Page_037.pro
dff667f7c23edded088c908bfdc799d0
bcac60a6b2fc3f4bb791a9390231f2e32d801708
119827 F20101119_AAACYC makkar_c_Page_102.jp2
6ba9cded3ea91b649a25ec305e8b69ed
5f0d2e50b05ddd0259d6ea4e076e42dfc86596e2
50221 F20101119_AAADCU makkar_c_Page_021.pro
fae6e5541d71eb8336abb3bd5c6ed8c6
3c371c50fd1389d734e10f899460ba86cfa122a0
107509 F20101119_AAACXO makkar_c_Page_087.jp2
4aa1e77393ec3e67677256cf91d393a8
d69567bab5bbbdb3428b0675aa33fda5d217bed3
17350 F20101119_AAADDK makkar_c_Page_038.pro
837aa7356a0f86206f7590e3fa8a5ddd
fa9b3ba579ce152b4fe884008671b095a30f7d24
135411 F20101119_AAACYD makkar_c_Page_103.jp2
c100b6786782f8307c2740e19bd42884
f604823963aa0978d53ac7302939bddb1da137ac
43809 F20101119_AAADCV makkar_c_Page_022.pro
caf75fca1615e8afdeb312d4aa03aac7
83c225e48f6461d3c12fac1ae690fb5524a90924
86939 F20101119_AAACXP makkar_c_Page_088.jp2
d4705cbe387444cdcc7840a738be1bfb
44a5ab8e49f4d57cf57ef1bd7e1650b12d6ab7d6
8140 F20101119_AAADDL makkar_c_Page_039.pro
6caf2535d41692c417f9c30afefd4474
f19449780628f424347912a8aaefc53f7f2cbcd3
136798 F20101119_AAACYE makkar_c_Page_104.jp2
076623f2d033bbf4154f0848388e93e6
dca7fe3391257c17e9b6d68744cda3dc38333aea
41296 F20101119_AAADCW makkar_c_Page_023.pro
8d6718cdc8238fb6f696c92525e2f140
dec3803cf8e8a168063da627c8bc9f0700f8cc8d
50021 F20101119_AAACXQ makkar_c_Page_090.jp2
27cc18cad35adf0eae60b623ed082a0f
1c482c3832420592b46e0ca272470f33e036fb80
134782 F20101119_AAACYF makkar_c_Page_105.jp2
9d5018a064da3cc85e3a1bfbacbba599
fa4cc42a18e84a79e00c4a6d0256145e86567cfe
12690 F20101119_AAADCX makkar_c_Page_024.pro
0e2f81c150661113870a5b46b5b1ac12
fa1ef27653032c7710a0050665115f6e696b1c08
47590 F20101119_AAACXR makkar_c_Page_091.jp2
c05a304a7b2dc3f6b183be1679f8d000
e839c8174ed9480cf04181eea7297478c697c4e5
7778 F20101119_AAADEA makkar_c_Page_055.pro
a98e068fdc3e938e08ba5b54736b3cfb
fe8a95acf58ac15ab9901ee9d8e070d3025b36cb
49271 F20101119_AAADDM makkar_c_Page_040.pro
60ee0e6b3d5dead02edf43ffaa8525dd
8fa85d49316a2c7dbc300e4d7927eedd9c8d9e06
128320 F20101119_AAACYG makkar_c_Page_106.jp2
9d6cbecbf19258ffc5d1b9260a1211c0
d60fadafed688577ad1a107ed5f4d2da68348780
14682 F20101119_AAADCY makkar_c_Page_025.pro
2be48e35054744937b98a0ea3b30c249
55bf2ea86489f0cdf48e83d83ce37803464fc7ed
53599 F20101119_AAACXS makkar_c_Page_092.jp2
acfbfb3f44ecfb0c519daa82f9c72201
b04bc1a32ea52505d4cfe414d8bf48a33718f81b
19219 F20101119_AAADEB makkar_c_Page_056.pro
b52ac1bb01f86adab9e2f3453b88bc4a
89fad23f44539eb1baefa45d806416d3be06e4f1
46370 F20101119_AAADDN makkar_c_Page_041.pro
e6e4b1f501adda8bd8727bc06d78e1ab
a4e709bff175cf13ec34fefc9a3508579ce8b6d2
36382 F20101119_AAACYH makkar_c_Page_107.jp2
23d119ad23a3d9b96df029681e6b3247
3fcf458d5f3894c1b26352282d3a647e361da047
37175 F20101119_AAADCZ makkar_c_Page_026.pro
296e35cbde336c76895258c671fd105a
3cc9eee32b258e5170364871a203e50b7b7a34dc
57226 F20101119_AAACXT makkar_c_Page_093.jp2
501ab9d3e222abbef1eccbf8f51e4875
24577e856abc0a00288db2dc870e3918c3272362
32898 F20101119_AAADEC makkar_c_Page_057.pro
91500a45f0c1e28343dad04e90183790
56bd48b84451bae27f535d2b5c4d4b73eeb87149
37293 F20101119_AAADDO makkar_c_Page_042.pro
2dd606e7683862f1fd4ac1bb211e53fb
472e81d062745f18297ff2003ea17a9b12c117b7
F20101119_AAACYI makkar_c_Page_001.tif
18234537fd81ae8b474bcaede6bafdc8
4e544f29bc3272e09a94e9ef6110789d8b7e0d09
48301 F20101119_AAACXU makkar_c_Page_094.jp2
df7ed6b13c5954c22486ea89e199a3c9
1a07a51657d07e694084ed14dd9fc52e8f5fde73
9847 F20101119_AAADED makkar_c_Page_058.pro
1ef7b03b07bc6a077468b26007e32f21
539a1916166aca13ca8f9772b164eb1e0953ea98
35574 F20101119_AAADDP makkar_c_Page_043.pro
c64759121225ae17190cf834e342c513
23d4cd6fdebf08aefbdaaca67e45af29e3047664
F20101119_AAACYJ makkar_c_Page_002.tif
d9bc74ee5aad24728a79a19934183621
eeecfcbb95c70e511b40d01a481463b7e6d26bb5
56528 F20101119_AAACXV makkar_c_Page_095.jp2
2272322773ea5149e5ff76d4fa0fe186
bdf7f68474c5ef1bc2951a1a2090409531715cd1
14818 F20101119_AAADEE makkar_c_Page_059.pro
9726795fe7bced50c2392c4f38fd1af8
b699e3fa0023717a1b329ce772edb350791a4bf7
30788 F20101119_AAADDQ makkar_c_Page_044.pro
cc4ef899bd86c61850b436cf223371a6
32c57f92e5f0dfbfd68a30eb3a678e35e0537316
F20101119_AAACYK makkar_c_Page_003.tif
9181e1fe2d1b9885fc3c122d9abc0f41
3a45a659c84bc67c325a5e53dca5f02af3f033f4
54117 F20101119_AAACXW makkar_c_Page_096.jp2
e2d2d46fe58d84496ed5955e8936b6a8
c3d0debe4e7fefcf1617abbe516cea3cf3aeefa5
6947 F20101119_AAADEF makkar_c_Page_060.pro
f75324edcdaf83b594c9c57b8aee2752
2b699f9ca3783399aa032d1d2088c50b265be871
F20101119_AAACZA makkar_c_Page_020.tif
9337c4f0a2e6655417a63398254e7c2b
0859477e1755eada0a0b388d41a47592ae45b395
29076 F20101119_AAADDR makkar_c_Page_045.pro
d46a6ed11179f830dd7123bf93421ebb
e92ddaca3623c1e9de4712d5c743bfbbdb375692
F20101119_AAACYL makkar_c_Page_004.tif
67d836601891f13820ae0e58d19abb06
957d71b06186fca6ca2686bdef824543f2814153
66848 F20101119_AAACXX makkar_c_Page_097.jp2
af2a08d444a4fcf7324ddfbf42becc24
ac23d99ec4700b170e40175c9e8f1aa8531db528
7065 F20101119_AAADEG makkar_c_Page_061.pro
640906cfc96c2d9ab841a92efc470da5
25fcfcfffa7955b51fc023374fb5d6ea2494ac18
F20101119_AAACZB makkar_c_Page_021.tif
72a51558b22bee988cb20fbe0e85899e
6f2bae1836a9decc8c60943c23a427627050de6f
30375 F20101119_AAADDS makkar_c_Page_046.pro
4190ba6412b428380ecd94ff7059d39e
8289a008547d050ec04c12ca0746e2438bd89134
25271604 F20101119_AAACYM makkar_c_Page_005.tif
4d8d6f9a383da1e420a7cce0c836828c
ced10d909292e4b917e8a625052a4f10879ef088
42431 F20101119_AAACXY makkar_c_Page_098.jp2
9dbe1b652388ac5b4a314e1dc842f647
79d144064da417a0eddb17473dbc3262a64fb838
10358 F20101119_AAADEH makkar_c_Page_062.pro
8891c108d35ce8ecc02e0c6bede3d6c6
5bf6778ecee7305fd1d43882472666f6f7df120e
F20101119_AAACZC makkar_c_Page_022.tif
6f3d9fdab488b4f3587c09747cd1a413
9ac7a4d85a77a697ba102978328885ff6ab76868
39692 F20101119_AAADDT makkar_c_Page_047.pro
4b606cf2fb03681cec714da321457600
533834f9b3874cbff39ea4450c717401a34537d8
F20101119_AAACYN makkar_c_Page_006.tif
05164bc6d0c5eae3669a4c8873036034
0e022c8fc55a1fd496a5c7fbf10536ee8599bf82
47272 F20101119_AAACXZ makkar_c_Page_099.jp2
7fef69195ae311844afc84ba7ae42c56
a06a14c4b36a84562bedf03b6b34770b04c7c72e
37151 F20101119_AAADEI makkar_c_Page_063.pro
edd430e7b150cb8fb8e98ddf1e9749eb
a49f2b7e2bc5f98c87e24df449ebe4f0e35558cf
32185 F20101119_AAADDU makkar_c_Page_048.pro
ebd1777803f04312dd5a896390972c0e
0e0f6b1e68094c7d3ae8a8999992d4777fa424de
F20101119_AAACYO makkar_c_Page_007.tif
f7773fa69eca9cd639f6f57814178726
f0fb7b3fcd112d6e5eaee16cb42e97e7f6ceab18
14061 F20101119_AAADEJ makkar_c_Page_064.pro
7ece42d13267bfabf55c0104b8c503cd
fff80083ce560f34b56e133f7c493b4936beab7e
F20101119_AAACZD makkar_c_Page_023.tif
fa8213d2e9fdc8179cbef3b86a1db5f9
eee3f9093bc52db0dcaa44855beb9d84dedf3635
35718 F20101119_AAADDV makkar_c_Page_049.pro
29e3df348f79d333c0538901653064f7
a878d381fff14a92fbe85c70a79625fe1e658fd0
F20101119_AAACYP makkar_c_Page_008.tif
8a288caa72bd4cf66a82556b97eaa3ed
3443f0406a95e9ed04992ccc27312a97d234b85b
29699 F20101119_AAADEK makkar_c_Page_065.pro
2b49adf93637a368ddc44cdb6c2872c2
e4fbfc8c4cd0b302d7f274fd83e48a729b138850
F20101119_AAACZE makkar_c_Page_024.tif
d6b8db6f4708c76a4c98dd6efb139636
0b2e7bcdcdb984ca11bf1ba2a53ad505c137fdfb
27355 F20101119_AAADDW makkar_c_Page_050.pro
7127678f71318d51cb54b4d94f208eab
89a04bd925d5f40a2744f115071ab2722c6af902
F20101119_AAACYQ makkar_c_Page_009.tif
0a2637fc68c914cec3112b833be9fbe7
3fbdb52140f024db38b2a7517e1702a40b197dac
50296 F20101119_AAADEL makkar_c_Page_066.pro
a41d73a7aa4cbeb688ca6d010b11f507
899d5ab5bb50a8440207ec1be37385cf59c3b3b1
F20101119_AAACZF makkar_c_Page_025.tif
813ed5ac39fef074bc6abb24d8a9363a
9a100647f9c92770f6d92bb465ca33632a15e966
32068 F20101119_AAADDX makkar_c_Page_051.pro
888cf8aae7aa9ae839dae1b91d1266e0
9201af972681f2eb6345b879adfd6f74e575fb8f
F20101119_AAACYR makkar_c_Page_010.tif
6a79b3488398e98a059fe040d5c3f8d0
bba75fea819bf16941020cc84bb6fc1ea7fa900d
33222 F20101119_AAADFA makkar_c_Page_081.pro
8b962c1df523a340bb60fdc82bd52612
d90950c5a7a30455b1e370125d655e196ab12072
56401 F20101119_AAADEM makkar_c_Page_067.pro
dc3fbdf194366f63b11cfbe3fe8abfe6
2e29b41130af14b31260c532f47aa2808996bfac
F20101119_AAACZG makkar_c_Page_026.tif
f10e00f590838355f010692b7ecaf018
6515474d8955b39ddde74dd933537f9d50d6fd56
9565 F20101119_AAADDY makkar_c_Page_052.pro
4e98c34033a61353ae13db10e880eb62
6b863307aea212912dda844d76947a68a6232a34
F20101119_AAACYS makkar_c_Page_011.tif
37c1aaf7a98221fbc93973fc82b5597a
38c0b3179be381ef274a983a5f23d72d4416cb44
24560 F20101119_AAADFB makkar_c_Page_082.pro
8a7eda6ae9d6186e2a6f4f24896c1eaf
502f10537516c1479732799834b212481b88ff53
F20101119_AAACZH makkar_c_Page_027.tif
d06ce63dcb6c730a59dd9fd3de145b91
8cf9fd385c644d33c72bf79c4c2e7adae84f5435
9402 F20101119_AAADDZ makkar_c_Page_054.pro
8b3f056b9eaa65cb83fb16a8216a3f81
541b150e8872b7c9e0cb33deb324ac41edb5cc42
F20101119_AAACYT makkar_c_Page_012.tif
79b331b6f653001de881f76e3a79ba4d
9a0c58bba27374f6ead2fca6dbb1f54190aad8fe
7205 F20101119_AAADFC makkar_c_Page_083.pro
3bd87b8b136f4541249ea2064da0794c
6fe7ba5273b6389b2bbd55f97e18d76445e5500c
34841 F20101119_AAADEN makkar_c_Page_068.pro
d211626173249f59e496a7be9b7b935d
c89e4948a2273123b036d24149bf02e01e4ce23f
F20101119_AAACZI makkar_c_Page_028.tif
c0acf0debbb66da784ea50eaec2621a3
bbe3845755b7401e6bc68457cb0771c01eb1d568
F20101119_AAACYU makkar_c_Page_013.tif
e642ecc800e0c09bf4536e89c0ce3acb
c5e8b9c2951f182147ec246a0d18d634d8b6226a
33454 F20101119_AAADFD makkar_c_Page_084.pro
cb49e09be3351fab980d563e537babd9
8f7d00952b2619f414510e6d6c562551656b1d2c
8860 F20101119_AAADEO makkar_c_Page_069.pro
d433652663f49dec7bd89596a6a1775a
eb01a809d5a1bfb457d1b3c60e2b45ea9d7e53a2
F20101119_AAACZJ makkar_c_Page_029.tif
8e28029c924e0b5c6d3e350aa41cfb29
a5e27895cf34212f8362924ab7c23f5768097319
F20101119_AAACYV makkar_c_Page_014.tif
5038d16e42a5a6768d3770c62a21aded
74ba82796184f32d29c4f37b29498d5a4acf7270
9074 F20101119_AAADFE makkar_c_Page_085.pro
a92ffd036e65a616ce83f3cc9e0d5aa5
fa531fd6024d8883891d46f0d2ef3bb3c7e8099f
46565 F20101119_AAADEP makkar_c_Page_070.pro
b20bfb76ce8e6faa131be06c5843dde1
ae8cafa42e4459d7c948930cd6cde24694a5e71a
F20101119_AAACZK makkar_c_Page_030.tif
93260482d146e025ea14d67ac7368785
9840d5d51645f75706b45e74369df3f956afb5f0
F20101119_AAACYW makkar_c_Page_015.tif
9feb058e3d740223afe7b7dd57e70dd1
55b2ec6db915228a250a6150b3640c8f3abb7073
23972 F20101119_AAADFF makkar_c_Page_086.pro
55567a4a86518d1343b50154a42fb55e
1e61db024b3fccedab7de336bc053e00677f1c91
45698 F20101119_AAADEQ makkar_c_Page_071.pro
2f769a39c5465fbce387e92b047ab29c
836f55189560fbb7c8bfe9110201aa49e7b37c7a
F20101119_AAACZL makkar_c_Page_032.tif
1b31b3ddb94973264070e62ebdb7408f
cda5ea0c5b1e11f7b1bcbafb3bbd6715ed958576
F20101119_AAACYX makkar_c_Page_016.tif
367f9a5a8f29b5f9bc170dafae7d268b
4bb4969d98e0cdb7e50865833a1c8d832d8337d4
50793 F20101119_AAADFG makkar_c_Page_087.pro
2c6443359f0364b4338e6abae6a4c119
b227c4e2d091a341e20c8600217d7a2f56cd8b29
40126 F20101119_AAADER makkar_c_Page_072.pro
94c498369d7f95306fff1b0af13e7b87
d2ae4ba88c0ed40f0a600e2b30ac326f2ba36d83
F20101119_AAACZM makkar_c_Page_033.tif
18ddbaf4d7c605bc07fd2897e25431f5
d9a9f5f4fbac887fd50c00c1c6f08a10de579cbc
F20101119_AAACYY makkar_c_Page_017.tif
521179001c3e304a2bf96ec05bc18077
1f90bcfc8827372aa1f6e3c468904579a438da46
40160 F20101119_AAADFH makkar_c_Page_088.pro
b08d8d4d4ce0e01d38c3f85a9c5fc613
3e3e3795b63b80b2ea73c2315785e83f5aeb232e
38790 F20101119_AAADES makkar_c_Page_073.pro
eaf576e94c383ca828511de592c583d3
3c0b320a7ce6c822305a363b00999617b645e25b
F20101119_AAACYZ makkar_c_Page_018.tif
b5398856b86e6481947722015fc6e717
3cdc5fd2d300701ccad7b269973d6c111321eeac
30042 F20101119_AAADFI makkar_c_Page_089.pro
c75c3a59791f86da651fa5395235752e
bb66e5ad2b2e1a355de7a7f7d959bfe485d68799
37862 F20101119_AAADET makkar_c_Page_074.pro
c0c880ec3e6c3fe9bbc14af342db1bd8
c06a9cdc3ca689d4fdb47209bbd9e0e74166fdb4
F20101119_AAACZN makkar_c_Page_034.tif
f24d8eb0adb610e07ebe908cda4deb3f
9f12e264297dd585b8ba5f182828ce88906bba8f
21622 F20101119_AAADFJ makkar_c_Page_090.pro
2ba210d5b1b312878a16bbd2c0923c94
81c865f61de01c98550ba552f3f9d30046fb7056
37240 F20101119_AAADEU makkar_c_Page_075.pro
0283f9bacbe28af55fa7649da5465717
bd1f8b2d9099bb2dc9097c01971d0d6cf0b146ec
F20101119_AAACZO makkar_c_Page_035.tif
450edb3a21a835b16439bff87f4240c9
fa6e1e898f2ea3b5fd7ce900a7222c70cc6e1e56
22527 F20101119_AAADFK makkar_c_Page_091.pro
e3eab945b589cd82f486e88695778ced
ab1a5861a3704ac5f6fcda242aae2ec4e2742f4d
31050 F20101119_AAADEV makkar_c_Page_076.pro
296128108da3a1ba92443fae840a74a0
ef076e4a95417d49180cfc0c7fa8f3a8d503b4b6
F20101119_AAACZP makkar_c_Page_036.tif
eead8780a827066c6db6fcec8e305aae
221d735c70b12ff8690fe8f7808756e0d2a591a2
25603 F20101119_AAADFL makkar_c_Page_092.pro
87eef7db436eaafd0aa0035049eb75ce
0326a399baf6099fefd10ed8f51846c24a1f392c
30646 F20101119_AAADEW makkar_c_Page_077.pro
0d6d187e9056ac6ff53797cb369b2c97
5af5506f074b3a0bab36b4a7681adad09e135359
F20101119_AAACZQ makkar_c_Page_037.tif
2bc1cd20e670be1e6f2a2b8cbef3da8c
e6aa43ce9754a42a287a122ec6c9ad97090e104b
108 F20101119_AAADGA makkar_c_Page_002.txt
fdb82c9ef3874c46031271d891a93b37
d77607cfe4f38eed557f5ae08acc6793d94bea06
24104 F20101119_AAADFM makkar_c_Page_093.pro
57a7e586c22236326add4881f5870b78
fbaa5b6a4f6a9688d481f09bc4c78ead92db930f
33962 F20101119_AAADEX makkar_c_Page_078.pro
8c71e71f58e6de75c45fa80f00cd661e
9b1ce8afffb6f85e686f22bee653053fd0ea8917
F20101119_AAACZR makkar_c_Page_038.tif
3635d8357b2059dd880dac2ad974ecae
58cf8e3131b8cb775b3e2237fe12bbcf886314e8
177 F20101119_AAADGB makkar_c_Page_003.txt
087b6a0a6bf9ffaa6eb9797f9d12cc67
7e91127265bd8db81728e95660905a5788acf18b
20802 F20101119_AAADFN makkar_c_Page_094.pro
24d93809a410f0b785d8c881690cbce0
4c6e3f43ddf38940f01280e7cfba91062df621ed
43864 F20101119_AAADEY makkar_c_Page_079.pro
4cae6dd90138b3e42730f0e300ecd60c
b9f7799372cc466512fb743ebbfdea00f3ab45e5
F20101119_AAACZS makkar_c_Page_039.tif
c49492b7159b180fd62cdf111a260003
b8ce3159401c3bdab9bdee5672ac4d56318d91fe
1373 F20101119_AAADGC makkar_c_Page_004.txt
ebb9eeeeeb1b15647c8ca5041f101c99
8038ef49b7d3f3ae05812de842720e8772e6cad5
31005 F20101119_AAADEZ makkar_c_Page_080.pro
eb4f794ad2ae83ac3aebdb11e9d3e2d5
4411d199de5c99e0050aa0d47e9f352d3a4745f3
F20101119_AAACZT makkar_c_Page_040.tif
28b54818f6e01b17db7d6fcb450d477a
565e73865f4b75ded79581eeb22458d6ffa64930
1914 F20101119_AAADGD makkar_c_Page_005.txt
d2f2ed4a8a2ba9ec14e50598a81affe4
ac8b69e7a69bec8e15ab07ef173127be71d6dd89
21734 F20101119_AAADFO makkar_c_Page_095.pro
fbe2227039020cb3f17766f30111a8de
8cf57337d4c81a201a8da4c1cf40dccb1c855958
F20101119_AAACZU makkar_c_Page_041.tif
69c23031d2c7e1e1ba7785d5dd376a9d
dffe26d0057badfee58a5612a5f4b2594ce2230a
771 F20101119_AAADGE makkar_c_Page_006.txt
5e8690159451b379348e9e2873d3e4cf
843e75992745f0543b9b011ecb008e86e265b0d2
23781 F20101119_AAADFP makkar_c_Page_096.pro
a92876367df02ec235021c80715b2027
9dece688ea2059de8905e578a7988a469198ea3d
F20101119_AAACZV makkar_c_Page_043.tif
5d8ffd35e1efe5bdb3c9457486b51076
b4e4858f7fccbcc93d372b862edd942ada1d81c5
480 F20101119_AAADGF makkar_c_Page_007.txt
1400cdcc28d09504d6a3dfde2d707ab9
c9d935523adb030775eaff47abccb16b3423a3d6
35318 F20101119_AAADFQ makkar_c_Page_097.pro
ebf11c9eacde1b3ecdc8e7ca0271b992
f01e948b8a0dafda177180bd4bfaef97dd2663cc
F20101119_AAACZW makkar_c_Page_044.tif
ce5e542248cb8ccc70a5a462ae72a814
d9a09f3fde03156716bbb138731328a4c71a1fab
2584 F20101119_AAADGG makkar_c_Page_009.txt
149d4632eff559bdd824c240cd2090c2
6962767536ec85bbd5e6d66a6dc8dca94e086498
16600 F20101119_AAADFR makkar_c_Page_098.pro
50b901659865f6d4ada17ec7eb6379cb
2048e36d2b84f11304199c0fc846bd04a00070c7
F20101119_AAACZX makkar_c_Page_045.tif
698053401fd5e26d225fc9abbbe07e6d
8e4cf1395ad7290b785217928ca6e0379785cdc6
2501 F20101119_AAADGH makkar_c_Page_010.txt
a88189b2c766da08de0f1a5508c4d465
85e7e7ad3a8a1cb1c8dfb9df4b433df4d4e0e93f
19501 F20101119_AAADFS makkar_c_Page_099.pro
8b3ad294a6f5110ca40778b3109e300e
088f7bb9541d08bc404c0bc8d27bd905942d67e2
F20101119_AAACZY makkar_c_Page_046.tif
5f2a56f46b7e11833de48d66fad282ff
5f40f8394fa073fbc5a9d888c7272f28c08674bd
1909 F20101119_AAADGI makkar_c_Page_011.txt
cb61228e8aed26ae2f49cbc184f5790e
e02f50a6296a81f3c866cba48ebd31e259384aa3
10599 F20101119_AAADFT makkar_c_Page_101.pro
c3f596864d822cecdc7dc0bf49bb47a0
f0324c77c1d0078ca607eda97be9299c1a73976a
F20101119_AAACZZ makkar_c_Page_047.tif
0e6e88db9d92f6eec4cb0390eb3cb6bc
7e7e3f3def37401f55161986e38ece69cb802795
1644 F20101119_AAADGJ makkar_c_Page_012.txt
bb11d92fa7124736697be40961ef197f
858ef3412867f00805074684ff30750c50324ced
54689 F20101119_AAADFU makkar_c_Page_102.pro
f358b11ad5801e7cc661c9a6cde78330
a4438736f6340fb9642286352dbcd09d942c77a1
1771 F20101119_AAADGK makkar_c_Page_013.txt
26c0f0714bc24d5f0bce5472a12fce8e
d33262e754e135ec5bbb4685e0438faafbffc39c
63066 F20101119_AAADFV makkar_c_Page_103.pro
76d5a49d8b9691e4effa31e17becf7f2
e2e5adc88edf2f1fce7ab5aada20ef1834a0a870
63503 F20101119_AAADFW makkar_c_Page_104.pro
de86b9e7d7dab4bbfb35d05f94c0aad8
f6a0209db15a65d82ea8b3ca35a1eddabce16a16
2293 F20101119_AAADGL makkar_c_Page_014.txt
1e64a5db97bd01837ba7e354d138d0ad
b205d5bcec34a96766169573e79b6fc493ae18f0
61976 F20101119_AAADFX makkar_c_Page_105.pro
8d7160496c7938a217d9c44431efd6fb
5c7c712db8224e12da85d3547e3e18e7ac2d7abd
584 F20101119_AAADHA makkar_c_Page_030.txt
d05ca8fd9b6ee9f6cbf99de671c6c42c
11a56d537fcdf40a439f3c7968d65fd52a530dd9
2217 F20101119_AAADGM makkar_c_Page_015.txt
623e7c85df0ae88c2a7e08ef37ebe001
150af9146693193f299f7a037a24073a3b491177
58461 F20101119_AAADFY makkar_c_Page_106.pro
b58d7cf58b86284d83c560439c1733ad
e12268be7d5e84307cf279fb8010fef1be79ec94
388 F20101119_AAADHB makkar_c_Page_032.txt
6b9121fca4b39bade8fd4d72230c2513
7a6ec9569768d6afdc78a3b95ee5782081e10989
2271 F20101119_AAADGN makkar_c_Page_016.txt
f13db6a2c001c31e63725e4df6f14382
fc2ad558f50c0a760a22f32dc44b4ec7baa394a8
15056 F20101119_AAADFZ makkar_c_Page_107.pro
95f1f387d8af6946d89da07fe40cfeff
37dc04849fa904b59bf80831b74f740df946b393
457 F20101119_AAADHC makkar_c_Page_033.txt
112b5db062069145b1d71571a53b41a1
b402f2d2d4c3693b786c5dbed7297bab04206e3f
2290 F20101119_AAADGO makkar_c_Page_017.txt
47c32367997af320f4abbe57852eb405
f8a93d0aeb09302290b672e186e824e86b9c8818
476 F20101119_AAADHD makkar_c_Page_034.txt
1fcbe5398f1483f29dde820b56884589
31227c0ab622ff7c0b328ad5b0ffd016c0356b72
412 F20101119_AAADHE makkar_c_Page_035.txt
f480902e918bc6269e58465295edbb59
86c13964414a5dd26f79141a0163df9f7048f772
2188 F20101119_AAADGP makkar_c_Page_018.txt
b37d4c85236133608c520b6d06866a08
ec4477bffc1b819ecc3384ece71481a7545dc7aa
2005 F20101119_AAADHF makkar_c_Page_036.txt
3fbf2e799f54cd754663dde40c1bc38e
b223a3b966ced486be85510e48e9d00988330688
2184 F20101119_AAADGQ makkar_c_Page_019.txt
beac8e97bd7c96d35688e531cd4772d3
3e3edd88da8f5f8731daa230b948f80d039fc7a4
1335 F20101119_AAADHG makkar_c_Page_037.txt
44c02c34ea6e757b7ffa008e8a496191
dce8f89b85fba00c787e6d44c1211ece0ad8e8ba
434 F20101119_AAADGR makkar_c_Page_020.txt
a8df651935f15068bdeeb405fc9319f9
d8ed6d2a0793815547162fd9bb3e1f234232fb16
711 F20101119_AAADHH makkar_c_Page_038.txt
fcf11db5e4c4a5bcf1e6ad96b722f1c8
6aa33da3e2f69f56ae1d3b669f65ff2311ad3de5
2028 F20101119_AAADGS makkar_c_Page_021.txt
a4edba3b04c8d1aa29549333d0b1edad
13724c95c6cebbf4046611b80a796090acd35103
365 F20101119_AAADHI makkar_c_Page_039.txt
0a3f3c8acd3941bea2e7e45185fc46e1
451481ccea0203b58cb7f49f22f4fb2893306f19
1819 F20101119_AAADGT makkar_c_Page_022.txt
5837a575f967a2eb80786179005250bb
0b1957a493cfd97452ef6a6812f98c1760264c5d
2010 F20101119_AAADHJ makkar_c_Page_040.txt
55d6e0da63c72f317cdc564c13e63ef9
24d8a48e2ac8b9476445358ebcc429c0a874452d
1811 F20101119_AAADGU makkar_c_Page_023.txt
63296b7e7edc87ea1a0819e74a21b620
b212721b1d4391feb4f484c2a5738ad94cc1ce88
1717 F20101119_AAADHK makkar_c_Page_042.txt
287c6e08bdc5767e3eed72d2dfe4092b
a3faf55696d5f75040208e2096abc9446508ad31
665 F20101119_AAADGV makkar_c_Page_024.txt
b055a00129534d82f49176f711fb423d
8f33d7e39ecfa30ec80a7af1bd3f0d88184aa183
1448 F20101119_AAADHL makkar_c_Page_044.txt
b68f714d3e2f8f7fd1798a76e1e8d6d1
e24303de03fd56b81a314a43ce9b0eb22b43c80a
653 F20101119_AAADGW makkar_c_Page_025.txt
32152a875cf7658a9a823c6174148556
ce25b7970c0cad67fd110d4cb739bb611f74a470
441 F20101119_AAADIA makkar_c_Page_061.txt
a026c32f6d31c003c9547213854a9e7f
209f7f7d5efbc5ce5c35968a1f9d832e25eefc93
1545 F20101119_AAADHM makkar_c_Page_045.txt
ac5cf24c30eb0e9268deff160650b733
d9a4cc4426b4b7f641de7913d122e5afc38c9dab
1661 F20101119_AAADGX makkar_c_Page_027.txt
6eeda0b0f500846bd706a54534c81ecf
882d4707342393bde1879de4ad8fd2482765a891
766 F20101119_AAADIB makkar_c_Page_062.txt
9358f73b86d55c9135de1ef6a49cc64e
65d0bcdaf058739d5299505212f225a09e8c3375
1822 F20101119_AAADHN makkar_c_Page_047.txt
4f0a755aa80a965cc333b9fbed1cce14
a60db7883880b2d5e8f76f939fe24f83e122835b
623 F20101119_AAADGY makkar_c_Page_028.txt
0753df4daf99a25164b8025d49bdef7e
133f3ecbd82acaa9c021811a93a625684a452dad
1639 F20101119_AAADIC makkar_c_Page_063.txt
dc4672dd7c778d24297511ede524a052
5e158119d94a30baf7a99c3846c9b3040974574b
1378 F20101119_AAADHO makkar_c_Page_048.txt
726889a1771369c9a991d1a3f5968dcb
2fc72b98d003203e0b6c0a5c56717f6d9bb2a2fc
395 F20101119_AAADGZ makkar_c_Page_029.txt
f8899adbf4d5feb53321e61278548f26
688414a1b327f473430879a115291d63fde73149
862 F20101119_AAADID makkar_c_Page_064.txt
0f60e6e91e22545afa4b9798eb83d1cf
5436aa46dd99799640f9d15b3ae4cab7840136f4
1604 F20101119_AAADHP makkar_c_Page_049.txt
cc960eb365aebb27ea1e970ace9354d3
5ff4e1cf4384a1bb598382909218e36171270f65
1538 F20101119_AAADIE makkar_c_Page_065.txt
09f637e56dc25340a3ab6902ff495818
ab9905cd9f99ecb15e4d95ba28949ee0f30edba6
2241 F20101119_AAADIF makkar_c_Page_066.txt
be8f1d3c16cca46147b4209ac1a4898b
1f1b24520cc002a3542b7eead512ebf4dd60d490
1345 F20101119_AAADHQ makkar_c_Page_050.txt
8621af117e9b10e1dc4e6b64169962f9
1a9827923553dba0bdbb5d71a53bb9176b440e97
2222 F20101119_AAADIG makkar_c_Page_067.txt
5c252559c84baba457c243584aa842ce
a5a93a5e0525970c6f64b325d7fdad04ea7e878d
1586 F20101119_AAADHR makkar_c_Page_051.txt
0a06282e9f2931f9475994db024ef841
d6330cef0d5317e43f708f13d3041a06954e0c79
1423 F20101119_AAADIH makkar_c_Page_068.txt
a31eb2fea946d0a20bfbe2d7995b6a50
71bc20f8d19a2dbc2e5bc4ead822ad87ef996de0
723 F20101119_AAADHS makkar_c_Page_052.txt
dbfbf0f89ee97f28958b7924bd08ac70
cbd96a41741b5c8a6ce483b152007f36416c2dda
392 F20101119_AAADII makkar_c_Page_069.txt
40c26afd45f1e4b5b793ded03b7ff7c6
39adc5fc2e2a2b003006f406c47b28d62467d89d
1012 F20101119_AAADHT makkar_c_Page_053.txt
d16f44b237a6adce8e4e050596cbe287
a28fe2723f58ab8a753f3bc12e69c4cdf03a5ced
1910 F20101119_AAADIJ makkar_c_Page_070.txt
ce8a60fde7a2455620613cd2600e660e
0adaae76f03b9094915d48c723b5dd7aaf3d0b99
608 F20101119_AAADHU makkar_c_Page_054.txt
8ef0e2c40928de677ffdb2302b8e8057
259cf2397fda4eb94e5ede72e4b75c0913842def
1982 F20101119_AAADIK makkar_c_Page_071.txt
575cd4115de9809a5c0d09a074f7db45
baad21737bd3136e6f79e8b1c5224447e700f1b0
453 F20101119_AAADHV makkar_c_Page_055.txt
96dd275a830dc4e4eda671b64b93fb5d
8968999e5f6e313a35b09c9f0d114670f4deb311
1820 F20101119_AAADIL makkar_c_Page_072.txt
02956ca090c6f305819b2b6f78fd674c
b9e6b7d7199c5846b8861ac8d6ac2b7c1988627e
1534 F20101119_AAADHW makkar_c_Page_057.txt
01874cf17246ce94d17e41b9e329be39
984758fba284b41a7ffd35effa978f17bd57619f
1703 F20101119_AAADIM makkar_c_Page_073.txt
e488e6203883347c66457f89f2299d49
0a87e2e77b067a103d1bd373dabd66493b90cc54
914 F20101119_AAADHX makkar_c_Page_058.txt
4129e93ba76343f4cd0860493bd07212
0c32ed63b37188fe3544c7f6bdd59e6ef008eef0
1399 F20101119_AAADJA makkar_c_Page_089.txt
79f86a685ac02f0c83a9d11a8d397a55
299c2edea0fa68b4027315fe9da56e393f4e43e7
1762 F20101119_AAADIN makkar_c_Page_075.txt
db327ed887e41937ba0e1a93d936d1b3
0267211db6fe44421331a2440ecae7299d239aa6
1403 F20101119_AAADHY makkar_c_Page_059.txt
5df696097f73103554ae1c04450ee3fa
f366bc3d1c3590de90478ec465a809d447086da4
1396 F20101119_AAADJB makkar_c_Page_090.txt
0a17d3fe482b0eeb9839572249b7ae48
35edb2d4056a4863b2ff06b7861ce2d90861f3f9
1581 F20101119_AAADIO makkar_c_Page_076.txt
bbbfa2aefcd022923a01bc2baf6814e5
9a5bb456967405036d91d9360625c7d85d55fc84
341 F20101119_AAADHZ makkar_c_Page_060.txt
403dfc17f0e262f1696c6dcafa9fa949
1b600fb23dc122773bf5f7427392fd20ce0ad110
1263 F20101119_AAADJC makkar_c_Page_091.txt
e8509cda9f89c2e3c9347266b24d2c55
e9c8b1e845944ee28030b374e104c41fd428cb42
1593 F20101119_AAADIP makkar_c_Page_077.txt
0f045a25dc59cafbcc9c65926e813af9
4021412fa0e5dda66859284bf8544f5f0154f052
1203 F20101119_AAADJD makkar_c_Page_092.txt
1613c8fb3ee7edb11fdf5b9a458ae1fd
9a0632a1adb88b09d63f2625af422e7885799695
1599 F20101119_AAADIQ makkar_c_Page_078.txt
853046106f074254daa245578592051c
57cd886258f51f7cb3a043eee3b6c4f0df03734d
1366 F20101119_AAADJE makkar_c_Page_093.txt
435ee7e16d2e2e975f6c322f8c6de30e
ee20ca84317a277a673574118c2b5cb60db343c1
1167 F20101119_AAADJF makkar_c_Page_094.txt
b3015e1e7690511a4ac3173fa7f20eca
c45d65f24b1b259453d8455c97568e18904a21f2
1795 F20101119_AAADIR makkar_c_Page_079.txt
17d8196004a9be06489f14a1c30f8114
84c34f86b8dc0095c23f4a870e36b13ddc95e612
1187 F20101119_AAADJG makkar_c_Page_095.txt
0009d0d55f051299bb716a2e44614aa7
658921f1bc8f39a33b2a86e0a38e40af625dbaa6
1291 F20101119_AAADIS makkar_c_Page_080.txt
af6cbf31e826a57988c36820a88127cc
beee58858ab25bdd418eee8b22b0a78951a45215
1190 F20101119_AAADJH makkar_c_Page_096.txt
e75eb1354f7ba0ecc40d4b2912adc79d
66362cdaa0f7ae5b97c6f5eff6e77c0cf2c3ae72
1565 F20101119_AAADIT makkar_c_Page_081.txt
8d47829a6422bd408145d6d9ff93d0a0
bd5cc818a3d239fb9e3494bb45db41b160d5ddff
1907 F20101119_AAADJI makkar_c_Page_097.txt
d898665a4117634cd9f182d0f2a56382
f1fbee594a7c608aa6dfa7b8037e83c45cd066b5
1346 F20101119_AAADIU makkar_c_Page_082.txt
baad3889dcf457f74f314f62b1a6d880
5d83edec7e711f5570dd0fdbd8a45ad82bbfe6ad
985 F20101119_AAADJJ makkar_c_Page_098.txt
ab22ba4b7d06916f9043c18528fd1944
b0938f7f062be93bef823207340e4dc51f26406a
490 F20101119_AAADIV makkar_c_Page_083.txt
f81cd7517417c83673bd43a5f3333a01
5c631f4ec04daa5c476ec2d7b3b60ed4084439a5
981 F20101119_AAADJK makkar_c_Page_099.txt
e20d3dfeb51d70b670d6c948de58a988
e6aaabe5b7607e3ce55928eff3e3ea3bef8e51d9
1546 F20101119_AAADIW makkar_c_Page_084.txt
a6b2b07f5819175bb433164747c1371e
a58eb62f6c6e4021563dc2f9f5a0a94fe55d8322
1021 F20101119_AAADJL makkar_c_Page_100.txt
926cf4425708b788b461a1daeb76cfcb
149eaeaba11e2648794462fec12e41b4d1a1dd08
884 F20101119_AAADIX makkar_c_Page_085.txt
f471462676cc10da5b39458235e96dfc
e07f496f2c13a2bcffbf0d045c8801d11da01616
21150 F20101119_AAADKA makkar_c_Page_005.QC.jpg
88b3daa3066741d181f035d94b039ce1
302cea74029854f9290d2a95a027aae1d65f96ba
585 F20101119_AAADJM makkar_c_Page_101.txt
6f8fc03390a5f6dbac496a92e85003e3
af51208a588dac2470b862e047acdea705e8bcdd
2060 F20101119_AAADIY makkar_c_Page_087.txt
a56d04878da2cb73a6a8bee0628bb867
30b64ba03e9e441024deecbaa8a26216b3a7fc74
5291 F20101119_AAADKB makkar_c_Page_005thm.jpg
40d7fc1a7f2d1f77376d2454d103cd59
67210c9a8cf546131b60a5b976060cb8706053d1
2532 F20101119_AAADJN makkar_c_Page_103.txt
a14d664110e7c0b93b7f2bb064ccca3f
6fc3a3de7fa28652af8737c7ce140953a9d7f760
1611 F20101119_AAADIZ makkar_c_Page_088.txt
670b0a129e7a66125787d55c8c3a7648
f5a3822f7988ce8f95e0c1d168b9b9f6f1568e54
12295 F20101119_AAADKC makkar_c_Page_006.QC.jpg
706778f4a7fca53549ef0f8a533cf892
b3fa23f8b24e8ef3808b2d5705bb0f95c4cdb617
2550 F20101119_AAADJO makkar_c_Page_104.txt
6e0432850ab70fb363295f5c54b2c0f6
fa5a54b54747d3834cecfa8a7d4009809221b0ed
3426 F20101119_AAADKD makkar_c_Page_006thm.jpg
6f0f4cef91ad709036590fa42c45c6e5
c57106ddee0abd81ade0c9223f5dfd4b14e06041
2493 F20101119_AAADJP makkar_c_Page_105.txt
5ed15bf176009ebf91ab35a1a4b89583
2b34ad2c72a6178638031543d1e92c31b60c8671
5784 F20101119_AAADKE makkar_c_Page_007.QC.jpg
03b8820b3980a8c18011ea4e15e1ecb9
5ad904bf76f23565a53ebf1a2382c4b92e5b8a69
2363 F20101119_AAADJQ makkar_c_Page_106.txt
bb4353ea54d39ba2dc38c6b3334ad88c
dd84bf24f00658f63e7d8f36ac5e1c42b758d67e
2160 F20101119_AAADKF makkar_c_Page_007thm.jpg
0e5e7f1f1fd63a94d720ec7085d31e07
2f9ed8599d85b97193a8883118ac72b7984c563e
648 F20101119_AAADJR makkar_c_Page_107.txt
2fd218dbe31183e2626e3aa95b5cd520
7b9416373de2e1bb14c2ae8d4e095ea5201de7ca
20276 F20101119_AAADKG makkar_c_Page_008.QC.jpg
8e64e2dbe848929d3f71b48555282ca1
fb604690743c0f8da4d30c25a42ea7324f736c77
5385 F20101119_AAADKH makkar_c_Page_008thm.jpg
67685753f1c5c96da05c108402921b21
83dac8141daaeeee9db1597aacfda9b1a8593c53
2418 F20101119_AAADJS makkar_c_Page_001thm.jpg
0796a149c0868f6fe43b656207854533
71fccb22f12831e80af0805e1c121713adeac172
26330 F20101119_AAADKI makkar_c_Page_009.QC.jpg
2134148ed5a51dbc837678afe3a75404
71e3f052595723ee53f9d1032c402c78569915da
3446908 F20101119_AAADJT makkar_c.pdf
9e8c1f714e33fcac372ad37865112c26
2be1e4bbe87d8c415422f2281e65f8521454c781
6628 F20101119_AAADKJ makkar_c_Page_009thm.jpg
d5fb0385f38c1d658198a751f3557812
6701a868b1a099e995e3a153e70f758c69e46fe3
7109 F20101119_AAADJU makkar_c_Page_001.QC.jpg
c3f7f9b61f26d4da1d8b236131f9be14
d73d5efc9b6ab57ee410b704e0f3920b9c6f7d52
26246 F20101119_AAADKK makkar_c_Page_010.QC.jpg
907c3371bbd736e6a6ece8bee9f58faf
fd73b32d2330954636e66a7b11bce11cea9de743
3167 F20101119_AAADJV makkar_c_Page_002.QC.jpg
dc54f02f8050fd8a43b6decf22228385
f000c1449b1fb0a7013df3623d4931cb59bd11e0
6699 F20101119_AAADKL makkar_c_Page_010thm.jpg
e6e740025f615f5787ad1eb070ad17e0
8600f3d4dd9ec0f00547f6d7bf14cb1935ee95a7
1393 F20101119_AAADJW makkar_c_Page_002thm.jpg
ac09dcce5bb901135bc4c2112e7a8599
3b5e15fb83ce84fa8377fb3c2766392b76a11b9c
25641 F20101119_AAADLA makkar_c_Page_019.QC.jpg
4f0c75e2f3120722478e99e647d36f98
200b168de3ec820779ac80302a50e053513d67a4
20955 F20101119_AAADKM makkar_c_Page_011.QC.jpg
b9ef35c3b467ec4eae2aa4922516a4ac
45f3385c934bc60ca830bf6cfc7f6eb5573abc02
3954 F20101119_AAADJX makkar_c_Page_003.QC.jpg
b32de71349a4008b7baa97395ee320e2
420ce0a98a0e79f01150f149f022f3301cef5666
6487 F20101119_AAADLB makkar_c_Page_019thm.jpg
30bf9bec02b2588db7c020f9e2ca8dea
3439aa95f294270bc6277e344a3449fca7877fe0
5470 F20101119_AAADKN makkar_c_Page_011thm.jpg
3c43a0ab01bff99ea715f2be85a7f5fd
9cd54ed641c930d5698e65480ed8207ec6d1b5d9
16718 F20101119_AAADJY makkar_c_Page_004.QC.jpg
9042447bd5e41ef9854663b44b4238a5
52ddce2a8e21c52934cb968641064cf2218020d8
6752 F20101119_AAADLC makkar_c_Page_020.QC.jpg
9273fcdb1f5fdd16486054f7d8f15f36
03fa46fdd44ae4ee2ca4d12f0f59d64bddaa6d50
19646 F20101119_AAADKO makkar_c_Page_012.QC.jpg
636c9a3826785760c4713cfc5284ca6d
5f8b5e3f8e39e587008ebd1231f6428a43557445
4649 F20101119_AAADJZ makkar_c_Page_004thm.jpg
8ec3c9d7eeacc1fe4a6b8cafe15cd5fd
62abbe18654ae02950e5148aef18f239cf62df6c
2165 F20101119_AAADLD makkar_c_Page_020thm.jpg
38d8a791841f6e2faf1f5271ed59e9e8
e2420a79e85ff4e2f96fc298ab7841d54812ffec
19955 F20101119_AAADKP makkar_c_Page_013.QC.jpg
02b6ac4e30e0a1080050f7ae9f8c4050
5d44fa52a8f602bd643a173cd952dd595c25aa20
22712 F20101119_AAADLE makkar_c_Page_021.QC.jpg
5670d7de6e0274a64160c26279892bcf
476de828b738d47738d556e1c0e045ae35e01a10
5444 F20101119_AAADKQ makkar_c_Page_013thm.jpg
cfbc23519a335d24d731e67b3ae81243
d00945db6242930ec35157344c772dafaa516aab
5970 F20101119_AAADLF makkar_c_Page_021thm.jpg
b15626ff1863913635a035359a487d4c
db3076bec0829cfffb0cbbad50bbc71a2fdd6100
26846 F20101119_AAADKR makkar_c_Page_014.QC.jpg
5724130a8da003b44a1c25048f45d69e
c2e212f4ef4fdc7606fe85de57115e29521f44a2
5742 F20101119_AAADLG makkar_c_Page_022thm.jpg
de552d4ec096ff49277d43a11faed007
cbdc76b34d655fd1bd1c7b1a1d6501c5ee16ad49
25985 F20101119_AAADKS makkar_c_Page_015.QC.jpg
2f493961f057cc4b1d409cd0993beb58
1d64d29e5a9effa422a1f150f5b521657de33fc9
19530 F20101119_AAADLH makkar_c_Page_023.QC.jpg
21546a81b7156b2336c84a9a75b5fd58
a03f66808963b00e187694b1e8a68a9582af2e56
6521 F20101119_AAADKT makkar_c_Page_015thm.jpg
a36986ce93cf955ec0d83d9090bedf9e
e028ea44074f9e2571760bdbe06fb15b3058aa57
8158 F20101119_AAADLI makkar_c_Page_024.QC.jpg
11affb43ae8902f2e685f68212aeeb46
2c16470d3e89df163539767bd29c4a3649e61dca
26669 F20101119_AAADKU makkar_c_Page_016.QC.jpg
20efdf422efe94e1609875c81bedbba6
ad2611996d034c8d931a582b064d7b16f036ab3e
2929 F20101119_AAADLJ makkar_c_Page_024thm.jpg
d98da766270fc5174c61e540a700e982
ed9e1665ad257d506ead70cc1208b77ca82b2921
6674 F20101119_AAADKV makkar_c_Page_016thm.jpg
b3c828c3eb3eedb310a0ca3f9b19fa33
e85a2b7d30b892c1ac7195ed9cf89ff9cf54df72
10650 F20101119_AAADLK makkar_c_Page_025.QC.jpg
6dfe7a2b8e8bc2982814390cfcef78e3
52873e06cb7d78e04cbd5bc91a7702345bb36a6f
27018 F20101119_AAADKW makkar_c_Page_017.QC.jpg
75a78a71682ad174bc7913645ca609c7
e6d41ef7ba9fef9f32e47070e4dbe2322bfe2336
3497 F20101119_AAADLL makkar_c_Page_025thm.jpg
a62f4e8f6d8b9e5490851f8e30dc6879
dcfe92121108772aa0f6a97e845aadb90ce55615
6694 F20101119_AAADKX makkar_c_Page_017thm.jpg
21b98aa044ca0e959a5d2f9464c1778b
430317483c7ce09ad7cb1818e893e00d0ad2439e
9930 F20101119_AAADMA makkar_c_Page_034.QC.jpg
1821dea2440c6959ac7d6df1d9b1577d
a6068450c5840dd029bafa243c91901ae743737b
20246 F20101119_AAADLM makkar_c_Page_027.QC.jpg
b5e2acbe2b7e129e449509364c04f375
37ad32a6848a3b47dd5240765c13d03c8fd1524c
25863 F20101119_AAADKY makkar_c_Page_018.QC.jpg
e2296d5124f32cdebb1b7eddec14c767
85c689365f7aca23c65dc02e34fd9bd65803739e
3324 F20101119_AAADMB makkar_c_Page_034thm.jpg
cc83d57fb84f7f1cdc00857bf071a2a8
447349e1633bcb89771fc5df9e56409284f3d097
5641 F20101119_AAADLN makkar_c_Page_027thm.jpg
3d0a0fe8cd55a1241da80cfcf25cea77
89dd7740c8f58046304b50f064e67d949c7fb130
6561 F20101119_AAADKZ makkar_c_Page_018thm.jpg
c598557609f09534e70504915cdf0b28
a74ef43b3ca2f75be495034a04ef81b6b2d8cbc4
11494 F20101119_AAADMC makkar_c_Page_035.QC.jpg
064428bcca7936f335cf3d4c2835402a
d8c1e20829d5cfd7848af3fd2b6454f24e8be434
10618 F20101119_AAADLO makkar_c_Page_028.QC.jpg
8d705a9fc67658625cb2ae0726532a11
2e7a487ef761d20c768f1e94364098fb3f675427
3445 F20101119_AAADMD makkar_c_Page_035thm.jpg
03a09a4fb4cc77ef164531c7c248858a
a0d8afaa5ac21e16f37b90c82d35e84853ac8e58
3329 F20101119_AAADLP makkar_c_Page_028thm.jpg
567b2cfa2beb5d85402959b2db792ed1
c298ef496faad65a4cdad767b4dd86824e688887
23851 F20101119_AAADME makkar_c_Page_036.QC.jpg
0e179da1cda30162be4768d967017b14
9e45f14b9b28d5e2ee0ecb66155565479a7cc7cf
8041 F20101119_AAADLQ makkar_c_Page_029.QC.jpg
3dddd0ef495e1c404c75454f653b5401
90f202df6c556a2130f1a79f877f5acc810d377a
6228 F20101119_AAADMF makkar_c_Page_036thm.jpg
751d3c8c3a09362baf073e6b58644600
8c4a1016b2e20c90dcca22ed9fa56cfc5c5216c5
2679 F20101119_AAADLR makkar_c_Page_029thm.jpg
0f49b7da7f4647f816ff29ac3bdaa5dc
1fc0871dac608d8a7729d884be01a1985bd13200
22503 F20101119_AAADMG makkar_c_Page_037.QC.jpg
cc4be1c6ae1769bdf50f63b6008acfa4
f78b4e318946a74c4aa9bce8e26140b4811532b4
8609 F20101119_AAADLS makkar_c_Page_030.QC.jpg
e38b19a4dbfbd672ebbc9a43f8955db6
d0672ec056c9e677db3447da9216bf2a612e1187
6185 F20101119_AAADMH makkar_c_Page_037thm.jpg
80befe6c62bfebc419036492baf1a242
3959539bf3a33726f45a3dd4ae60849a3279435c
2816 F20101119_AAADLT makkar_c_Page_030thm.jpg
ddc4ae04c688a5abcb3df85034a490b3
8320d4159ad22d6aa44fda15eeb027460996defa
13789 F20101119_AAADMI makkar_c_Page_038.QC.jpg
93c6eb1543be3efd3989d560caabde4c
77848e60dafd49f6fb018bb21a6952a8ba2a6e1e
4127 F20101119_AAADMJ makkar_c_Page_038thm.jpg
2be37931a421ba922ae511091ded23c9
96a665e0f5cd3c0a394e0c237fb81e59cf9da1f6
19456 F20101119_AAADLU makkar_c_Page_031.QC.jpg
bb9e29ce5560efab8f86cac9111ac6ad
470e2ee3f940efe7d8aeb03a026bc144be72e793
6081 F20101119_AAADMK makkar_c_Page_039.QC.jpg
bbd288b7572e73b1e7b1e12e8e6ea94b
990ddcaae190dbe3e9f2320f438eb01e505f75c6
5268 F20101119_AAADLV makkar_c_Page_031thm.jpg
9e27415ae1b684891d6dec2b7357ba51
73993e46b60ed0855662830aa1d5b1bfe301a53b
1997 F20101119_AAADML makkar_c_Page_039thm.jpg
b60dba47a30c5bf5933f6d47e21986ca
b7414ff64bf1ee31db331686fd9142386c96f162
10261 F20101119_AAADLW makkar_c_Page_032.QC.jpg
264520f774aec106c2ce5c90e7864692
0e32f6e2ff5b9475e2794b31831100d49c5f6e4e
15628 F20101119_AAADNA makkar_c_Page_048.QC.jpg
ee80f6f0ea6c77a3b66b7cd87461501b
f60e510012b599e70f49788fc0b5b066ec83f2e7
23044 F20101119_AAADMM makkar_c_Page_040.QC.jpg
5b6e0eeb4f32af3681dd4fcbf38b3571
e695091e914f5ae67605318b7edad18c593122d1
3444 F20101119_AAADLX makkar_c_Page_032thm.jpg
9417095c0d76cbbee64ea2c48f96152f
5de9f2f6952b236897f24b2a10c2a75d4cf2184e
4802 F20101119_AAADNB makkar_c_Page_048thm.jpg
eb4b0a312f417a1c5e276a501787661b
19e15349dd06fa21ba574671372f4a322ba82020
6205 F20101119_AAADMN makkar_c_Page_040thm.jpg
65274a1d8c60780df313ea8d26f9f68a
e7c667945094946fd47f7b091405f16b55c8ad3b
12528 F20101119_AAADLY makkar_c_Page_033.QC.jpg
37dd9bfb8afd29bcd4b8aaf3ec29a8d7
6718a7a0eb9d62ba3c7088e27125bf60ae16de58
17259 F20101119_AAADNC makkar_c_Page_049.QC.jpg
87ee37364230c93c16cf00d91575e64c
13db491947489cd20208358d1963876cbcfe8550
22138 F20101119_AAADMO makkar_c_Page_041.QC.jpg
85eb8ed5f2b2acc5aee33362984a6b2f
bed148dac876f663c51467fb175d7867d4ea5bc9
3655 F20101119_AAADLZ makkar_c_Page_033thm.jpg
6c375562ece8b10f6979f785b306de11
00e2e7d9787a3f1646865b832a2c65034b995820
5043 F20101119_AAADND makkar_c_Page_049thm.jpg
509a1549f348b16eb148bb4a0daeb056
952aaeb6df95c755a88128a81eb0183d020c9029
5823 F20101119_AAADMP makkar_c_Page_041thm.jpg
dffd036e07a63361f9aeee3fd6573e8d
a5ff8a086f9b8b0ab57c1eb4d95a7e74e6165306
17170 F20101119_AAADNE makkar_c_Page_050.QC.jpg
b333a8192bc1bb6a36bb942bdfdaefe5
22019c9f341ae41db4568287eea64601a6ea5f00
18575 F20101119_AAADMQ makkar_c_Page_042.QC.jpg
5452d1912509317ffe47111ddc2c9359
b01af5dfe289440451ac68f0059960bdb14fed2b
5032 F20101119_AAADNF makkar_c_Page_050thm.jpg
5da707e61cda3603921f048604f12c8f
f82cbb04eab6fc2d55ab854537ea344d06f190b3
17002 F20101119_AAADMR makkar_c_Page_043.QC.jpg
ac7b97cc90653eaec9d1cc3d66cef553
064834f42bb5112e2cf8f942854bf945834a06ee
15985 F20101119_AAADNG makkar_c_Page_051.QC.jpg
0f80830795961dac40c8fb28309c1cfd
3f0dad324a9c407e9742fa36053aa1cf8cc9a220
5116 F20101119_AAADMS makkar_c_Page_043thm.jpg
ece55de32b348c0ecba59bb4acb89397
35c3e1a799769982a67bca8765e2106818b8cc00
12393 F20101119_AAADNH makkar_c_Page_052.QC.jpg
f4c2d7ba3ef0c1163c7e94d2afbe10c4
2bb356bee59a308a3b9a57930912084880636353
15433 F20101119_AAADMT makkar_c_Page_044.QC.jpg
db85d9245742473aa09c87d774d59c2a
4d3c068150271b7d7df4d02783bed2505e49e600
4290 F20101119_AAADNI makkar_c_Page_052thm.jpg
037e621029e6244704fe6c25175e4d64
fb754456b6df0e25832742b31fe3b7f99b64413b
4877 F20101119_AAADMU makkar_c_Page_044thm.jpg
c4ac2a2d659a3d25a0d50666750cd37d
4e347883dccce385cbcfa5680ec65a5887db0885
13686 F20101119_AAADNJ makkar_c_Page_053.QC.jpg
a23aefe2d2b075f1bd0060bf45375b43
c9cf9f96af9e59507481fb715574cc27a740c4de
4158 F20101119_AAADNK makkar_c_Page_053thm.jpg
e89f21b00969e7760f3ec8d856de9cc0
8b3dbeef287fa93283baed657190e93ebf9747ae
4531 F20101119_AAADMV makkar_c_Page_045thm.jpg
5d891055c2a5e3e20224960fd93dabc9
89bcd04481fc8f38d6a1980f582458b18e56db7a
11537 F20101119_AAADNL makkar_c_Page_054.QC.jpg
8a2649bf53fb9a6351397bd299fb52af
4c7b12bb989e5fed62f8623a7e27e5fea51334ec
15044 F20101119_AAADMW makkar_c_Page_046.QC.jpg
440c6a7cb503b61f9cc7336ecf219b9f
043ec00d7b5c62c2f0e7ce2b085c378c1acb758f
3976 F20101119_AAADNM makkar_c_Page_054thm.jpg
279589366118bbab9f2a2db208708163
00e89680aae32ca07026c7e6f04689c031dabf47
4986 F20101119_AAADMX makkar_c_Page_046thm.jpg
ab222575c63ab3d63727f70759b342e4
1272fbcdcd393b9972fc56cd314ff4fce67665f6
12724 F20101119_AAADOA makkar_c_Page_062.QC.jpg
a23ba5131c76a8147c9b1159dfb03903
79a72b10ccfffb68c29f557f24e436fc621d4bf3
12542 F20101119_AAADNN makkar_c_Page_055.QC.jpg
6121cc1441ac2c7bf7a7a2e2510cde0c
8b0e1ac6e3154c0e06d09e56129f6271352addd5
18775 F20101119_AAADMY makkar_c_Page_047.QC.jpg
59809011cc17b54f78b6ad8816ba19a5
90ef233f0bdcd29e3ecda80d55d1f3b87d66caa2
4481 F20101119_AAADOB makkar_c_Page_062thm.jpg
f4dbc0bfd7e58fde79ce431a67d2422b
febcf3d1906ee489df914fcdce174244a4c603e6
14630 F20101119_AAADNO makkar_c_Page_056.QC.jpg
ba58251a5f6a377d774431ac5ace7e15
a7e2e6f8f0955d9b5c504191a6fccfbc42b157b7
5398 F20101119_AAADMZ makkar_c_Page_047thm.jpg
3e0b46bdde9c61789eee34682aee618b
befff00e3f26ef1e4ef346ab314d2fdccb32044b
21479 F20101119_AAADOC makkar_c_Page_063.QC.jpg
a667244194b21d14577e53f212af579b
6c2a55b9d5a1cb8f0238286df326dc75d7eaa106
4712 F20101119_AAADNP makkar_c_Page_056thm.jpg
392264880406f2b5b032c3768de22a86
5b85ce03b7bbb89bdf8184396989c502b3ef12c7
6043 F20101119_AAADOD makkar_c_Page_063thm.jpg
b322c83315c78282dad0f083888997e1
c0df7edba584d50f8e1beeb4da958491305f9faa
20433 F20101119_AAADNQ makkar_c_Page_057.QC.jpg
590d6dc048e841d078a3e7800ac981f5
cf8e829f177c97e5761638af2adf8dd0d60845ba
15385 F20101119_AAADOE makkar_c_Page_064.QC.jpg
4317dd610ba408152190cd63467403c3
6f5c81700219711a01f69b4e35448f5ba1849c86
F20101119_AAADNR makkar_c_Page_057thm.jpg
f2d7fb383027fdb1ba95475013524596
f66a63d1bd12e9ac67dacea9f0d8f7f5cc039997
4677 F20101119_AAADOF makkar_c_Page_064thm.jpg
9503303f130fc0faa0a3364942be3542
fadd449a2834888cfec7eda17610798dbaf8f136
12249 F20101119_AAADNS makkar_c_Page_058.QC.jpg
c380a0e2dece98320bfbd254563048b3
12850d773664c7a476e9d8134dd7369ebe1f9aee
17454 F20101119_AAADOG makkar_c_Page_065.QC.jpg
c71819c4492c6a077d5a534580e3ec95
d3824bc00beca1cf5cc711e58dbb4651ff8f9d2d
4053 F20101119_AAADNT makkar_c_Page_058thm.jpg
79ed31c7d211dc593e27807e5f2f54de
c00e874cfd8438c4c2eb574a6eda9314a96b9c49
5279 F20101119_AAADOH makkar_c_Page_065thm.jpg
37781bae217e03cab9baf8dd833aad28
d2178fd974bfd45a323e914a33c8a4147350e914
13634 F20101119_AAADNU makkar_c_Page_059.QC.jpg
096bbbc6ea15ca156a23367cd1165a90
59db71086f65c0c6263fad017e2dc7add3929a4f
24590 F20101119_AAADOI makkar_c_Page_066.QC.jpg
ba7531462b530868ff4436068b1779ab
3385894db81528d1369ed5c85e903590d60287a6
4463 F20101119_AAADNV makkar_c_Page_059thm.jpg
7c1a567bb6c440681bd29851f925a96d
f06382d132b32e8fb3c281d958baf282141ff3d6
6234 F20101119_AAADOJ makkar_c_Page_066thm.jpg
2a07f8bb451025785aa2bc8d1f6a8394
9fb105dde7bccbce48fabdb141ef3a4b4ae88d4a
25792 F20101119_AAADOK makkar_c_Page_067.QC.jpg
91f6e844c04c3f1a3fd351147ff60333
8b59b906ab863d2b413af871b5af01d8dbb386c5
12955 F20101119_AAADNW makkar_c_Page_060.QC.jpg
1a002274442850818602224fe4d16eca
1e1320472a07eb504695b84700f13bbcc9acce54
6542 F20101119_AAADOL makkar_c_Page_067thm.jpg
a1936e1bb76a3d5a2efb9475a12b8d0c
9a520f6aee22daefd559c6f3416f66add03a1bee
4466 F20101119_AAADNX makkar_c_Page_060thm.jpg
5bfcfdb816dbcac07b85e6c4b5ea96e9
8073f886c80621bd5a01abbe448a817884c3837b
5071 F20101119_AAADPA makkar_c_Page_075thm.jpg
6e02b6b2f989439165ab2501ea51e983
501a123796c45a8694ae9808e4a9e8229d8474d8
21433 F20101119_AAADOM makkar_c_Page_068.QC.jpg
03b68d0d85d96c39c94673c83bf122a7
b5f16d1457749384c1836b37c2c83b3e09b53a38
11143 F20101119_AAADNY makkar_c_Page_061.QC.jpg
6702720eecdb54e5aa343e6f1293d94b
9288cc24a016baf48c414c037ecc620e48ba5313
14880 F20101119_AAADPB makkar_c_Page_076.QC.jpg
79776860671a1aae5af43e903c34374a
e6a8ca10c6be2a2b5c4dc0bb659430b4e335234e
6069 F20101119_AAADON makkar_c_Page_068thm.jpg
f8006719459d1f6f2b39d3a64ec05a91
870e6c46839c86ccbc036ed356a25a469f5a0520
3952 F20101119_AAADNZ makkar_c_Page_061thm.jpg
ae4f22cf71f9ec0f23d1a2467876a2f4
0344b9c8336b8d621fa83f56dc9eaaa32bd475d1
4847 F20101119_AAADPC makkar_c_Page_076thm.jpg
46d95d3fe2e2a6e98bd77d36533b5c38
4532c168ffe6e1c1e1ad091fa99a4f2c8d69659a
9040 F20101119_AAADOO makkar_c_Page_069.QC.jpg
c8dab877b5ea1da72f69cb6663535049
1e3358e476652b6423d9ad3eeac6bfd6f6d0bdcf
14844 F20101119_AAADPD makkar_c_Page_077.QC.jpg
14ed03ca1ddd8f41bfb932f7368ae4b7
c4b83dd10256e62737edddda9500c371ec61a4f7
22750 F20101119_AAADOP makkar_c_Page_070.QC.jpg
eeb039b6760c6951f083bdfc679b4af8
46f99766c81761d4550a7d587d1f0396c807e659
4699 F20101119_AAADPE makkar_c_Page_077thm.jpg
61e67745e5cdd9a86b8ef18bd2e905e5
dd464d45dd7a20eeb1cc7578653b269adb94b33b
5961 F20101119_AAADOQ makkar_c_Page_070thm.jpg
29cf11ba200a9b3d7a35cbfe3d3ba0ea
2319a0639e8f0bf80e94e9340935c2874d5c2c76
16370 F20101119_AAADPF makkar_c_Page_078.QC.jpg
4c6188849437804b9678c3e9914d1da0
3bc96d7914ca97c8b99486624d4f8f5d8d7a6984
20772 F20101119_AAADOR makkar_c_Page_071.QC.jpg
68a6f7a209c96d22a9061f7657974ea8
cf73228ab5899dc1d289eda539b9307c20f9d02f
5028 F20101119_AAADPG makkar_c_Page_078thm.jpg
0aace824b83d53c1157e44644301f61a
2727139ce4e6415873917477320ddf80cdec7e83
5846 F20101119_AAADOS makkar_c_Page_071thm.jpg
6e70657c38851d09d769d443c0ad3d4d
e3917d4920b237315a0e97f06d29116ef8124ca7
20822 F20101119_AAADPH makkar_c_Page_079.QC.jpg
bd857fc03e03cba61fd5de485eb9beda
2f5341ca5c3174413cb6165235fa056b66385d60
18286 F20101119_AAADOT makkar_c_Page_072.QC.jpg
b25c931fa2465d5e2b2dd30783a4e696
a7e0e9db7a86c1bb3209489937cdb979ba655b9a
5806 F20101119_AAADPI makkar_c_Page_079thm.jpg
a8926c588c1c1c0882e374335622e0fc
a294aeac1e16a0c408c695ffb109abe1bf7b0f12
5471 F20101119_AAADOU makkar_c_Page_072thm.jpg
b65310a903da45aac18215ff5b474f4f
c1f2d70755f330613812dbb30298b8e63c46f030
20365 F20101119_AAADPJ makkar_c_Page_080.QC.jpg
2e37d44a4fe77523c80d762c8ef03b0a
cef3669bc6c114548cb1c5953513d8c4149a065f
17688 F20101119_AAADOV makkar_c_Page_073.QC.jpg
657997dece9aed41ff30d1a8b79d5d32
34c406478dd63c27b5d56811995326cb4b7f2777
5930 F20101119_AAADPK makkar_c_Page_080thm.jpg
2830cf10e32f9b1283a4a3381c39fe1e
6c8db9512e661ee5f9bc34f4c238fd86fc7de493
5175 F20101119_AAADOW makkar_c_Page_073thm.jpg
9f72592d62a87a8218db76c72046f237
8916bf6ae507d199f40b09e15e5c85bbbb080fb5
15765 F20101119_AAADPL makkar_c_Page_081.QC.jpg
014137d4faf13c61efa816f7f1447fe7
d22befbf69734294aeaacff6784dafc42f49aba5
11015 F20101119_AAADQA makkar_c_Page_090.QC.jpg
d2dba4cbd62a9ddb4868f416bd78b008
4d38bdd5657709330e8007f95416f5b546c1c052
4685 F20101119_AAADPM makkar_c_Page_081thm.jpg
7199284e5636868a42a415022693a4fe
542ffbb9e1ac75615e2dd73dc9c4d9546fda8573
17757 F20101119_AAADOX makkar_c_Page_074.QC.jpg
a9c3c9752171aca7539bf2d8654acd58
36732339c1780f13ddaaf1f6ddf621e6a9eda803
3627 F20101119_AAADQB makkar_c_Page_090thm.jpg
0002742da48becdccc18e85b22f4d2c0
a9b107ede59681ac21342c801de13c770aca9424
4916 F20101119_AAADPN makkar_c_Page_082thm.jpg
3ec967eb141976aa46b1de1e573a0524
f2372ebcfa378aeb3110714d77d83f844ad6d82c
5105 F20101119_AAADOY makkar_c_Page_074thm.jpg
25c0245eecd53400e33b00820a62533f
46177e027818eec2c660c7384f835c66e4ad1d88
10750 F20101119_AAADQC makkar_c_Page_091.QC.jpg
bd38704e187321a46cb7f73c3990b751
c8fd45c76ae1d82fc20d96f046c8e1ddddc06c65
14362 F20101119_AAADPO makkar_c_Page_083.QC.jpg
2022e62b833b3b7acf7e0f8b07dc05b3
2b078bc9fb9153475a4ee0020b184eed75bd4062
15722 F20101119_AAADOZ makkar_c_Page_075.QC.jpg
9eb452dbabaa923ebcd7ab3285d564c2
8a6c6ae20d3584b2e99ce3cdcc8fd42b72876c7a
3364 F20101119_AAADQD makkar_c_Page_091thm.jpg
d5bd507f7c06f73ea45810e7cf4e8d3b
27305076c35a8e3af62b80839943923c9848b972
20785 F20101119_AAADPP makkar_c_Page_084.QC.jpg
9c78d778c7a9144d5572d97f39d1a884
1ae5277c85effbfaf0710393758912dcf5c00c22
12717 F20101119_AAADQE makkar_c_Page_092.QC.jpg
55383f19021e8b68ceeabf00d61d2d6b
f0ad6013d97bf0cb54b1e7ea644f7684176cfab2
5830 F20101119_AAADPQ makkar_c_Page_084thm.jpg
a676303e1c6f67f4f420fb91ff9581d1
7104bb9055da94a8ef8da207b583fd00182adc3d
13220 F20101119_AAADPR makkar_c_Page_085.QC.jpg
eef5ef81fe61dd5d7747e995222e2136
70afc5409f7de6994ec1bcca49b1402c55c19367
3909 F20101119_AAADQF makkar_c_Page_092thm.jpg
14e7f399ad7891b2fb36e366a9e669f3
2c93bd66c36b0ecb3e7aee5bc38a833c49ade30e
4162 F20101119_AAADPS makkar_c_Page_085thm.jpg
d565f9e22e1419c47ee4bfc48fe31463
2545b0c85169a5db0a68a5042dc581b8e6d3aef4
12671 F20101119_AAADQG makkar_c_Page_093.QC.jpg
7796dfb56c42af11bf7144eb8b38757f
7dba0a1bc7f4f5b6609b4eaf3c3f1917d800bdf8
12799 F20101119_AAADPT makkar_c_Page_086.QC.jpg
63d9ea1ddddeb23c4af51f39e1e041fa
5d5028f14eba440c5ddc81dbb0e1d291c29721ba
4102 F20101119_AAADQH makkar_c_Page_093thm.jpg
0920d148ccdb6adc248b1128fefc8b83
a243f6b92d059276b2ef937914912a5fc1b23d18
3636 F20101119_AAADPU makkar_c_Page_086thm.jpg
e40ce1ea394d3f81d2a779d38c019ae7
64799636f0a2dde49109bcaa483f8505d3118a2a
3599 F20101119_AAADQI makkar_c_Page_094thm.jpg
66d9c37c09416294244dbd807359205d
0c9e011be77333102e2cc567043f868af1a36805
23144 F20101119_AAADPV makkar_c_Page_087.QC.jpg
996855ef8241be1949ed520a6168f020
34ab75cbd40d57c87b3a182ac6867ef29832e357
12861 F20101119_AAADQJ makkar_c_Page_095.QC.jpg
d73770ae5652ebe1f0f0f8e56894ec45
17a414c788524ae74c46b0d0d37721346120c0a4
6022 F20101119_AAADPW makkar_c_Page_087thm.jpg
bd21a616e6214d00ddf97538220e1325
210f87ecb72459cc07a80f38327c2e6d15199054
4181 F20101119_AAADQK makkar_c_Page_095thm.jpg
b21782d25505bfcf8f29fc21ff1dc794
dc0fdc4abaf80dd35982ed2468d0858057b93928
19682 F20101119_AAADPX makkar_c_Page_088.QC.jpg
b3559b5efd0df9bcffec035c71336519
a55ce6c856ba2f8eaa2d6578b180abfc694eb83a
12395 F20101119_AAADQL makkar_c_Page_096.QC.jpg
86fcd15827ba3684e2b93c8097b30e96
339f59dc4383f3c58c0f245941e2d20287bad083
26709 F20101119_AAADRA makkar_c_Page_104.QC.jpg
d63d584b61a71605757b0f2ffd55d08f
0154bf0fbaa04c63efb2bddbc9f312919e5acfc7
4124 F20101119_AAADQM makkar_c_Page_096thm.jpg
622086804e473ea0017aee2fc3d11e59
8e79195cf3d95c05413ab69015835b2ed4599d28
15219 F20101119_AAADPY makkar_c_Page_089.QC.jpg
4b0ab3369a4b26ac0b2f6d96a46fa273
28dabd751f8d30047cf072524cfa6c8b22a25d7e
6784 F20101119_AAADRB makkar_c_Page_104thm.jpg
071aa8970f24200c87519f73d8d39d6e
de75f8fc353ee6543b6f1c4524b340194c8e37a4
14467 F20101119_AAADQN makkar_c_Page_097.QC.jpg
b76bd98f75666425e18cec4159801df7
ee1d6fe2369ed4b8c0084a897b78729f94ff19b4
4628 F20101119_AAADPZ makkar_c_Page_089thm.jpg
7b2f07511dc081e128726003ec45a5fc
a70cc005c7d5cd513007bd9aa65921030f340815
25827 F20101119_AAADRC makkar_c_Page_105.QC.jpg
461890112de550269fd54ce521d6e9fa
e690afe87a11ae199635b9cca7c003c4a484bcc3
10346 F20101119_AAADQO makkar_c_Page_098.QC.jpg
c687d596378c99756efb09a297e6813d
7ff4c51a61823943ca216f0e636a5b679879d903
6878 F20101119_AAADRD makkar_c_Page_105thm.jpg
def131e6ee53588d05c08065895400aa
5556ddea4fe8935a9a5cd17ee5041a245d73a4b5
3154 F20101119_AAADQP makkar_c_Page_098thm.jpg
2ff12c88d57704c95f8624ce0fc320a9
143124afd1023464a885364c88eab98d7c297097
24418 F20101119_AAADRE makkar_c_Page_106.QC.jpg
d769cfc296ebd1101b4a5529158744d4
0196b8452a085763ed094231dabd5e0a73d1578e
10382 F20101119_AAADQQ makkar_c_Page_099.QC.jpg
f4430d25fd71d70011fbc2a7d2c36dfe
0c09616110db224c6c7251dc8b9d69002d559ebc
6347 F20101119_AAADRF makkar_c_Page_106thm.jpg
02d3ff64a2b01b454f2dd40e07b4bf56
78be5f5b9f50e3de8ca438ec268e0d397d87a888
3416 F20101119_AAADQR makkar_c_Page_099thm.jpg
1a47b70f92de40052748ba7fcd29365a
deae9d52b92c5fac77e6d032220ce7befd106b2b
8908 F20101119_AAADRG makkar_c_Page_107.QC.jpg
c833881e811b2abb7f6dc6d1f7d08620
f7f6a0cf52decba681992c54c947c189dd18b098
10356 F20101119_AAADQS makkar_c_Page_100.QC.jpg
00fc3ccbc9253d99b122f81dffce9100
e854ed96d51721cc133aecd134c7cbdc6b202790
2763 F20101119_AAADRH makkar_c_Page_107thm.jpg
b92f03b741f2427b3444c780745757ff
2d762c378065f9a92c0a6bd19e3aff01632c42d1
3279 F20101119_AAADQT makkar_c_Page_100thm.jpg
e23051e7c42351b12ef822d8289f654a
04f8e461fba247cc2be702d5cfbd79c53da55449
124344 F20101119_AAADRI UFE0014262_00001.mets FULL
35255d51fbb51d7d5266bebc3d875dcc
e6f679bc135d71b90927e5ae3f211644ebd90d84
7106 F20101119_AAADQU makkar_c_Page_101.QC.jpg
25f6985ce7aecf392714c3ba2b7b24d3
f8036b91f8bc83d69cb5781071006e7cda990e7a
2320 F20101119_AAADQV makkar_c_Page_101thm.jpg
3911e911da403bfccdce28bf7afcfaa8
a8ff63735a79a3dc7857657269a9fc7cc167f6f4
23441 F20101119_AAADQW makkar_c_Page_102.QC.jpg
9b0b9c4c9e0dc06436e330d20dfe27f9
0534ef604618061dce05610b1829023509b6b2e3
4212 F20101119_AAACOJ makkar_c_Page_097thm.jpg
c79db972e2650bed0570edbe6326813d
cc7508370275544759ad99792f782c2295c772f5
6067 F20101119_AAADQX makkar_c_Page_102thm.jpg
82f968df7dc667cd2d73fe5d13ccee42
ae2cf653e8f880e1c3f21ad5dc6576c98cb05e8f
3217 F20101119_AAACOK makkar_c_Page_069thm.jpg
75ba102491f9b4b12b5ee01e6a596ae0
2409371d47eded2f8baf074daf402995f124dad0
26415 F20101119_AAADQY makkar_c_Page_103.QC.jpg
93684c7c99002a2197b2bd214b5b63ed
78ebc6468106cc9f80f6c89f72b0208f0fe012f2
1595 F20101119_AAACOL makkar_c_Page_003thm.jpg
1ba37eb47b37d85c6659d2803875df1d
4eaba76cc596555b8ed5a6c88fd3ffc55b33306e
1788 F20101119_AAACPA makkar_c_Page_074.txt
792bce8be76005afe4c98120474c8c8f
c411e4a3c0155580bfe2c6e0a27cc6193b4a24f5
13964 F20101119_AAACOM makkar_c_Page_045.QC.jpg
a858b95c5a4ac9b749ac4f1987c4d6dc
68ef7e1b8edca10804651550e5efac5521366647
6819 F20101119_AAADQZ makkar_c_Page_103thm.jpg
8f47dd9020b636a61b382fef62d3fd2f
85d46717b421fe92732295754f7569f237998887
15790 F20101119_AAACPB makkar_c_Page_082.QC.jpg
a641092f0722926a9e920df6b3498af2
77d8b9a71d95493f184e7f979e9e7ed4ced7f63e
F20101119_AAACON makkar_c_Page_065.tif
4cf85b7ca1e0725b0326b0d090217400
a75dc149c175937f396d6c359d7792ddfb6d4a87
5184 F20101119_AAACPC makkar_c_Page_042thm.jpg
34afc33faa9cb7f75ab2be562006ca68
cf8bc05bcb048fc18d7bbf094e5bbb82a778797e
455 F20101119_AAACOO makkar_c_Page_001.txt
3db15574f6dd643c8ae0d301d94d216c
88d814e147cf37a9228b6242622848cb1ed28588
52246 F20101119_AAACPD makkar_c_Page_064.jpg
0cf9c8de29d5c0e692fb3a16c06196ad
dbf3179d35414ab261585766b9de2ce536bf482c
F20101119_AAACOP makkar_c_Page_042.tif
53c15072654b8279543c940b4890ab9d
c99359edd83cd3358046af1403fc18c5bc2d31e1
2298 F20101119_AAACPE makkar_c_Page_008.txt
d1fafb6644a07345a5350f6d553b3ac8
3569c6515349c036ec5fbe9c5058916c4f43f789
1085 F20101119_AAACOQ makkar_c_Page_056.txt
4acc96d23d0d47a8e574f104e48ac363
c41e1272eb3843e410d3b4987920cd509558183d
50751 F20101119_AAACPF makkar_c_Page_004.jpg
967651d0d829bb17ddadf429757fda9b
c481f30949cd6a9d10ce39be2f34f6058c3ecd4d
89502 F20101119_AAACOR makkar_c_Page_013.jp2
248cf0f86ff1b4aef136ef154b0fe2c3
62ff481fb35fa7eda887525d1c2814749086814a
2210 F20101119_AAACPG makkar_c_Page_102.txt
d7ba3202830b7116df2e5031455da385
4fdb14cdff7e1b1373befc250d7c46fb3d642176
69164 F20101119_AAACOS makkar_c_Page_089.jp2
18369f4a6a241919015ea7d3351df590
2ea99d94d8bf8839761b57e08159d6fdc75e7a22
34097 F20101119_AAACPH makkar_c_Page_031.pro
e87ce76c5f96398a1187a3c4724b2299
d1912e4731c161ed9b39233c9b723b14d9170982
F20101119_AAACOT makkar_c_Page_068.tif
d453445eb43a2aff003e9f28ca81199d
e02fb970ab996a519c5d13e76ce2542a537a0148
117210 F20101119_AAACPI makkar_c_Page_018.jp2
77225cf50fad401e5d84ec1cc6ea9dc5
137e27b10927a18f3e0de13ccc81019489f8585c
6688 F20101119_AAACOU makkar_c_Page_014thm.jpg
336ea191038b456791ca175b695216d3
de4e7a8a5a0f6b871c1909e0f89288e97b11fbe3
1980 F20101119_AAACPJ makkar_c_Page_041.txt
b90587ff4223a92cd9876339062e1d41
792dd32ae7cdab9fe2b24dd4f207dc4dd5a8042f
20565 F20101119_AAACOV makkar_c_Page_022.QC.jpg
1dffa033d803191d038218dab62703be
aa0a539e8e473e10a7f1eeea2d49391a0a1af915
1686 F20101119_AAACPK makkar_c_Page_043.txt
e16ae3f32da3bbbe61698a234ab03c6d
2a524997ccb49a6d6606af4814dc7efb1c50810f
962400 F20101119_AAACOW makkar_c_Page_084.jp2
b725d9489dc857167ac0aa9ba673d929
31e997a949d06ad902aaa328280c60658da2781f
F20101119_AAACPL makkar_c_Page_031.tif
4bc042196826501f8873a146a947c04a
89437fc76ee3596d935f9320c56a1a508c0f333e
5178 F20101119_AAACOX makkar_c_Page_012thm.jpg
6ec4049635a2addb49c79fcd7d2a1b7f
b7cf57cb08576deec23bd96bc7a0d5e4a532d60b
35641 F20101119_AAACPM makkar_c_Page_061.jpg
5712aa10a64aa118daa58e4af667de36
494784bd575c5896d05749df1c7ad9fa1d46ac70
F20101119_AAACQA makkar_c_Page_098.tif
cc9347e03b08403885dd37abd08c10b7
de7bb7c3ac99a64b0e7712ebb06eca5d747021eb
5193 F20101119_AAACPN makkar_c_Page_088thm.jpg
0b12bb13969806b4eff99850c10ad78a
1ebdee0020bb2fc3eb78ded071a4526904c5c2c0
5517 F20101119_AAACOY makkar_c_Page_026thm.jpg
d123d0d673117760b002979c787100fd
40e9d3827b087c0336ff35c43d6baa2b25dbe6b6
79262 F20101119_AAACQB makkar_c_Page_018.jpg
a96b919bd32cc076df2e8eafcdf86d30
7d3fb9ad44ec9edc2962093c2bb7c5e64ca42387
F20101119_AAACPO makkar_c_Page_019.tif
1c83b80311d84669b3fb2aaee5de4ac3
b40af493fe4e591b061daa688865263e43367b43
4316 F20101119_AAACOZ makkar_c_Page_055thm.jpg
8cbe244f6be8c56a4cc60db8a46fe50f
3bdc41ba899e2a7efa43197473deb10b61543945
5339 F20101119_AAACQC makkar_c_Page_023thm.jpg
0be353a33bb77c61c61058ee8af554f9
412f9c98b6475f5418b91a39d3215c95f10d8f1b
F20101119_AAACPP makkar_c_Page_106.tif
7862bea353e70dd959d107f8e1ea4d61
4d1757b839019ae39c4ca9c80fc08b32d40b6c38
51393 F20101119_AAACQD makkar_c_Page_043.jpg
e82ea14f3a5df4e7523e8458f1603735
04b4ba7899658148e950e24c9ee16baa7cda611f
85593 F20101119_AAACPQ makkar_c_Page_106.jpg
fa32375985eedda3710702c857f2bc81
33ca7f15b03760f0ef5b5c2f6258c605dbb5bc72
987 F20101119_AAACQE makkar_c_Page_086.txt
c6d50a78e241a278564fba7bb41d5a59
8a4b61e2e32ab5adf4e7c9c82e1b02d259062b84
58399 F20101119_AAACPR makkar_c_Page_014.pro
235f0ebf5a5a3bcf9c4f423cd0f732a0
21145cda29bff2247e2d7cd3f2ddd339e83683b4
18239 F20101119_AAACQF makkar_c_Page_100.pro
e702d479bc1f9d8bcfcf07976e780938
c830ceeead23e17e2ad73d3033f7a9ba84c5537c
F20101119_AAACPS makkar_c_Page_056.tif
c277ac24f5341abb1e7fa9a4d7e6d8fa
9b94db6b47aea2ff6277fb67e453b9c4120ce4ae
4320 F20101119_AAACQG makkar_c_Page_083thm.jpg
155a47d48bd7bc9e93bbf7e4b4de168f
b0a10d5e32f6df2e5f366f2392d2d407e29f2f39
61481 F20101119_AAACQH makkar_c_Page_013.jpg
0899d7df74b82a7bcea390efafd60987
9751a193ed668d283c47d146beaaa6e502e6c7ac
20570 F20101119_AAACPT makkar_c_Page_026.QC.jpg
5209e1169e15d0f1b98809444a21bfbf
4264a1dc450b52e870ccb7d73ed8294c0d5f93cb
1051968 F20101119_AAACQI makkar_c_Page_008.jp2
59663a35a8b263bf7e910e7b59aa4ab2
f74649adc913d49554be2c4a53405a3a59545ce6
56465 F20101119_AAACPU makkar_c_Page_015.pro
7b0318f035f0b5cd18dcbedb5f147df7
fba3deb4ed18a9fec3fa7ad2853b291092155afb
1411 F20101119_AAACQJ makkar_c_Page_031.txt
c1120cab055e810a3d25a24eeb34b2a5
49292e3b6519d9dda13a6c24789e430f55780ce9
10344 F20101119_AAACPV makkar_c_Page_094.QC.jpg
64872b79c49ed76a1de507cc9941f412
5f0f307c259b629fa96f64f88b596cb2b7023d38
1582 F20101119_AAACQK makkar_c_Page_046.txt
0131c4d79eca78e235483a7f034e5f9b
4815ddf69aa2cfed983240b36d1cd73755b5e8f5
1590 F20101119_AAACPW makkar_c_Page_026.txt
2ffe2b40311bcef0dfcd7080aec7510e
0826e6e48ce471934cf03d4175db65df964be858
160923 F20101119_AAACQL UFE0014262_00001.xml
c2e406716fd472ffe9e0fbdc34c1d413
0d1e815ff94ee2a8ae90b26bd4c6f7cf11a415af
12459 F20101119_AAACPX makkar_c_Page_053.pro
253e19da0ef541d2ba7fa4542ba24f95
5bc8c168da750a3aada103e1153fa3b854e6be8e
79449 F20101119_AAACRA makkar_c_Page_015.jpg
74c6f26d8b9cdd6c1d9587627ed2de63
67c590dfdb1687b686c3839fb296d2428fa2c5ee
F20101119_AAACPY makkar_c_Page_091.tif
23f26be9851728039ccb6b1dafebf251
8e5fc29197a5520a7fd561c8fbcabd16dcfc4af3
82256 F20101119_AAACRB makkar_c_Page_016.jpg
282f612b668dd2b31cb6a0f4cfc02b17
ba6084a0831c33536d5a8a0cfb5db9f6032c111b
82473 F20101119_AAACRC makkar_c_Page_017.jpg
518a2835f904aae5aa8819d5775253cd
bc5a6450351f51b40498d884dd5b9fee96da4b25
24197 F20101119_AAACQO makkar_c_Page_001.jpg
d7209de786b3f8396bcb359037a98212
de391f78a6dbc6068a8f566f8f4766ee32f7151d
4586 F20101119_AAACPZ makkar_c_Page_051thm.jpg
607a77aae82259fc69059a4ffaef7741
9448c7a5209ee03e3ae72a482922a2a314c1d916
78480 F20101119_AAACRD makkar_c_Page_019.jpg
26ea418440810b02c9ab5d0596cb2beb
dc425c724f384375ceab607029ac467307942cb1
10165 F20101119_AAACQP makkar_c_Page_002.jpg
9b2030c05a4d2eedacfe3c6edf4a6e1d
8199f9e250aad537103d93174fcbb2f8dd62af84
20084 F20101119_AAACRE makkar_c_Page_020.jpg
b9a2e7c90e3268ec2818b13d5f0aac08
dfb1d719bfb962223236f8553440f6f2ed25fcd0
12953 F20101119_AAACQQ makkar_c_Page_003.jpg
ac4f29cc5c0d0e0982450dde1793413c
ded54280c2a5d2218650a73ac0d658e47c628951
71318 F20101119_AAACRF makkar_c_Page_021.jpg
0bf95c5461369bac3fedf57ca06332a0
2f035720c4ffbbed22010a52116365bf381ce486
75603 F20101119_AAACQR makkar_c_Page_005.jpg
7be0f2794c30956bb8a467c20643fb0d
e5b7df53565ade72522b1c1b4da7539debdd0c20
65591 F20101119_AAACRG makkar_c_Page_022.jpg
a1b20671d37ef1fdf634122a9de54e2b
440d36543c3587aab6349abb3de51624b6cb6325
42311 F20101119_AAACQS makkar_c_Page_006.jpg
7e5199dd826fc2c7c59687d5a9120026
458a1476f34cf2f35854caedfd0fea13120408eb
58541 F20101119_AAACRH makkar_c_Page_023.jpg
a866016a18276d98b88fe7d7a45ae85c
289a36472ceeafbfb4ba5c64bac855ea1d4e7086
21118 F20101119_AAACQT makkar_c_Page_007.jpg
53581ff9596952f5dabcb85eb624dcbd
1548f362a1f17696180952b7a4ac2efbf12c9ddd
26892 F20101119_AAACRI makkar_c_Page_024.jpg
74599f6da44104a14a495d19bdb5a6d8
0515167c030f40758e188c7b318d746d9b97ccee
70092 F20101119_AAACQU makkar_c_Page_008.jpg
6810373babdb3c0974bfe33978f3debd
1e786a8aecfc6f44e3b919f0a14e845288a7fcdc
35505 F20101119_AAACRJ makkar_c_Page_025.jpg
b63e637e84296625dc9da49b3b1356c6
2b8dc274624bb9a45a78cd4e55ab5d86c4d34af1
97738 F20101119_AAACQV makkar_c_Page_009.jpg
24c51fe81dac6e747abcabfd655f1128
44d15af32f9e8c88d87f68cd5569c8294ee42d20
62284 F20101119_AAACRK makkar_c_Page_026.jpg
5b9ddd2d6318766859e11baf4c0027d0
cac75a939bb3ed173f4434c1d420fed4bef039a2
93362 F20101119_AAACQW makkar_c_Page_010.jpg
aebf054ce6f25ef89275f3efb2065b95
1fe623248eabbe6739ba3e67b8b193c8b3096799
63190 F20101119_AAACRL makkar_c_Page_027.jpg
640611d7d11124de1143c3ff11250076
893e6f4072579d5d79ae3ad3ae1ebaa4802de530
67175 F20101119_AAACQX makkar_c_Page_011.jpg
34832eff2470105af2ef817c0fa127c6
f43cdfa5dccf5ab7912905d27de903c60fe3ece7
35081 F20101119_AAACRM makkar_c_Page_028.jpg
5b29b642c63ef2e36f8c368b1b5218eb
4a725ab72e0e8bc794f45998804ca4dde12861b6
60323 F20101119_AAACQY makkar_c_Page_012.jpg
960b5216df4428abcad6888068788c5d
decef2b973cdb6b0b53a34916bb78c651be1437e
55010 F20101119_AAACSA makkar_c_Page_042.jpg
c449990c78d3d733d824999855371061
79caa64720af2477210dad046c1ce87a4c97f9b1
25324 F20101119_AAACRN makkar_c_Page_029.jpg
225bc2166438384ac3600b2240db21dd
7f5e3b4456ad1c145a23eeb0f2908b4876e65018
81618 F20101119_AAACQZ makkar_c_Page_014.jpg
c1702916365e213d6286fcbdad948c3a
5512e9f1b8d4376f9ce57986666a4f88f5b72d3f
47315 F20101119_AAACSB makkar_c_Page_044.jpg
f6bfaea2f4cee2550c5c43fd0f45c1a5
882ea9c16ad651b30f58486dde662a2231b75f31
29753 F20101119_AAACRO makkar_c_Page_030.jpg
ef3d87e621e8443a9985844d70ae1ab9
178c571af362803eef1919948378ed255cc96a80
43432 F20101119_AAACSC makkar_c_Page_045.jpg
7b930ca7f858e0bab15c1b2584df1e72
c165ca2da8b3683a829d58fda9cbc4af03412e4a
61613 F20101119_AAACRP makkar_c_Page_031.jpg
6c2d8707c834a7d9c7f65524de1b3640
920a12a0462b16c33c857f58cb9bc55dffb97880
45123 F20101119_AAACSD makkar_c_Page_046.jpg
157de8f964fb6052a226fdf8a71ac061
2950e0f9543417282d7d184c529f673ced5ec661
33803 F20101119_AAACRQ makkar_c_Page_032.jpg
fb6e7d09c39adc273322e5c41c4cec81
43bef7e6429cf9047d6f9b7571963e0c6d2142a0
58291 F20101119_AAACSE makkar_c_Page_047.jpg
b5ca6549ed66861a7bdf088fc6475949
981b6ba87ea1ab5d99602435532354acef3ef5ec
39887 F20101119_AAACRR makkar_c_Page_033.jpg
55f5eab7ae3bd3ae40c60b40b19e6392
65a6218dd52ba5618503a52a820cb3a297b5b6bf
47600 F20101119_AAACSF makkar_c_Page_048.jpg
2d505757937e8b40d007b1cb7ccb4640
43848b702dd5295465a12a54e86513f1de4d67b9
34198 F20101119_AAACRS makkar_c_Page_034.jpg
af816d7d477ef78f77250179f2de332f
0d295cb68de9c12d6bf78acb5f4882219c8557a2
55031 F20101119_AAACSG makkar_c_Page_049.jpg
0fc9ab09101cf943fe52fe335bea2d3d
a2faedfd9f53ae179ce32134444a279cc8cf1444
36272 F20101119_AAACRT makkar_c_Page_035.jpg
8ae263c75c0d3776771404009e59f671
d2308703ade013ac26c7dd59b6630637e37a4cb4
54768 F20101119_AAACSH makkar_c_Page_050.jpg
79025eb34fd3130e0a56622c9227069f
266903767d996696a8d6f930a4710835efd1a8d4
74110 F20101119_AAACRU makkar_c_Page_036.jpg
4500c09cee8cf81c99ba13e877b7f0ce
585f46772c98a8a3f69ff78cf0526a3466b4be26
48590 F20101119_AAACSI makkar_c_Page_051.jpg
0bb71ad2e8732240ea30a8bbf5a4145a
a970cedbd7f6ac8d64f22ba8daa8194edc8ccd5e
71851 F20101119_AAACRV makkar_c_Page_037.jpg
30a3b43599e5ed432b90b2fae1637365
912be2e9c064764ea7c93c38638ce021df86da01
40383 F20101119_AAACSJ makkar_c_Page_052.jpg
2378e32845281c6b963bde2b29cde2af
ed6f38413038bdd2ca8c68096d26b9038b6e42f6
48357 F20101119_AAACRW makkar_c_Page_038.jpg
5690a427ad489787fcc7408288ea2f6b
3442f88b9c39f971271436fa4e649d86780baa29
44043 F20101119_AAACSK makkar_c_Page_053.jpg
f3971fe088108356f4cc9a4b26b643bc
72e26cce087df88c51c451542f11bbd376565dec
17669 F20101119_AAACRX makkar_c_Page_039.jpg
3757684b2484091db1346d53b4073259
c0b60ccfd0cc4995344932231ec2739bfce98d70
36396 F20101119_AAACSL makkar_c_Page_054.jpg
6cf9f79e86eb0c7bd8ad2a7ff07c05b7
18917f03c5273737fb2c8faafef55e55b2953298
71038 F20101119_AAACRY makkar_c_Page_040.jpg
a36edb7378fbe62a549b92706f5698f0
e7c931243343be231dfe54ad2d340547d4931961
64937 F20101119_AAACTA makkar_c_Page_071.jpg
108c666199bd7286258689be87e40e3c
277571340da3402888dcd090747d55f2d086b39d
38775 F20101119_AAACSM makkar_c_Page_055.jpg
e03c9d2d78f6e69564e14a0272ea9cc3
0170d5509ac086c68e3458913926bf2749f68591
67726 F20101119_AAACRZ makkar_c_Page_041.jpg
482fbbab3112f9f75593410bea360138
6872bb99253f20a7cb9f30b9c15d73646f342e5b
56948 F20101119_AAACTB makkar_c_Page_072.jpg
7e5eb2d3db992fc4b767276837681ead
7402cef21ff68ac870a654eafbff30331a50d455
46538 F20101119_AAACSN makkar_c_Page_056.jpg
78a43f7bd643f3f208bea65baef85017
7060e7e9934af9e654ef1172b0bafcc9eb06e3a1
56175 F20101119_AAACTC makkar_c_Page_073.jpg
5b6ef0234a2d98a9cc392552dd39c71b
69b06fb210404dfc51d55f0a23147c7cef6b2579
64344 F20101119_AAACSO makkar_c_Page_057.jpg
aff1daa31b45e7adf1aaab0b6845133a
502c543c877374b6734ed8f87c470d8fada45bb9
53258 F20101119_AAACTD makkar_c_Page_074.jpg
8a282e537a8530b74c889c6e79621080
356691c4ebc38bb2dd766d79fce61384e1a397f2
39405 F20101119_AAACSP makkar_c_Page_058.jpg
11720ca63c2be69e5b7d7df20ed72ac7
4dc4c77a746387704679076d9f8e6ed9a7cce2be
49740 F20101119_AAACTE makkar_c_Page_075.jpg
141608713ce04e197507a85ff4a432f5
a6db6335e95871d168d6975415cfaf58cf3ee4b9
48391 F20101119_AAACSQ makkar_c_Page_059.jpg
df7115a3aa009c9d43a49edc318037e8
cd0fadc6da9814fc43be881558128b29f38954cb
45779 F20101119_AAACTF makkar_c_Page_076.jpg
2575efad8a5c3165be7faa7d63a7cd30
ec317e5d3749d29e1727f89c51daa321ff9f11b1
41856 F20101119_AAACSR makkar_c_Page_060.jpg
1070d991f1dae7e0015b89acee49202b
7484aaaab2f20694f3682d19e5c60766003026a5
45569 F20101119_AAACTG makkar_c_Page_077.jpg
8a1228cee5f27a1b2cd93e68bcd35f81
91e2d1962810c67991de4ef7e58a35ee4cb12422
41925 F20101119_AAACSS makkar_c_Page_062.jpg
2403661134b5edfc78f288ed81cfb36a
cb14578bad01958cb79e39c94997ea620eb4e3c2
50567 F20101119_AAACTH makkar_c_Page_078.jpg
1f79a4bbc2ad01f82a7b0b7d7a542e69
544a69a5cdac3ed0c99c4880c430096b4ab29d98
65973 F20101119_AAACST makkar_c_Page_063.jpg
830fdca98f7d694cb0557cab750c4676
605cfcac8c820644a88d885fdd2105ad3b319dca
64018 F20101119_AAACTI makkar_c_Page_079.jpg
e5b07d1c44ec81adbfb5d71446db4846
3001ce96beeb97b0c2fd628b19ffc2da13d3461a
57424 F20101119_AAACSU makkar_c_Page_065.jpg
9195d3758c609662f97d489f5d34b7f4
ae73405daf2deb720ddbbab1b90054eb6748bd78
67901 F20101119_AAACTJ makkar_c_Page_080.jpg
1fcbeefd0b201702eb3c3ab56e73f291
a46952f6b73185548deafc6ae4ec3fc047355dfd
78194 F20101119_AAACSV makkar_c_Page_066.jpg
922786a8cb78048bcd46e1aa6eaa920b
6593a968cf5876e0c9b3eadbb84980f703908721
48012 F20101119_AAACTK makkar_c_Page_081.jpg
eab8e755aeaedc3a1df49bed99ba696a
29ef62ebbe70587da250a97660544d8ec1664cac
79186 F20101119_AAACSW makkar_c_Page_067.jpg
7eaf2ae95b892dce37a5744182892980
ebf76a06eceb5c76584a2996cdb5d3959601af20
52510 F20101119_AAACTL makkar_c_Page_082.jpg
7eb0b1e6bb3b26abd8110fea4764539a
84414ca9aaa8a10c50c60eec1f72b5f56cdd79ac
66206 F20101119_AAACSX makkar_c_Page_068.jpg
253ef0969b31f6f0b9d1a0bf98933258
d0e631164688ba2492c5576ba0440be953659c9c
45929 F20101119_AAACUA makkar_c_Page_097.jpg
901e396e7c2c7e2779202342a5a8cde4
fb7dcaf061fafeb8846ae9a670dff8f970c5673b
45304 F20101119_AAACTM makkar_c_Page_083.jpg
7a4fb23aff3bb35896d8df1daabfe82a
ebef1f3b519fdddb296aa986a738cb1dff9dedc4
28360 F20101119_AAACSY makkar_c_Page_069.jpg
c9d16ad2c660f29b9c12f2e67c8a031e
9a64da892db7de88d12398fe33fc1a44fd4588e9
31852 F20101119_AAACUB makkar_c_Page_098.jpg
1be75fcbcd4444566e7523b74a73c0c1
72ca16e3541ad9c70eb62c739be63fc3c979faa4
63493 F20101119_AAACTN makkar_c_Page_084.jpg
1d44654ab3b25e71c1d9c381e12df242
e197d6e47c6d82353ffea36cde7af61b075d3fda
73735 F20101119_AAACSZ makkar_c_Page_070.jpg
349cab29b584646ba9fcdae69564d466
335ddf8b4e25444bf2c0e7f4fdc6703a2e1b90fd
33083 F20101119_AAACUC makkar_c_Page_099.jpg
8f2d5210378572638b376d1af69b04e5
2571548ef3db50533e9e626eee258f734b3ea71d
41586 F20101119_AAACTO makkar_c_Page_085.jpg
3c184fe1c3557fc21cd6e0f4babb3f07
5d18c1371b2bdc7dbc5608b2340db500eb66d058
32022 F20101119_AAACUD makkar_c_Page_100.jpg
22b56d87b8ad76ed1fa172904576c492
66a35f6c85d816c00c615d8f7e027134a17f90b4
39510 F20101119_AAACTP makkar_c_Page_086.jpg
9ec6df0996af4f24e3261934cfb8c334
4291a6a9d45c50904e5aa3d6a0f6c80454c32c02
21643 F20101119_AAACUE makkar_c_Page_101.jpg
43c8bdecc6b2b3a00d9cf8063598ba82
fad45ca14b631803db2dac757f5c649b67c4aace
72708 F20101119_AAACTQ makkar_c_Page_087.jpg
111508359b20794c4b124066fc5386fe
218aa2f141bf6b6f63bd180d6d1925aeb2bdd5d8
79902 F20101119_AAACUF makkar_c_Page_102.jpg
8a8affd0a8537da62c119338a0ed27ce
7aa82befd5c301afc37b4eea394ad585f29f0eb0
59280 F20101119_AAACTR makkar_c_Page_088.jpg
c6300254f2f270143241b278b54b2132
9233786918cfc979959eda1fb54ca8f8741cbff6
F20101119_AAADAA makkar_c_Page_048.tif
b83417a1dba243e6828068dc8ee7588f
6069b964cf635f96c22567797f62f45bdcf4216c
90399 F20101119_AAACUG makkar_c_Page_103.jpg
24216e98a39706ae5a608b08433390c9
720359ad6665c2e0461a51984e23826b91d56298
47764 F20101119_AAACTS makkar_c_Page_089.jpg
75018d7a218346e5c68e5d1415206dec
cbe18fd53a047c0dfbf33d2ad56cb2cf2390e44c
F20101119_AAADAB makkar_c_Page_049.tif
94f49cf00a602c5697afae1ae8d093a1
e3c18799340e6679aa95b11f532053bd6f5cefee
90574 F20101119_AAACUH makkar_c_Page_104.jpg
7a11ac331b8a8c0c49ae89d65d3a7ade
3a6c2d83fa2e0189f199ca91d788ab034b67968f
34618 F20101119_AAACTT makkar_c_Page_090.jpg
7a8b4a781070887f1cae5df18031eba5
175f30b1702cd1f4cfde8a1a54c0f95b45150516
F20101119_AAADAC makkar_c_Page_050.tif
86d26ce8c7c24e06aa2f4b2b91a8cb60
db1254e9f2e61dce07e3b3545392737aaa478b05
89355 F20101119_AAACUI makkar_c_Page_105.jpg
89d6731f9b9c75cc95365afa4c5522c5
ee8f686c42dd2a5cc949d7520b318a5ab7e431ee
34051 F20101119_AAACTU makkar_c_Page_091.jpg
461ab0882af9df4855ad7abf768470f5
392264505aa622a9b7fae7825f725edb2e16329c
F20101119_AAADAD makkar_c_Page_051.tif
796c209e2ee9b8cc9e6505f0ae907827
2cafbaa9e9b3d84be99d4f8dde0120bfd60338d7
27780 F20101119_AAACUJ makkar_c_Page_107.jpg
f406bbf71accec6c333ee0686c3a87d2
00ecafdc08c107d9dd057172b9cc2ea3bd5c5a8d
39384 F20101119_AAACTV makkar_c_Page_092.jpg
815f0fdd85e370bcb8b29863605047e3
d02747e695403336f07ae62c2ecef6e930063939
F20101119_AAADAE makkar_c_Page_052.tif
936fcdcd97bde8f7c4fc9b73bf7ded07
9842b5a20cf6d9ae362ef5f2f4432cc7f242c445
24201 F20101119_AAACUK makkar_c_Page_001.jp2
f7a450ba8b4cb2fe30dbd46eafc21f0a
6f9d5f60e2a007665c4d1ca57550fb5516042e41
40727 F20101119_AAACTW makkar_c_Page_093.jpg
9567145248ffdc2b51f0f73b5a264aa4
8e1873ec2f06f6c66c81023d06514abef50702c0
F20101119_AAADAF makkar_c_Page_053.tif
f721d6d61d834db4448a638980c5a809
0e4a6c0ecbc0efeda72d65a5ea644f1bead711a3
5595 F20101119_AAACUL makkar_c_Page_002.jp2
90f6d84db4f6cdc907a1107397f17dab
6ddf55cb63d17dea5aeb5bba1aa96c5be9e17b0f
32879 F20101119_AAACTX makkar_c_Page_094.jpg
ade03d8614337b9565a845011523e110
2f54a274e26c27037cc830cbddd4c7507990894d
F20101119_AAADAG makkar_c_Page_054.tif
e27ea60581a94bf2c704255a7b0329bf
8feceb2476a80e81fc1a482849fdac359a0c65d6
9974 F20101119_AAACUM makkar_c_Page_003.jp2
490deb0bcb07186af9e9e877f1818e50
0eaa19af8d335bb579fe51616a97af24d953e095
40076 F20101119_AAACTY makkar_c_Page_095.jpg
d6ff5155e8088ccd23a7151de15da9f0
20380996cf2bb73072729d3b21644099c1252823
F20101119_AAADAH makkar_c_Page_055.tif
aff4109d11d0a516b39e326cfbe1b74e
91bf5cd80ad772c05a79836ab9817543bb6a63cd
23842 F20101119_AAACVA makkar_c_Page_020.jp2
a424d4f3087152af32c243d8f41d642f
76575b6c545201e1541446835b20a2086855623b
73056 F20101119_AAACUN makkar_c_Page_004.jp2
490f8b4d1d59353282a0f7d164d53610
9c6deeb65d53601335e18de638eddda1ec914e6b
39236 F20101119_AAACTZ makkar_c_Page_096.jpg
a2f5a032b2211a28054bfde61fee7b37
dfe26c1fe0681b64a11cfe34eec8e488e9059b2a
F20101119_AAADAI makkar_c_Page_057.tif
694864fc48de0567a20c6c887519ff1a
929a887457b6592f05d840e0caf92264b9cc44f7
105740 F20101119_AAACVB makkar_c_Page_021.jp2
d8a014a84f2c8b9fbd91801336db6c70
a86445a43293de77ed1d39f7df6ff2631a76a07e
1051981 F20101119_AAACUO makkar_c_Page_005.jp2
25e0d0dfb8f3b479491595fd614e24fc
5c81bd36e1cf82727882fcc86674ecb5817e3eba
92159 F20101119_AAACVC makkar_c_Page_022.jp2
9552e030e26796687009a5c0eee54d58
8a60d232efdda46a4104590790077321f5aa75af
880796 F20101119_AAACUP makkar_c_Page_006.jp2
7b3b5b6a967ef5bd23f4c676502485e5
408c63c1ae105c42c648bf0381e69c017b3ce776
F20101119_AAADAJ makkar_c_Page_058.tif
7d3c3bdda69846aacba8156b6a0ce261
28cb289c1c5e0d4dcfcd35ef6bc290a6355fb8b7
87344 F20101119_AAACVD makkar_c_Page_023.jp2
4eb41866506289a2384ad05b21f7c2b0
82dad818f7bf9e2a7cfce2d06d75f2cf64245a99
F20101119_AAADAK makkar_c_Page_059.tif
c19aca0265d8ba1dd5b90c92158c586a
0df5986505ec0f1aaee0a9dce3861802d43dd5e8
285513 F20101119_AAACVE makkar_c_Page_024.jp2
3081ef983987f138ab4b4b011621a36c
441bc0865cdce223bf3c802c30dd9e0a6e70ea1e
324937 F20101119_AAACUQ makkar_c_Page_007.jp2
c5926348a53481c53c9385c8898b0e47
da6fbf5233c9ada63c20d4c64a19a39db9bfdf08
F20101119_AAADBA makkar_c_Page_077.tif
91414f6464d7230338eb984918f5a4d4
90cead9585640faa7287b289d2c9d8fd78ddfbb1
F20101119_AAADAL makkar_c_Page_060.tif
7848f8ec84ff3f925be59d839add0012
acc0887efba61d63960823f99d2841474b3b4a75
383593 F20101119_AAACVF makkar_c_Page_025.jp2
f4e2c0b024c5d31504cd5a49c7bb0435
4d1b70bf69e348cd5df034a5012e237d890f7edd
1051984 F20101119_AAACUR makkar_c_Page_009.jp2
4365051a89116ecc0d41152493a1a1f1
aaa6991c04f936c548eef2c6318b734e6c7253e8
F20101119_AAADBB makkar_c_Page_078.tif
a96e3511b716a0befaa7aa17cd6f347b
400b058428603f88d448591e81c3bc73058a0241
F20101119_AAADAM makkar_c_Page_061.tif
d24838ac9ac69059664b7861f682b590
42d8d6713b6e1812ebc4fc4bd71bb57cf3eb32ea
834577 F20101119_AAACVG makkar_c_Page_026.jp2
432c896ef22102fb4322b6ed68025623
188f58ce72bf11bf6e60a16227923975b019dd71
1051983 F20101119_AAACUS makkar_c_Page_010.jp2
d7d0dc369762fe080530af69788c9f68
a76afa322e302b0450171229b586878b8c595f2f
F20101119_AAADBC makkar_c_Page_079.tif
c405affbdd1be0e61ac2755c076c29f8
7ed02f77dd42bfd1c066df0972e44364080fbb6d
F20101119_AAADAN makkar_c_Page_062.tif
c23b8615d23e5f2010b16eab09a7d12a
a51a244d5d414c1fb357b9ad66c44d6a29cf8e6c
851333 F20101119_AAACVH makkar_c_Page_027.jp2
371f74f0af27bec323ed3039e31251de
bf2051d4bf3a89846c1455d4daf64c0bc03f5b8c
94848 F20101119_AAACUT makkar_c_Page_011.jp2
ce857c5293dc3592c7740f476f3bc70f
1925ad2c566fa8d80774951f1bee06dc68545f65
F20101119_AAADBD makkar_c_Page_080.tif
36b804d3d3175ac5fd512d1c8005f016
0515ceaa8f26211d8cbdb5422af10f10c11bd760
F20101119_AAADAO makkar_c_Page_063.tif
8c605ef01a4e3402909c64743d51b7e8
5635e86f569cafd424643b2d13cb6409dceb0b96
40892 F20101119_AAACVI makkar_c_Page_028.jp2
4284f685e4dd525dafe7ef519e195b6c
f24a8488934f2c00e7c9a7e93c5763758e510b7a
88868 F20101119_AAACUU makkar_c_Page_012.jp2
e9df59c3a8028f50b1e158ca1a3d8150
2677e124af3d886419f86a8bfe0619ebb9073ea7
F20101119_AAADBE makkar_c_Page_081.tif
9142ada064f0c648757ebb3332880fb7
12f8b1ffb1c01ec6d0f1443c6a753b4932907ea2
F20101119_AAADAP makkar_c_Page_064.tif
12e59dc82b19bb5e3a6af1ec8e88c148
a62ecf17e7b135ee30e9339a0e3d9d659ff0614e
314095 F20101119_AAACVJ makkar_c_Page_029.jp2
0894843339f18e50364f49c46748d678
c7bad3007bbf48b45bfc9e991726a5bb78b746ef
121186 F20101119_AAACUV makkar_c_Page_014.jp2
277af571f6065c1c08fa0e67ff33d0b8
59c0b6f62188b11f0daf2d33abe76632a175b606
F20101119_AAADBF makkar_c_Page_082.tif
9849facc2a84ca1f55c71b347c5e38b8
785d96b382044e0b5fe5f67d679dee37f7b2f5c2
F20101119_AAADAQ makkar_c_Page_066.tif
0693b50cf73811dbff0a009e3fcb1390
833c37df1f799d4ac758d33672301cf3160f38a2
33443 F20101119_AAACVK makkar_c_Page_030.jp2
b39da13294fa47c66dc84b302b964f97
b098204cf19921e7eb6e540bc57efc9c14d7250f
115758 F20101119_AAACUW makkar_c_Page_015.jp2
6854cc1f05dd7a6db247a39053ce617c
087a649271dd122b0c8986bd2d6f496e4c0a9f43
F20101119_AAADBG makkar_c_Page_083.tif
b2894896b9c4c38647d0a221d0e62615
2a41321d5970a2faaa54bd3b980deb87ad01ea4b
F20101119_AAADAR makkar_c_Page_067.tif
da239d566909f97d0ae874b75f41a193
bcd4557b6b252b0660a4dd281c0077f8504fd289
848406 F20101119_AAACVL makkar_c_Page_031.jp2
14111092b671a3d88f035d6081ab5a83
476226130c97da88f8798c7ce16e3b1bef249dd8
119964 F20101119_AAACUX makkar_c_Page_016.jp2
e66c673aaf823440f28879b6f7f04043
dab8be3b589f7715284db12be3a70e6b9da32a95
F20101119_AAADBH makkar_c_Page_084.tif
7e7efb994b16f518a9e53c288eb2e793
25dd784490ceba44a7d7ec83bdbf905a71cc60b8
66424 F20101119_AAACWA makkar_c_Page_046.jp2
993557575aac386b9a07bb7f9000de1c
4d27ee97834e4ad6cf2fe435ca29e05611268af8
F20101119_AAADAS makkar_c_Page_069.tif
6922da6f17831167f4f10a19381e4471
07095a237cbec26a41491e9b9cd4f3527125431e
437145 F20101119_AAACVM makkar_c_Page_032.jp2
7d027b2b9fc018430af4454e3c24ea9b
74ee0671d8c41c81ed809d9bea7840009d88f632
121524 F20101119_AAACUY makkar_c_Page_017.jp2
c8561133d2edb5e81244c40ff6a659e2
6a1f9cef154257754a8fe547611626845738ae00
F20101119_AAADBI makkar_c_Page_085.tif
4f1f4dc08f38ce948a222c4a1aa0fcd4
16c6d3725c7d2bc126f5d421e1df510dcc687738
85194 F20101119_AAACWB makkar_c_Page_047.jp2
32e9377f25b47c7176479d674ad7e18f
1182135d9e2c66eefa201454662013d1294784d2
F20101119_AAADAT makkar_c_Page_070.tif
1b218bb8d7f706e498a4aba81502057b
0d7a63aaadb89a2c65b510401e2fab2185deb766
594139 F20101119_AAACVN makkar_c_Page_033.jp2
cb6789bc2be8ca934acc69716f85efe5
f239e62503c958d3d7ad93d4e693fcbb0eae38e4
115027 F20101119_AAACUZ makkar_c_Page_019.jp2
35eac9af3302b41391bc009f94440d02
23df494788e5b383930e2a923e4cd35a1c78a8bb
F20101119_AAADBJ makkar_c_Page_086.tif
39571b8812ab31edf83d1a23f0fc3860
1182d1a0fb95e43e53610664f4cef4f711e355f2
70676 F20101119_AAACWC makkar_c_Page_048.jp2
c6a51f2488aa8ac3c0e3cf20334aec17
6f509de8e39cafec7209b78e0f6690d0991a8045
F20101119_AAADAU makkar_c_Page_071.tif
4a485b1cd2ad9d504df174a6c3517997
765e4f79981890ca6bbfe24b25ae4c513632ff27
415833 F20101119_AAACVO makkar_c_Page_034.jp2
338756a61924f1370ff2a3b605f8c7c6
1009e2a680159bc2390c16b1bcb09e73bb77d9cb



PAGE 1

NONLINEARMODELING,IDENTIFICATION,ANDCOMPENSATIONFOR FRICTIONALDISTURBANCES By CHARUMAKKAR ATHESISPRESENTEDTOTHEGRADUATESCHOOL OFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENT OFTHEREQUIREMENTSFORTHEDEGREEOF MASTEROFSCIENCE UNIVERSITYOFFLORIDA 2006

PAGE 2

Copyright2006 by CharuMakkar

PAGE 3

Thisworkisdedicatedtomyparentsfortheirunconditionallove,unquestioned supportandunshakenbeliefinme.

PAGE 4

ACKNOWLEDGMENTS Iwouldliketoexpresssinceregratitudetomyadvisor,Dr.WarrenE.Dixon, whoisapersonwithremarkablea ability.Asanadvisor,heprovidedthenecessaryguidanceandallowedmetodevelopmyownideas.Asamentor,hehelped meunderstandtheintricaciesofworkinginaprofessionalenvironmentandhelped developmyprofessionalskills.Ifeelfortunateingettingtheopportunitytowork withhim. IwouldalsoliketoextendmygratitudetoDr.W.G.Sawyerforhisvaluable discussions.IalsoappreciatemycommitteemembersDr.CarlD.CraneIII,Dr.T. F.Burks,andDr.J.Hammerforthetimeandhelptheyprovided. Iwouldliketothankallmyfriendsfortheirsupportandencouragement.I especiallythankVikasforbeingmystrengthandapillarofsupportforthelast twoyears.IwouldalsoliketothankmycolleagueKeithDupreeforhelpingme outonthosedi cuiltdayswhenIwasdoingmyexperiements,andotherwise aquaintingmewithAmericanculturealmosteveryday. FinallyIwouldliketothankmyparentsfortheirloveandinspiration,my sistersSoniaandMadhvi,mybrothers-in-lawAnilandRohitandmydarlingnieces Kovida,AdyaandHiaforkeepingupwihmeandlovingmeunconditionally. iv

PAGE 5

TABLEOFCONTENTS page ACKNOWLEDGMENTS.............................iv LISTOFTABLES........... ......................vii LISTOFFIGURES.... ............................viiiABSTRACT.............. ......................xi 1INTRODUCTION.. ............................1 2MODELINGOFNONLINEARUNCERTAINTY-FRICTION......9 2.1FrictionModelandProperties.....................10 2.2Stick-SlipSimulation..........................11 2.3ExperimentalResults..........................24 2.4ConcludingRemarks..........................25 3IDENTIFICATIONANDCOMPENSATIONFORFRICTIONBYHIGH GAINFEEDBACK.. ............................28 3.1DynamicModelandProperties....................29 3.2ErrorSystemDevelopment.......................30 3.3StabilityAnalysis............................32 3.4ExperimentalResults..........................37 3.4.1Experiment1..........................38 3.4.2Experiment2..........................44 3.4.3Experiment3..........................45 3.4.4Experiment4..........................51 3.4.5Experiment5..........................51 3.5Discussion.... ............................53 3.6ConcludingRemarks..........................56 4TRACKINGCONTROLINTHEPRESENCEOFFRICTIONBYHIGH GAINFEEDBACKANDAMODEL-BASEDFEEDFORWARDCOMPONENTANEXTENSION... ............................58 4.1DynamicModelandProperties....................59 4.2ErrorSystemDevelopment.......................60 4.3StabilityAnalysis............................64 v

PAGE 6

4.4ExperimentalResults..........................68 4.4.1Experiment1..........................70 4.4.2Experiment2..........................70 4.5Discussion.... ............................72 4.6ConcludingRemarks..........................74 5CONCLUSIONANDRECOMMENDATIONS...............75 APPENDIX AIDENTIFICATIONANDCOMPENSATIONFORFRICTIONBYHIGH GAINFEEDBACK.. ............................77 BTRACKINGCONTROLINTHEPRESENCEOFFRICTIONBYHIGH GAINFEEDBACKANDAMODEL-BASEDFEEDFORWARDCOMPONENTANEXTENSION... ............................84 REFERENCES....... ............................90 BIOGRAPHICALSKETCH............................95 vi

PAGE 7

LISTOFTABLES Table page 31Comparisonoftrackingresultswhennoexternalloadwasappliedtothe circulardisk...... ............................54 32Comparisonoftrackingresultswhenanexternalloadwasappliedtothe circulardisk...... ............................54 vii

PAGE 8

LISTOFFIGURES Figure page 21Frictionmodelasacompositionofdi erente ectsincluding:a)Stribeck e ect,b)viscousdissipation,c)Coulombe ect,andd)thecombined model................ ......................12 22Characteristicsofthefrictionmodel.....................12 23Modularabilityofthemodeltoselectivelymodeldi erentfrictionregimes: topplot-viscousregime(e.g.,hydrodynamiclubrication),middleplotCoulombicfrictionregime(e.g.,solidlubricantcoatingsatmoderateslidingspeeds),andbottomplot-abruptchangefromstatictokineticfriction(e.g.,non-lubricouspolymers)......................13 24Mass-springsystemfordemonstratingstick-slipfriction..........14 2Frictioncoe cientvsslipvelocity......................15 26Blockvelocityvstime.............................16 27Slipvelocityvstime..............................16 2Frictioncoe cientvstime...........................17 2Frictioncoe cientvsslipvelocity......................17 210Blockvelocityvstime.............................18 211Slipvelocityvstime..............................18 2Frictioncoe cientvstime...........................19 2Frictioncoe cientvsslipvelocity......................20 214Blockvelocityvstime.............................20 215Slipvelocityvstime..............................21 2Frictioncoe cientvstime...........................21 2Frictioncoe cientvsslipvelocity......................22 218Blockvelocityvstime.............................22 219Slipvelocityvstime..............................23 viii

PAGE 9

2Frictioncoe cientvstime...........................23 221Testbedfortheexperiment..........................25 222Theproposedfrictionmodelverycloselyapproximatestheexperimentallyidenti edfrictiontermobtainedfromtheadaptivecontrollerdevelopedinMakkaretal.[41].Topplotdepictstheexperimentallyobtained frictiontorque,middleplotdepitcsthefrictionplotobtainedfromthe proposedmodelandthebottomplotdepictsacomparisonofthetwo withsolidlineindicatingexperimentallyobtainedfrictiontorqueand dashedlineindicatingfrictiontorqu eobtainedfromtheproposedfrictionmodel....... ............................26 31Desireddisktrajectory.............................38 32PositiontrackingerrorfromthePDcontroller................40 33Positiontrackingerrorfromthemodel-basedcontrollerwithfrictionfeedforwardtermsasdescribedin(3).... .................40 34Positiontrackingerrorfromtheproposedcontroller............41 3Comparisonofpositiontrackingerro rsfromthethreecontrolschemes..41 3Comparisonofpositiontrackinge rrorsfromthemodel-basedcontroller andtheproposedcontroller..........................42 37TorqueinputbythePDcontroller......................42 38Torqueinputbythemodel-basedcontrollerwithfrictionfeedforward termsasdescribedin(340)..........................43 39Torqueinputbytheproposedcontroller...................43 3Identi edfrictionfromtheadaptivetermintheproposedcontroller....44 311PositiontrackingerrorfromthePDcontroller................45 312Positiontrackingerrorfromthemodel-basedcontrollerwithfrictionfeedforwardtermsasdescribedin(3).... .................46 313Positiontrakingerrorfromtheproposedcontroller.............46 34Comparisonofpositiontrackingerrorsfromthethreecontrolschemes..47 3Comparisonofpositiontrackingerrorsfromthemodel-basedcontroller andtheproposedcontroller..........................47 316TorqueinputbythePDcontroller......................48 ix

PAGE 10

317Torqueinputbythemodel-basedcontrollerwithfrictionfeedforward termsasdescribedin(340)..........................48 318Torqueinputbytheproposedcontroller...................49 3Identi edfrictionfromtheadaptivetermintheproposedcontroller....49 30Positiontrackingerrorwiththeproposedcontrollerwhenthecircular diskwaslubricated. .................. ............50 3Torqueinputbytheproposedcont rollerwhenthecirculardiskwaslubricated......... ............................50 3Identi edfrictionfromtheadaptivetermintheproposedcontrollerwhen thecirculardiskwaslubricated........................51 323Netexternalfrictioninducedwithnolubrication.Thenetfrictionwas calculatedbysubractingtheidenti edfrictionterminExperiment1from theidenti edfrictionterminExperiment2.................52 324Netexternalfrictioninducedwithlubrication.Thenetfrictionwascalculatedbysubractingtheidenti edfrictionterminExperiment1from theidenti edfrictionterminExperiment3.................52 3Thefrictiontorquecalculatedfro mthemodelin(2)approximatesthe experimentallyidenti edfrictiontorquein(39)..............53 326WearingoftheNylonblockwhereitrubbedagainstthecirculardisk...56 327Wearonthecirculardisk...........................57 41Theexperimentaltestbedconsistsofa1-linkrobotmountedonaNSK direct-driveswitchedreluctancemotor....................68 42Desiredtrajectoryusedfortheexperiment..................70 43Positiontrackingerrorwhentheadaptivegainiszero...........71 44Torqueinputwhentheadaptivegainiszero.................71 4Positiontrackingerrorforthecontrolstructurethatincludestheadaptiveupdatelaw..... ............................72 4Torqueinputforthecontrolstruct urethatincludestheadaptiveupdate law...................... .................73 47Parameterestimateforthemassofthelinkassembly............73 x

PAGE 11

AbstractofThesisPresentedtotheGraduateSchool oftheUniversityofFloridainPartialFul llmentofthe RequirementsfortheDegreeofMasterofScience NONLINEARMODELING,IDENTIFICATION,ANDCOMPENSATIONFOR FRICTIONALDISTURBANCES By CharuMakkar May2006 Chair:WarrenE.Dixon MajorDepartment:MechanicalandAerospaceEngineering Forhigh-performanceengineeringsystems,model-basedcontrollersare typicallyrequiredtoaccommodateforthes ystemnonlinearities.Unfortunately, developingaccuratemodelsforfrictionhasbeenhistoricallychallenging.Despite opendebatesinTribologyregardingthecontinuityoffriction,typicalmodels developedsofararepiecewisecontinuousordiscontinuous.Motivatedbythefact thatdiscontinuousandpiecewisecontinuousfrictionmodelscanbeproblematic forthedevelopmentofhigh-performancecontrollers,anewmodelforfrictionis proposed.Thissimplecontinuouslydi erentiablemodelrepresentsafoundation thatcapturesthemajore ectsreportedanddiscussedinfrictionmodelingand experimentation.Theproposedmodelisgen ericenoughthatothersubtletiessuch asfrictionalanisotropywithslidingdirectioncanbeaddressedbymathematically distortingthismodelwithoutcompromisingthecontinuousdi erentiability.From literature,itisknownthatifthefrictione ectsinthesystemcanbeaccurately modeled,thereisanimprovedpotentialtodesigncontrollersthatcancancel thee ects,whereasexcessivesteady-statetrackingerrors,oscillations,andlimit cyclescanresultfromcontrollersthatdonotaccuratelycompensateforfriction.

PAGE 12

AtrackingcontrollerisdevelopedinChapter3forageneralEuler-Lagrange systembasedonthedevelopedcontinuouslydi erentiablefrictionmodelwith uncertainnonlinearparameterizableterms.Toachievethesemi-globalasymptotic trackingresult,arecentlydevelopedintegralfeedbackcompensationstrategyis usedtoidentifythefrictione ectson-line,assumingexactmodelknowledgeofthe remainingdynamics.ALyapunov-basedstabilityanalysisisprovidedtoconclude thetrackingandfrictionidenti cationresults.Experimentalresultsillustratethe trackingandfrictionidenti cationperformanceofthedevelopedcontroller. ThetrackingresultinChapter3isfurtherextendedtoincludesystemswith unstructureduncertaintieswhileeliminatingtheknowndynamicsassumption. Thegeneraltrendforpreviouscontrolstrategiesdevelopedforuncertaindynamics innonlinearsystemsisthatthemoreunstructuredthesystemuncertainty,the morecontrole ort(i.e.,highgainorhighfrequencyfeedback)isrequiredtoreject theuncertainty,andtheresultingstabilityandperformanceofthesystemare diminished(e.g.,uniformlyultimatelyboundedstability).TheresultinChapter 4isthe rstresultthatillustrateshowtheamalgamationofanadaptivemodelbasedfeedforwardtermwithahighgainintegralfeedbacktermcanbeusedto yieldanasymptotictrackingresultforsystemsthathavemixedunstructuredand structureduncertainty.Experimentalresultsareprovidedthatillustrateareduced rootmeansquaredtrackingerror. xii

PAGE 13

CHAPTER1 INTRODUCTION TheclassofEuler-Lagrangesystemsconsideredinthisthesisaredescribedby thefollowingnonlineardynamicmodel: ( )¨ + ( ) + ( )+ ( )= ( ) (1—1) In(1—1), ( ) R denotestheinertiamatrix, ( ) R denotesthe centripetal-Coriolismatrix, ( ) Rdenotesthegravityvector, ( ) Rdenotesfrictionvector, ( ) Rrepresentsthetorqueinputcontrolvector, and ( ) ( ) ¨ ( ) Rdenotethelinkposition,velocity,andacceleration vectors,respectively.Forhigh-performanceengineeringsystems,model-based controllers(seeDixonetal.[16])aretypicallyusedtoaccommodateforthesystem nonlinearities.Ingeneral,eitheraccuratemodelsoftheinertiale ectscanbe developedornumerouscontinuousadaptiveandrobustcontrolmethodscanbe appliedtomitigatethee ectsofanypotentialmismatchintheinertialparameters. Unfortunately,developingaccuratemodelsforfrictionhasbeenhistorically problematic.Infact,aftercenturiesoftheoreticalandexperimentalinvestigation, ageneralmodelforfrictionhasnotbeenuniversallyaccepted,especiallyatlow speedswherefrictione ectsareexaggerated.Infact,Armstrong-Helouvry[1] examinedthedestabilizinge ectsofcertainfrictionphenomena(i.e.,theStribeck e ect)atlowspeeds.Tofurthercomplicatethedevelopmentofmodel-based controllersforhigh-performancesystems,frictionisoftenmodeledasdiscontinuous; thus,requiringmodel-basedcontrollerstobediscontinuoustocompensateforthe e ects. 1

PAGE 14

2 Motivatedbythedesiretodevelopanaccuraterepresentationoffrictionin systems,variouscontrolresearchershavedevelopeddi erentanalyticalmodels, estimationmethodstoidentityfrictione ects,andadaptiveandrobustmethods tocompensatefororrejectthefrictione ects.Ingeneral,thedominantfriction componentsthathavebeenmodeledinclude: staticfriction (i.e.,thetorque thatopposesthemotionatzerovelocity), Coulombfriction (i.e.,theconstant motionopposingtorqueatnon-zerovelocity), viscousfriction (i.e.,whenfull uidlubricationexistsbetweenthecontactsurfaces), asymmetries (i.e.,di erent frictionbehaviorfordi erentdirectionsofmotion), Stribecke ect (i.e.,atverylow speed,whenpartial uidlubricationexists,contactbetweenthesurfacesdecreases andthusfrictiondecreasesexponentiallyfromstiction),and positiondependence (oscillatorybehaviorofthefrictiontorqueduetosmallimperfectionsonthemotor shaftandreductorcenters,aswellastheelasticdeformationofballbearings). Classicalfrictionmodelsarederivedfromstaticmapsbetweenvelocityand frictionforce.Fromacomprehensivesurveyoffrictionmodelsincontrolliterature (seeArmstrong-Helouvry[1]andArmstrong-Helouvryetal.[2]),someresearchers believethatdynamicfrictione ectsarenecessarytocompletethefrictionmodel. Severaldynamicfrictionmodelshavebeenproposed(seeBlimanandSorine[5] andCanudasdeWitetal.[8]).ThesemodelscombinetheDahlmodel(seeDahl [14])withthearbitrarysteady-statefrictioncharacteristicsofthebristle-based LuGremodelproposedbyCanudasdeWitetal.[8].Arecentmodi cationtothe LuGremodelisgivenintheLeuvenmodelbySweversetal.[52]thatincorporates ahysteresisfunctionwithnon-localmemoryunliketheLu-Gremodel.TheLeuven modelwaslaterexperimentallycon rmedbyFerrettietal.[21].However,a modi cationtotheLeuvenmodelisprovidedbyLampaertetal.[34]thatreplaces thestackmechanismusedtoimplementthehysteresisbythemoree cient Maxwellslipmodel.AnothercriticismtotheLuGremodelhasbeenrecently

PAGE 15

3 raisedbyDupontetal.[18],whounderlineanonphysicaldriftphenomenonthat ariseswhentheappliedforceischaracterizedbysmallvibrationsbelowthestatic frictionlimit.Recentlysingleandmultistateintegralfrictionmodelshavebeen developedbyFerrettiandMagnanietal.[22]basedontheintegralsolutionofthe Dahlmodel.However,thesefrictionmodelsarebasedontheassumptionthatthe frictioncoe cientisconstantwithslidingspeedandhaveasingularityattheonset ofslip.Unfortunately,eachoftheaforementionedmodelsarediscontinuous(i.e.,a signumfunctionofthevelocityisusedtoassignthedirectionoffrictionforcesuch astheresultsbyDupontetal.[18],Ferrettietal.[22],Lampaertetal.[34],and Sweversetal.[52]),andmanyothermodelsareonlypiecewisecontinuous(e.g.,the LuGremodelin[8]).Asstatedpreviously,theuseofdiscontinuousandpiecewise continuousfrictionmodelsisproblematicforthedevelopmentofhigh-performance continuouscontrollers. Chapter2andthepreliminarye ortsbyMakkaretal.[42]andMakkar andDixonetal.[43]providea rststepatcreatingacontinuouslydi erentiable frictionmodelthatcapturesanumberofessentialaspectsoffrictionwithout involvingdiscontinuousorpiecewisecontinuousfunctions.Theproposedmodel isgenericenoughthatothersubtletiessuchasfrictionalanisotropywithsliding directioncanbeaddressedbymathematicallydistortingthismodelwithout compromisingthecontinuousdi erentiability. Ifthefrictione ectsinasystemcanbeaccuratelymodeled,thereisan improvedpotentialtodesigncontrollersthatcancancelthee ects(e.g.,modelbasedcontrollers);whereas,excessivestea dy-statetrackingerrors,oscillations, andlimitcyclescanresultfromcontrollersthatdonotaccuratelycompensatefor friction.Giventhepastdi cultyinaccuratelymodelingandcompensatingfor frictione ects,researchershaveproposedavarietyof(typicallyo ine)friction estimationschemeswiththeobjectiveofidentifyingthefrictione ects.For

PAGE 16

4 example,ano inemaximumlikelihood,frequency-basedapproach(di erential binaryexcitation)isproposedbyChenetal.[12]toestimateCoulombfriction e ects.Anotherfrequency-basedo inefrictionidenti cationapproachwas proposedbyKimandHa[31].Speci cally,theapproachbyKimandHa[31]uses akindoffrequency-domainlinearregressionmodelderivedfromFourieranalysis oftheperiodicsteady-stateoscillationsofthesystem.TheapproachbyKim andHa[31]requiresaperiodicexcitationinputwithsu cientlylargeamplitude and/orfrequencycontent.Anewo inefrictionidenti cationtoolisproposed byKimetal.[32]wherethestatic-frictionmodelsarenotrequiredtobelinear parameterizable.Howevertheo ineoptimizationresultbyKimetal.[32]is limitedtosingledegree-of-freedomsystemswheretheinitialand nalvelocity areequal.Anotherfrequencydomainidenti cationstrategydevelopedtoidentify dynamicmodelparametersforpreslidingbehaviorisgivenbyHensenandAngelis etal.[26].Additionalidenti cationmethodsincludeleast-squaresasdevelopedby CanudasdeWitandLichinsky[10]andKalman lteringbyHensenetal.[27]. Inadditiontofrictionidenti cationschemes,researchershavedeveloped adaptive,robust,andlearningcontrol lerstoachieveacontrolobjectivewhile accommodatingforthefrictione ects,butnotnecessarilyidentifyingfriction.For example,givenadesiredtrajectorythatisperiodicandnotconstantoversome intervaloftime,thedevelopmentbyChoetal.[13]providesalearningcontrol approachtodampoutperiodicsteady-stateoscillationsduetofriction.Asstated byChoetal.[13],aperiodicsignalisappliedtothesystemandwhenthesystem reachesasteady-stateoscillation,thelearningupdatelawisapplied.Liaoet al.[37]proposedadiscontinuouslinearizingcontrolleralongwithanadaptive estimatortoachieveanexponentiallystabletrackingresultthatestimatesthe unknownCoulombfrictioncoe cient.However,ZhangandGuay[60]describe atechnicalerrorintheresultpresentedbyLiaoetal.[37]thatinvalidatesthe

PAGE 17

5 result.AdditionaldevelopmentisprovidedbyZhangandGuay[60]thatmodi es theresultbyLiaoetal.[37]toachieveasymptoticCoulombfrictioncoe cient estimationprovidedapersistenceofexcitationconditionissatis ed.Tomeiin[53] proposedarobustadaptivecontrollerwhereonlyinstantaneousfrictionistaken intoaccount(dynamicfrictione ectsarenotincluded). Motivatedbythedesiretoincludedynamicfrictionmodelsinthecontrol design,numerousresearchershaveembracedtheLuGrefrictionmodelproposed byCanudasdeWitetal.[8].Forexample,theresultbyTomei[53]wasextended in[54]toincludetheLuGrefrictionmodelproposedbyCanudasdeWitetal. [8],resultinginanasymptotictrackingresultforsquareintegrabledisturbances. RobustadaptivecontrollerswerealsoproposedbyJainetal.[29]andSivakumar andKhorrami[50]toaccountfortheLuGremodel.Canudasetal.investigated thedevelopmentofobserver-basedapproachesfortheLuGremodelin[8].Canudas andLichinskyin[9]proposedanadaptivefrictioncompensationmethod,and CanudasandKellyin[11]proposedapassivity-basedfrictioncompensationterm toachieveglobalasymptotictrackingusingtheLuGremodel.Barabanovand Ortegain[4]developednecessaryandsu cientconditionsforthepassivityofthe LuGremodel.Threeobserver-basedcontrolschemeswereproposedbyVedagarbda etal.[56]assumingexactmodelknowledgeofthesystemdynamics.Theresults byVedagarbdaetal.[56]werelaterextendedtoincludetwoadaptiveobservers toaccountforselecteduncertaintyinthemodel.Theobserver-baseddesignin Vedagarbdaetal.[56]wasfurtherextendedbyFeemesteretal.[20].Speci cally, apartial-statefeedbackexactmodelknowledgecontrollerwasdevelopedtoachieve globalexponentiallinkpositiontrackingofarobotmanipulatorbyFeemsteretal. [20].Twoadaptive,partial-statefeedbackglobalasymptoticcontrollerswerealso proposedin[20]thatcompensateforselecteduncertaintyinthesystemmodel.In addition,anewadaptivecontroltechniquewasproposedbyFeemsteretal.[20]to

PAGE 18

6 compensateforthenonlinearparameterizableStribecke ect,wheretheaverage squareintegralofthepositiontrackingerrorswereforcedtoanarbitrarilysmall value. InChapter3andinthepreliminaryresultsbyMakkaretal.[40]andMakkar andDixonetal.[41],atrackingcontrollerisdevelopedforageneralEulerLagrangesystemthatcontainsthenewcontinuouslydi erentiablefrictionmodel withuncertainnonlinearparameteriza bletermsthatwasdevelopedinChapter 2.Thecontinuouslydi erentiablepropertyoftheproposedmodelenabledthe developmentofanewidenti cationschemebasedonanewintegralfeedback compensationterm.Asemi-globaltrackingresultisachievedwhileidentifyingthe frictionon-line,assumingexactmodelknowledgeoftheremainingdynamics. ThecontroldevelopmentinChapter3isbasedontheassumptionofexact modelknowledgeofthesystemdynamicsexceptfriction.Thecontrolofsystems withuncertainnonlineardynamics,however,isstillamuchresearchedareaof focus.Forsystemswithdynamicuncertaintiesthatcanbelinearparameterized, avarietyofadaptive(e.g.,seeKrstic[33],SastryandBodson[47],andSlotineet al.[51])feedforwardcontrollerscanbeutilizedtoachieveanasymptoticresult. Somerecentadaptivecontrolresultshavealsotargetedtheapplicationofadaptive controllersfornonlinearlyparameterizedsystems(seeLinandQian[38]).Learning controllershavebeendevelopedforsystemswithperiodicdisturbances(see Antsaklisetal.[3]),andrecentresearchhasfocusedontheuseofexosystems bySerranietal.[49]tocompensatefordisturbancesthatarethesolutionofa lineartime-invariantsystemwithunknowncoe cients.Avarietyofmethods havealsobeenproposedtocompensateforsystemswithunstructureduncertainty including:variousslidingmodecontrollers(e.g.,seeSlotineandLi[51],andUtkin [55]),robustcontrolschemes(seeQu[46]),andneuralnetworkandfuzzylogic controllers(seeLewisetal.[36]).Fromareviewoftheseapproachesageneral

PAGE 19

7 trendthatcanbedeterminedisthatcontrollersdevelopedforsystemswithmore unstructureduncertaintywillrequiremorecontrole ort(i.e.,highgainorhigh frequencyfeedback)andyieldreducedperformance(e.g.,uniformlyultimately boundedstability). Asigni cantoutcomeofthenewcontrolstructuredevelopedbyXianand Dawsonetal.[57]isthatasymptoticstabilityisobtaineddespiteafairlygeneral uncertaindisturbance.ThistechniquewasusedbyCaietal.[7]todevelopa trackingcontrollerfornonlinearsystemsinthepresenceofadditivedisturbances andparametricuncertaintiesundertheassumptionthatthedisturbancesare 2withboundedtimederivatives.Xianetal.[58]utilizedthisstrategytopropose anewoutputfeedbackdiscontinuoustrackingcontrollerforageneralclassof nonlinearmechanical(i.e.,second-order)systemswhoseuncertaindynamics are rst-orderdi erentiable.Zhangetal.[59]combinedthehighgainfeedback structurewithahighgainobserveratthesacri ceofyieldingasemi-global uniformlyultimatelyboundedresult.Thisparticularhighgainfeedbackmethod hasalsobeenusedasanidenti cationtechnique.Forexample,themethodhas beenappliedtoidentifyfriction(seeMakkaretal.[40]andMakkarandDixonet al.[41]),forrangeidenti cationinperspectiveandparacatadioptricvisionsystems (e.g.,seeDixonetal.[17],andGuptaetal.[25]),andforfaultdetectionand identi cation(e.g.,seeMcIntyreetal.[44]). TheresultinChapter4andthepreliminaryresultsinPatreetal.[45]is motivatedbythedesiretoincludesomeknowledgeofthedynamicsinthecontrol designasameanstoimprovetheperformanceandreducethecontrole ortwhile eliminatingtheassumptionthatthedynamicsofthesystemiscompletelyknown. Forsystemsthatincludesomedynamicsthatcanbesegregatedintostructured (i.e.,linearparameterizable)andunstructureduncertainty,thisresultillustrates howanewcontroller,errorsystem,andstabilityanalysiscanbecraftedtoinclude

PAGE 20

8 amodel-basedadaptivefeedforwardterminconjunctionwiththehighgainintegral feedbacktechniquetoyieldanasymptotictrackingresult.Thischapterpresents the rstresultthatillustrateshowtheamalgamationofthesecompensation methodscanbeusedtoyieldanasymptoticresult.Experimentalresultsare presentedtoreinforcetheseheuristicnotions.

PAGE 21

CHAPTER2 MODELINGOFNONLINEARUNCERTAINTY-FRICTION Frictionforcediscontinuitieshavebeendebatedforatleastthreecenturies datingbacktothepublishedworksbyAmontonin1699(e.g.,seetheclassictext byBowdenandTabor[6]).Modernpublicationsassumetheslidingmotionbetween solidsoccursatalargenumberofverysmallanddiscretecontacts.Incontrast topopularandsimplemodelsthatassumeastructuralinteractionacrossregular andrepeatingsurfacefeatures,thecontactacrossengineeringsurfacesisknown tooccuronthetopsofasperitiesorsurfaceprotuberances,which,likefractals, aredistributedacrossalllengthscales.Thenumberofcontactsforengineering systemsisenormous,whichledtotheseminalworkbyGreenwoodandWilliamson [24],whotreatedthedistributionsofthesecontactsusingstatisticaldistributions. Thesefunctionswereintegratedtogiveacontinuousexpressionfortherelationship betweencontactareaandpressure.Duringsliding,andinparticularatthe initiationofgrossmotion(i.e.,pre-sliding),thedynamicsofindividualasperity contactsbreakingandformingisofgreattheoreticalinterest;however,dueto thelargenumberofcontactsinengineeringsystems,thedynamicsaretreatedas continuousfollowingclassicalstatisticalmethods. Thischapterprovidesa rststepatcreatingacontinuouslydi erentiable frictionmodelthatcapturesanumberofessentialaspectsoffrictionwithout involvingdiscontinuousorpiecewisecontinuousfunctions.Inunlubricatedor boundarylubricatedsliding,wearisinevitable.Theproposedfrictionmodel containstime-varyingcoe cientsthatcanbedeveloped(e.g.,modeledbya di erentialequation)tocapturespatiallyandtemporallyvaryinge ectsdueto wear.Thiscontinuouslydi erentiablemodelrepresentsafoundationthatcaptures 9

PAGE 22

10 themajore ectsreportedanddiscussedinfrictionmodelingandexperimentation. Theproposedmodelisgenericenoughthatothersubtletiessuchasfrictional anisotropywithslidingdirectioncanbeaddressedbymathematicallydistorting thismodelwithoutcompromisingthecontinuousdi erentiability. Thischapterisorganizedasfollows.Thefrictionmodelandtheassociated propertiesareprovidedinSection2.1.Thegeneralityofthemodelisdemonstrated throughanumericalsimulationinSection2.2.Speci cally,numericalsimulations areprovidedfordi erentfrictionmodelparameterstoillustratethedi erent e ectsthatthemodelcaptures.Section2.3describesthatthedevelopedmodel approximatestheexperimentalresultsobtainedinMakkaretal.[41]. 2.1FrictionModelandProperties Theproposedmodelforthefrictionterm ( ) in(1—1): ( )= 1(tanh( 2 ) tanh( 3 ))+ 4tanh( 5 )+ 6 (2—1) where R =1 2 6 denoteunknownpositiveconstants1.Thefrictionmodel in(2—1)hasthefollowingproperties. • Itiscontinuouslydi erentiableandnotlinearparameterizable. • Itissymmetricabouttheorigin. • Thestaticcoe cientoffrictioncanbeapproximatedbytheterm 1+ 4. • Theterm tanh( 2 ) tanh( 3 ) capturestheStribecke ectwherethefriction coe cientdecreasesfromthestaticcoe cientoffrictionwithincreasingslip velocityneartheorigin. • Aviscousdissipationtermisgivenby 6 • TheCoulombicfrictionispresentintheabsenceofviscousdissipationandis modeledbytheterm 4tanh( 5 ) 1Theseparameterscouldalsobetime-varying.

PAGE 23

11 • Thefrictionmodelisdissipativeinthesensethatapassiveoperator ( ) ( ) satis esthefollowingintegralinequality[4] Z 0 ( ) ( ( )) 2where isapositiveconstant,provided ( ) isbounded. Figures2—1and2—2illustratethesumofthedi erente ectsandcharacteristicsofthefrictionmodel.Figure2—3showsthe exibilityofsuchamodel. 2.2Stick-SlipSimulation Thequalitativemechanismsoffrictionarewell-understood.Toillustrate howthefrictionmodelpresentedin(2—1)exhibitsthesee ects,variousnumerical simulationsarepresentedinthissection.ThesystemconsideredinFigure2—4 isasimplemass-springsystem,inwhichaunitmass isattachedtoaspring withsti ness restingonaplatemovingwithavelocity( ) inthepositive direction,whichcausestheblocktomovewithavelocity( ) inthesame direction.Themodeledsystemcanbecomparedtoamassattachedtoa xed springmovingonaconveyorbelt.Theplateismovingwithavelocitythatslowly increasesandsaturates,givenbythefollowingrelation:=1 0 1 ThesystemdescribedbyFigure2—4ismodelledasfollows: ( )+ ( ) (( ) ( ))=0 wheretheterm( ) ( ) representstheslipvelocity,(i.e.,thedi erencebetween theplatevelocityandblockvelocityatanyinstantoftime).Todemonstratethe exibilityofthemodel,modelparameterswerevariedinordertocapturethe Stribecke ect,Coulombicfrictione ectandviscousdissipation.Forexample,

PAGE 24

12 Figure2—1:Frictionmodelasacompositionofdi erente ectsincluding:a) Stribecke ect,b)viscousdissipation,c)Coulombe ect,andd)thecombined model. Figure2—2:Characteristicsofthefrictionmodel.

PAGE 25

13 Figure2—3:Modularabilityofthemodeltoselectivelymodeldi erentfriction regimes:topplot-viscousregime(e.g.,hydrodynamiclubrication),middleplotCoulombicfrictionregime(e.g.,solidlubricantcoatingsatmoderatesliding speeds),andbottomplot-abruptchangefromstatictokineticfriction(e.g.,nonlubricouspolymers).

PAGE 26

14 Figure2—4:Mass-springsystemfordemonstratingstick-slipfriction. hydrodynamiclubricationinmanyoperatingregimesisviscous,lackingtheother e ects,whichareeasilysettozerointhemodel.SimpleCoulombicfrictionmodels areoftengoodforsolidlubricantcoatingsatmoderateslidingspeeds.Tocapture thise ect,thestaticandviscoustermscanbesettozero.Forsomestickyornonlubricouspolymers,thereexistsanabruptchangefromstatictokineticfriction, whichiscapturedbymakingtheStribeckdecayveryrapid. ACoulombicfrictionregimeisdisplayedinFigure2—5wherethefriction modelparametersin(2—1)weresetasfollows: 1=0 2=0 3=0 4=0 1 5=100 6=0 .TheCoulombicfrictioncoe cientisaconstant,opposing themotionoftheblockasseeninFigures2—5and2—8.Theblockvelocity,slip velocity,andthefrictionforceasafunctionoftimearedepictedinFigures.2—62—8.These guresindicatethattheblockvelocityslowlyrises,reachesamaximum andthenbeginstooscillate.Theslipvelocityalsorisesandthenoscillatesafter reachingamaximumvalue.These guresindicatethatthefrictionforcecauses theblocktomovealongwiththeplateuntilthespringforceovercomesthefriction force;hence,theblockbeginstoslipinanoppositedirectionoftheplatevelocity

PAGE 27

15 0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.1 0.05 0 0.05 0.1 [m/sec]Figure2—5:Frictioncoe cientvsslipvelocity. causingthespringtocompress.Asthespringreleasesenergybackintothesystem, theblockvelocityexceedstheplatevelocity.Themagnitudeoftheconstant frictioncoe cientresultsinaconstantoscillationbetweenthefrictionforceand thespringforce. TheviscousfrictionplotinFigure(2—9)isobtainedbyadjustingtheparametersasfollows: 1=0 2=0 3=0 4=0 5=0 6=0 01 .Theblock velocity,slipvelocityandthefrictionforcearegiveninFigures2—10-2—12.The blockvelocityinFigure2—10slowlydecreasesastheviscousfrictionincreasesas displayedinFigure2—9.Theviscouscoe cientoffrictionisanorderofmagnitude smallerincomparisontotheCoulombicfrictioncoe cient,asaresultthefriction forceisnotsu cientenoughtosustaintheoscillationsoftheblock.Theblock eventuallycomestorestandconstantlyslipsonthemovingplate. TheStribecke ectin(2—13)ismodeledusingthefollowingfrictionmodel parametervalues: 1=0 25 2=100 3=10 4=0 5=0 6=0 .The blockvelocity,slipvelocityandthefrictionforceareplottedinFigures2—14-2—16. AsseeninFigure2—13,theStribecke ectisseenasthehighbreakawayforceat

PAGE 28

16 0 10 20 30 40 50 60 70 80 90 10 0 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 [sec][m/sec]Figure2—6:Blockvelocityvstime. 0 10 20 30 40 50 60 70 80 90 100 0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 [sec][m/sec] Figure2—7:Slipvelocityvstime.

PAGE 29

17 0 20 40 60 80 100 0.1 0.05 0 0.05 0.1 [sec]Figure2—8:Frictioncoe cientvstime. 0.2 0 0.2 0.4 0.6 0.8 1 1.2 2 0 2 4 6 8 10 12 x 103 [m/sec]Figure2—9:Frictioncoe cientvsslipvelocity.

PAGE 30

18 0 10 20 30 40 50 60 70 80 90 100 11 0 0.1 0.08 0.06 0.04 0.02 0 0.02 0.04 0.06 0.08 0.1 [sec][m/sec]Figure2—10:Blockvelocityvstime. 0 10 20 30 40 50 60 70 80 90 100 0.2 0 0.2 0.4 0.6 0.8 1 1.2 [sec][m/sec] Figure2—11:Slipvelocityvstime.

PAGE 31

19 0 20 40 60 80 100 2 0 2 4 6 8 10 12 x 103 [sec]Figure2—12:Frictioncoe cientvstime. thebeginningofthemotionoftheblock,whichthenexponentiallydecreases. Theblockmoveswiththeplateduetotheinitialfrictionforce.Eventually enoughenergyisstoredinthespringsothatthespringforceovercomesthe breakawayfriction.Thefrictionforceexponentiallydecaysafterthebreakaway forceisreached.Aftertheblockovercomesthebreakawayforce,thespringforce becomesdominantandcausestheblocktoreverseitsdirectionandmovetowards thespring.Sincethereisnoforceopposingthismotion,alargeslipvelocityis exhibitedwhilethespringcompressesandreleases,pushingtheblockfasterthan theplate. Figure2—17illustratesthesumofthedi erente ectsinthefrictionmodelina stick-slipregimewiththefollowingfrictionmodelparameters 1=0 25 2=100 3=10 4=0 1 5=100 6=0 01 .Figures2—18-2—20showthevelocityof theblock,slipvelocityandfrictionforceasafunctionoftime.Theblockvelocity slowlyincreasesandstoresenoughenergytoreversethedirectionoftheblock. Thisstorageandreleaseofenergycausesthesystemtooscillate.Figure2—20shows thestick-slipphenomenon.

PAGE 32

20 3 2 1 0 1 2 3 4 5 0.25 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 0.2 0.25 [m/sec]Figure2—13:Frictioncoe cientvsslipvelocity. 0 10 20 30 40 50 60 70 80 90 100 4 3 2 1 0 1 2 3 [sec][m/sec] Figure2—14:Blockvelocityvstime.

PAGE 33

21 0 10 20 30 40 50 60 70 80 90 10 0 2 1 0 1 2 3 4 5 [sec][m/sec]Figure2—15:Slipvelocityvstime. 0 10 20 30 40 50 60 70 80 90 100 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 0.2 [sec]Figure2—16:Frictioncoe cientvstime.

PAGE 34

22 0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 [m/sec]Figure2—17:Frictioncoe cientvsslipvelocity. 0 10 20 30 40 50 60 70 80 90 100 3 2.5 2 1.5 1 0.5 0 0.5 1 1.5 [sec][m/sec] Figure2—18:Blockvelocityvstime.

PAGE 35

23 0 10 20 30 40 50 60 70 80 90 10 0 0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 [sec][m/sec]Figure2—19:Slipvelocityvstime. 0 10 20 30 40 50 60 70 80 90 100 0.2 0.1 0 0.1 0.2 0.3 0.4 [sec]Figure2—20:Frictioncoe cientvstime.

PAGE 36

24 2.3ExperimentalResults TheexperimentaltestbedshowninFigure2—21consistsofacirculardisk madeofAluminium,mountedonaNSKdirect-driveswitchedreluctancemotor ( 240 0 NmModelYS5240-GN001).TheNSKmotoriscontrolledthroughpower electronicsoperatingintorquecontrolmode.Themotorresolverprovidesrotor positionmeasurementswitharesolutionof153600pulses/revolutionataresolver andfeedbackresolutionof10bits.APentium2.8GHzPCoperatingunder QNXhoststhecontrolalgorithm,whichwasimplementedviaQmotor3.0,a graphicaluser-interface,tofacilitaterea l-timegraphing,datalogging,andadjust controlgainswithoutrecompilingtheprogram(forfurtherinformationonQmotor 3.0,thereaderisreferredtoLo eretal.[39]).Dataacquisitionandcontrol implementationwereperformedatafrequencyof1.0kHzusingtheServoToGoI/O board.A 0 315 m 0 108 m 0 03175 mrectangularNylonblockwasmounted onapneumaticlinearthrustertoapplyanexternalfrictionloadtotherotating disk.Apneumaticregulatormaintainedaconstantpressureof 15 poundsper squareinchonthecirculardisk.Thistestbedwasusedtoimplementatracking controllerwithadaptivefrictionidenti cationdevelopedbytheauthorsinMakkar etal.[40]. Theaimofthisexperimentwastomatchtheexperimentallyidenti edfriction torqueusingtheadaptiveterminMakkaretal.[40]andMakkarandDixonetal. [41]withthefrictiontorquecalculatedfromthenewproposedfrictionmodelin (2—1).Theexperimentallyidenti edfrictionisdepictedFigure2—22. Thecoe cientsin(2—1)werevariedtomatchtheexperimentallyidenti ed frictiontorque.Thefrictiontorquein (2—1)wascalculatedasafunctionofthe rotorvelocitywiththecoe cientschosenas 1=34 8 2=650 3=1 4=26 5=200 6=19 5 (2—2)

PAGE 37

25 Figure2—21:Testbedfortheexperiment. Figure2—22showsaplotoftheexperimentallydeterminedfrictiontorque,aplot offrictiontorquedevelopedfrom(2—1) andaplotcomparingtheexperimentally determinedfrictiontorquewiththedevelopedfrictionmodeloverlaid. TheexperimentallyobtainedfrictiontorqueinFigure2—22hasviscousand staticfrictioncomponentsandexhibitstheStribecke ect.Thesefactsweretaken intoconsiderationwhilechoosingtheconstantsin(2—1).Valuesfor 1and 4were chosentoaccountforthestaticfriction,and 6waschosentocapturetheviscous frictioncomponent.AsseeninFigure2—22,theanalyticalmodelapproximatesthe experimentaldatawiththeexceptionofsomeovershoot.Theexperimentalorigin ofthedirectionalfrictionalanisotropyisdiscussedindetailinSchmitzetal.[48] andisattributedtosmallmisalignmentbetweentheloadingaxisandthemotor axis. 2.4ConcludingRemarks Anewcontinuouslydi erentiablefrictionmodelwithnonlinearparameterizabletermsisproposed.Thismodelcapturesanumberofessentialaspectsof frictionwithoutinvolvingdiscontinuousorpiecewisecontinuousfunctionsandcan

PAGE 38

26 0 5 10 15 20 25 30 35 40 200 100 0 100 200 Time [sec]Friction Torque [Nm]Experimentally identified friction 0 5 10 15 20 25 30 35 40 200 100 0 100 200 Time [sec]Friction Torque [Nm]Identified friction from model 0 5 10 15 20 25 30 35 40 200 100 0 100 200 Time [sec]Friction Torque [Nm]ComparisonFigure2—22:Theproposedfrictionmodelverycloselyapproximatestheexperimentallyidenti edfrictiontermobtainedfromtheadaptivecontrollerdeveloped inMakkaretal.[41].Topplotdepictstheexperimentallyobtainedfrictiontorque, middleplotdepitcsthefrictionplotobtainedfromtheproposedmodelandthe bottomplotdepictsacomparisonofthetwowithsolidlineindicatingexperimentallyobtainedfrictiontorqueanddashedlineindicatingfrictiontorqueobtained fromtheproposedfrictionmodel.

PAGE 39

27 bemodi edtoincludeadditionale ects.Thecontinuouslydi erentiableproperty oftheproposedmodelprovidesafoundationtodevelopcontinuouscontrollersthat canidentifyandcompensatefornonlinearfrictionale ects.Thedevelopmentof onesuchcontrollerisdiscussedinnextchapter.

PAGE 40

CHAPTER3 IDENTIFICATIONANDCOMPENSATIONFORFRICTIONBYHIGHGAIN FEEDBACK Motivatedbythedesiretoincludedynamicsfrictionmodelsinthecontrol design,anewtrackingcontrollerisdevelopedinthischapterthatcontainsthenew continuouslydi erentiablefrictionmodelwithuncertainnonlinearparameterizable termsthatwasdevelopedinChapter2.Frictionmodelsareoftenbasedonthe assumptionthatthefrictioncoe cientisconstantwithslidingspeedandhavea singularityattheonsetofslip.Suchmodelstypicallyincludeasignumfunctionof thevelocitytoassignthedirectionoffrictionforce(e.g.,seeLampaertetal.[34], andSweversetal.[52]),andmanyothermodelsareonlypiecewisecontinuous (e.g.,theLuGremodelin[8]).InMakkaretal.[42],MakkarandDixonetal. [43],andChapter2,anewfrictionmodelisproposedthatcapturesanumberof essentialaspectsoffrictionwithoutinvolvingdiscontinuousorpiecewisecontinuous functions.Thesimplecontinuouslydi erentiablemodelrepresentsafoundation thatcapturesthemajore ectsreportedanddiscussedinfrictionmodelingand experimentationandthemodelisgenericenoughthatothersubtletiessuchas frictionalanisotropywithslidingdirectioncanbeaddressedbymathematically distortingthismodelwithoutcompromisingthecontinuousdi erentiability. Basedonthefactthatthedevelopedmodeliscontinuouslydi erentiable,a newintegralfeedbackcompensationt ermoriginallyproposedbyXianetal. [57]isexploitedtoenableasemi-globaltrackingresultwhileidentifyingthe frictionon-line,assumingexactmodelknowledgeoftheremainingdynamics. ALyapunov-basedstabilityanalysisisprovidedtoconcludethetrackingand frictionidenti cationresults.Experimentalresultsshowtwoordersofmagnitude 28

PAGE 41

29 improvementintrackingcontroloveraproportionalderivative(PD)controller, andaoneorderofmagnitudeimprovementoverthemodel-basedcontroller. Experimentalresultsarealsousedtoillustratethattheexperimentallyidenti ed frictioncanbeapproximatedbythemodelinMakkaretal.[42]andMakkarand Dixonetal.[43]. Thischapterisorganizedasfollows.Thedynamicmodelandtheassociated propertiesareprovidedinSection3.1.Section3.2describesthedevelopmentof errorsystemfollowedbythestabilityanalysisinSection3.3.Section3.4describes theexperimentalsetupandresultsthatindicateimprovedperformanceobtained byimplementingtheproposedcontrollerfollowedbydiscussioninSection3.5and conclusioninSection3.6. 3.1DynamicModelandProperties Theclassofnonlineardynamicsystemsconsideredareassumedtobemodeled bythegeneralEuler-Lagrangeformulationin(1—1)wherethefrictionterm ( ) is assumedtohavetheformin(2—1)asinMakkaretal.[42]andMakkaretal.[43]. Thesubsequentdevelopmentisbasedontheassumptionthat ( ) and ( ) aremeasurableandthat ( ) ( ) ( ) areknown.Moreover,thefollowing propertiesandassumptionswillbeexploitedinthesubsequentdevelopment: Property3.1: Theinertiamatrix ( ) issymmetric,positivede nite,and satis esthefollowinginequality ( ) R: 1k k2 ( ) ( ) k k2(3—1) where 1 R isaknownpositiveconstant, ( ) R isaknownpositivefunction, and kk denotesthestandardEuclideannorm. Property3.2: If ( ) L,then ( ) ,and 2 ( ) 2existandarebounded. Moreover,if ( ) ( ) Lthen ( ) and ( ) arebounded.

PAGE 42

30 Property3.3: Basedonthestructureof ( ) givenin(21), ( ) ( ) and ( ... ) existandareboundedprovided ( ) ( ) ( ) ... ( ) L. 3.2ErrorSystemDevelopment Thecontrolobjectiveistoensurethatthesystemtracksadesiredtrajectory, denotedby ( ) ,thatisassumedtobedesignedsuchthat ( ) ( ) ( ) ... ( ) Rexistandarebounded.Apositi ontrackingerror,denotedby 1( ) R,isde nedasfollowstoquantifythecontrolobjective: 1, (32) Thefollowing lteredtrackingerrors,denotedby 2( ) ( ) R,arede nedto facilitatethesubsequentdesignandanalysis: 2, 1+ 11(33) 2+ 22(34) where 1, 2 R denotepositiveconstants.The lteredtrackingerror ( ) isnot measurablesincetheexpressionin(34)dependson ( ) Afterpremultiplying(34)by ( ) ,thefollowingexpressioncanbeobtained: ( ) = ( ) + ( ) + ( ) (35) + ( ) ( )+ ( ) 1 1+ ( ) 22where(11),(32),and(33)wereutilized.Basedontheexpressionin(35)the controltorqueinputisdesignedasfollows: ( )= ( ) + ( ) + ( )+ ( ) 1 1+ ( ) 22+ ( ) (36) where ( ) Rdenotesasubsequentlydesignedcontrolterm.Bysubstituting (3)into(3),thefollowingexpressioncanbeobtained: ( ) = ( ) ( ) (37)

PAGE 43

31 From(3—7),itisevidentthatif ( ) 0 ,then ( ) willidentifythefriction dynamics;therefore,theobject iveistodesignthecontrolterm ( ) toensurethat ( ) 0 .Tofacilitatethedesignof ( ) ,wedi erentiate(3—7)asfollows: ( ) = ( ) ( ) ( ) (3—8) Basedon(3—8)andthesubsequentstabilityanalysis, ( ) isdesignedasfollows: ( )=( +1) 2( ) ( +1) 2( 0) (3—9) + Z 0[( +1) 22( )+ ( 2( ))] where R and R arepositiveconstants.Thetimederivativeof(3—9)is givenas ( )=( +1) + ( 2) (3—10) Theexpressionin(3—9)for ( ) doesnotdependontheunmeasurable ltered trackingerrorterm ( ) .However,thetimederivativeof ( ) (whichisnot implemented)canbeexpressedasafunctionof ( ) .Aftersubstituting(3—10)into (3—8),thefollowingclosed-looperrorsystemcanbeobtained: ( ) = 1 2 ( ) ( +1) 2 ( 2)+ ( ) (3—11) where ( ) Rdenotesthefollowingunmeasurableauxiliaryterm: ( ) ( ) 1 2 ( ) + 2 (3—12) Tofacilitatethesubsequentanalysis,anotherunmeasurableauxiliaryterm ( ) Risde nedasfollows: ( ) ( ) ¨ = 12¨ 12¨ k tanh( 2 ) k2 13¨ (3—13) + 13¨ k tanh( 3 ) k2+ 45¨ 45¨ k tanh( 5 ) k2+ 6¨

PAGE 44

32 Thetimederivativeof(3—13)isgivenasfollows: ( )= 2 ( ) 2 ¨ 2 + ( ) ... = ... ( 12 13+ 45+ 6) (3—14) 12... k tanh( 2 ) k2+ 13... k tanh( 3 ) k2 45... k tanh( 5 ) k2 2 12 2|| ¨ ||2tanh( 2 )[1 k tanh( 2 ) k2]+2 12 3|| ¨ ||2tanh( 3 ) [1 k tanh( 3 ) k2] 2 42 5|| ¨ ||2tanh( 5 )[1 k tanh( 5 ) k2] Afteraddingandsubtracting(3—13),theclosed-looperrorsystemin(3—11)canbe expressedasfollows: ( ) = 1 2 ( ) ( +1) 2 ( 2)+ ( )+ ( ) (3—15) wheretheunmeasurableauxiliaryterm ( ) Risde nedas ( ) ( ) ( ) (3—16) Basedontheexpressionsin(3—13)and(3—14),thefollowinginequalitiescan bedeveloped: k ( ) k || ¨ ||| 12+ 45+ 6 13| (3—17) ( ) || ... ||| 12+ 45+ 6 13| + || ¨ ||2(2 12 2+2 12 3+2 42 5) (3—18) 2where 2 R areknownpositiveconstants. 3.3StabilityAnalysis Theorem3.1: Thecontrollergivenin(3—6)and(3—9)ensuresthatthe positiontrackingerrorisregulatedinthesensethat 1( ) 0 provided isselectedaccordingtothefollowingsu cientcondition:

PAGE 45

33 + 1 22(3—19) where and 2areintroducedin(3—17)and(3—18),respectively,and is selectedsu cientlylarge.Thecontrolsystemrepresentedby(3—6)and(3—9)also ensuresthatallsystemsignalsareboundedunderclosed-loopoperationandthat thefrictioninthesystemcanbeidenti edinthesensethat ( ) ( ) 0 Proof: Let D R3 +1beadomaincontaining ( )=0 ,where ( ) R3 +1is de nedas ( ) [ ( ) p ( )](3—20) where ( ) R3 isde nedas ( ) [ 1 2] (3—21) andtheauxiliaryfunction ( ) R isde nedas ( ) || 2( 0) || 2( 0)( 0) Z 0 ( ) (3—22) where R isnonnegativebydesign.In( 3—22),theauxiliaryfunction ( ) R is de nedas ( ) ( ( ) ( 2)) (3—23) Thederivative ( ) R canbeexpressedas ( )= ( )= ( ( ) ( 2)) (3—24) Providedthesu cientconditionintroducedin(3—19)issatis ed,thefollowing inequalitycanbeobtainedas Z 0 ( ) | 2( 0) | 2( 0)( 0) (3—25)

PAGE 46

34 Hence,(3—25)canbeusedtoconcludethat ( ) 0 .Let ( ): D [0 ) R beacontinuouslydi erentiablepositivede nitefunctionde nedas ( ) 11+ 1 2 22+ 1 2 ( ) + (3—26) thatcanbeboundedas 1( ) ( ) 2( ) (3—27) providedthesu cientconditionintroducedin(3—19)issatis ed.In(3—27),the continuouspositivede nitefunctions 1( ) 2( ) R arede nedas 1( )= 1k k22( )= 2( ) k k2(3—28) where 1, 2( ) R arede nedas 1, 1 2 min { 1 1} 2( ) max { 1 2 ( ) 1 } where 1, ( ) areintroducedin(3—1).Aftertakingthetimederivativeof(3—26), ( ) canbeexpressedas ( )= ( ) + 1 2 ( ) + 2 2+2 1 1+ Afterutilizing(3—3),(3—4),(3—15),and(3—24), ( ) cansimpli edasfollows: ( )= ( ) ( +1) k k2 2k 2k2 2 1k 1k2+2 21 (3—29) Because 2( ) 1( ) canbeupperboundedas 21 1 2 k 1k2+ 1 2 k 2k2 ( ) canbeupperboundedusingthesquaresofthecomponentsof ( ) as follows: ( ) ( ) ( +1) k k2 2k 2k2 2 1k 1k2+ k 1k2+ k 2k2

PAGE 47

35 Byusingthefact || ( ) || ( || || ) k k ,theexpressionin(3—29)canberewritten asfollows: ( ) 3k k2 ¡ k k2 ( || || ) k kk ) k ¢ (3—30) where 3, min { 2 1 1 2 1 1 } andtheboundingfunction ( || || ) R isa positivegloballyinvertiblenondecreasingfunction;hence, 12mustbechosen accordingtothefollowingconditions: 1 1 2 2 1 Aftercompletingthesquaresforthesecondandthirdtermin(3—30),thefollowing expressioncanbeobtained: ( ) 3k k2+ 2( ) k k2 4 (3—31) Thefollowingexpressioncanthenbeobtainedfrom(3—31): ( ) ( ) (3—32) where ( )= k k2,forsomepositiveconstant R ,isacontinuouspositive semi-de nitefunctionthatisde nedonthefollowingdomain: { R3 +1|k k 1(2 p 3) } Theinequalitiesin(3—27)and(3—32)canbeusedtoshowthat ( ) Lin D ; hence, 1( ) 2( ) ,and ( ) Lin D .Giventhat 1( ) 2( ) ,and ( ) Lin D ,standardlinearanalysismethods(e.g.,Lemma1.4of[15])canbeusedtoprove that 1( ) 2( ) Lin D from(3—3)and(3—4).Since 1( ) 2( ) ( ) Lin D ,theassumptionthat ( ) ( ) ¨ ( ) existandareboundedcanbeusedalong with(3—2)-(3—4)toconcludethat ( ) ( ) ¨ ( ) Lin D .Since ( ) ( ) Lin D ,Property2canbeusedtoconcludethat ( ) ( ) ( ) ,and ( ) Lin D .From(3—6)and(3—9),wecanshowthat ( ) ( ) Lin D .Giventhat

PAGE 48

36 ( ) Lin D ,(3—10)canbeusedtoshowthat ( ) Lin D .Property2and Property3canbeusedtoshowthat ( ) and ( ) Lin D ;hence,(3—8)can beusedtoshowthat ( ) Lin D .Giventhat ( ) Lin D ,then(3—2)-(3—4) canbeusedtoconcludethat ... ( ) Lin D .Since 1( ) 2( ) ( ) Lin D ,thede nitionsfor ( ) and ( ) canbeusedtoprovethat ( ) isuniformly continuousin D Let S D denoteasetde nedasfollows: S ( ) D| 2( ( )) 1 1(2 p 3) 2 (3—33) Theregionofattractionin(3—33)canbemadearbitrarilylargetoincludeany initialconditionsbyincreasingthecontrolgain (i.e.,asemi-globaltypeof stabilityresult)asinXianetal.[57].Theorem8.4of[30]cannowbeinvokedto statethat k ( ) k2 0 ( 0) S (3—34) Basedonthede nitionof ( ) ,(3—34)canbeusedtoshowthat ( ) 0 ( 0) S (3—35) Hence,from(3—3)and(3—4),standardlinearanalysismethods(e.g.,Lemma1.6of [15])canbeusedtoprovethat 1( ) 0 ( 0) S Theresultin(3—35)canalsobeusedtoconcludefrom(3—7)that ( ) ( ( )) 0 ( 0) S

PAGE 49

37 3.4ExperimentalResults Theexperimentaltestbedusedforimplementingthecontrollerisdescribedin Chapter2.Thedynamicsforthetestbedaregivenasfollows: ( )= £ +0 5 2¤ [¨ ] | {z } ( )¨ + ( ) (3—36) where (rotormomentofinertia) =0 255 kg-m2, (massofthecirculardisk) =3 175 kg, (radiusofthedisk) =0 25527 m,andthefrictiontorque ( ) R isde nedin(2—1).Thecontroltorqueinput ( ) givenin(3—6)issimpli ed(i.e., thecentripetal-Coriolismatrixandgravitytermswereomitted)asfollowsforthe simpletestbed: ( )= ( )¨ + ( ) 1 1+ ( ) 22+ ( ) (3—37) where ( ) istheadaptivefrictionidenti cationtermde nedin(3—9).Thedesired disktrajectory(seeFigure3—1)wasselectedsimilartotheoneusedinFeemsteret al.[20]asfollows(indegrees): ( )=11 25tan 1(3 0sin(0 5 ))(1 exp( 0 01 3)) (3—38) Thissoft-sinusoidaltrajectorywasproposedin[20]toemphasizealow-speed transitionfromforwardtoreversedirections.Forallexperiments,therotorvelocity signalisobtainedbyapplyingastandardbackwardsdi erencealgorithmto thepositionsignal.Allstateswereinitializedtozero.Inaddition,theintegral structureoftheadaptivetermin(3—37)wascomputedon-lineviaastandard trapezoidalalgorithm.

PAGE 50

38 0 5 10 15 20 25 30 35 40 15 10 5 0 5 10 15 Time[sec]Desired Trajectroy [Degrees]Figure3—1:Desireddisktrajectory. 3.4.1Experiment1 Inthe rstexperiment,noexternalloadfromthethrusterwasappliedto thecirculardisk.Inadditiontothecontrollergivenin(3—36)and(3—37),aPD controllerandamodel-basedcontrollerwerealsoimplementedforcomparison.The PDcontrollerwasimplementedas: ( )= 1+ 1(3—39) where R isthederivativegainand R istheproportionalgain.The model-basedcontrollerwasimplementedwithstandardfrictionfeedforwardterms as: ( )= ( )¨ + ( ) 1 1+ ( ) 22+ ( )+ + (3—40) where R istheCoulombfrictioncoe cient, R istheviscousfriction coe cient,and R isthestaticfrictioncoe cient. Thegainsforeachcontrollerthatyieldedthebeststeady-stateperformance weredeterminedasfollowsforthePDcontroller,model-basedcontroller,and proposedcontroller,respectively:

PAGE 51

39 • PDcontroller =2600 =2600 (3—41) • Model-basedcontroller 1=315 2=315 =0 0828 (3—42) =0 0736 =0 104 Proposedcontroller =10 =5 1=100 2=600 (3—43) Thecoe cient and in(3—40)werecalculatedfromthefriction identi cationplotinFig.3—10.Thecoe cient iscalculatedfromthepeak frictiontorqueattheonsetofacycle, iscalculatedfromthe atportionofthe curveaftertheinitialpeak,and iscalculatedfromthehighpeakafterthe at portionofthecurve. ThepositiontrackingerrorfromeachcontrollerisplottedinFigures3—2-3—4 respectively.Acomparisonoftheposit iontrackingerrorfromeachcontroller isseeninFigure3—5.Figure3—6depictsanenlargedviewofthecomparison ofpositiontrackingerrorsfromthemodel-basedcontrollerandtheproposed controller.ThetorqueinputbyeachcontrollerisdepictedinFigures3—7-3—9 respectively.Thefrictionidenti cationtermin(3—9)fromtheproposedcontroller obtainedfromtheexperimentisgiveninFigure3—10. Thesignumfunctionforthecontrolschemein(3—9)wasde nedas ( 2( ))= 1 2 0 1 2 0 0 2=0

PAGE 52

40 0 5 10 15 20 25 30 35 40 2 1.5 1 0.5 0 0.5 1 1.5 2 Time [sec]Position Tracking Error [Degrees]Figure3—2:Positiontracking errorfromthePDcontroller. 0 5 10 15 20 25 30 35 40 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 Time [sec]Position Tracking Error [Degrees]Figure3—3:Positiontrackingerrorfromt hemodel-basedcontrollerwithfriction feedforwardtermsasdescribedin(3—40).

PAGE 53

41 0 5 10 15 20 25 30 35 40 0.08 0.06 0.04 0.02 0 0.02 0.04 0.06 Time [sec]Position Tracking Error [Degrees]Figure3—4:Positiontrackingerrorfromtheproposedcontroller. 0 5 10 15 20 25 30 35 40 2 1.5 1 0.5 0 0.5 1 1.5 2 Time [sec]Position Tracking Error [Degrees] PDController Modelbased controller Proposed controller Figure3—5:Comparisonofpositiontrackingerrorsfromthethreecontrolschemes.

PAGE 54

42 0 5 10 15 20 25 30 35 40 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 Time [sec]Position Tracking Error [Degrees] Modelbased controller Proposed controller Figure3—6:Comparisonofpositiontracki ngerrorsfromthemodel-basedcontroller andtheproposedcontroller. 0 5 10 15 20 25 30 35 40 100 80 60 40 20 0 20 40 60 80 Time (sec)Torque(N)Figure3—7:TorqueinputbythePDcontroller.

PAGE 55

43 0 5 10 15 20 25 30 35 40 150 100 50 0 50 100 150 Time (sec)Torque(N)Figure3—8:Torqueinputbythemodel-basedcontrollerwithfrictionfeedforward termsasdescribedin(3—40). 0 5 10 15 20 25 30 35 40 150 100 50 0 50 100 150 Time [sec]Torque [Nm]Figure3—9:Torqueinputbytheproposedcontroller.

PAGE 56

44 0 5 10 15 20 25 30 35 40 80 60 40 20 0 20 40 60 80 Time [sec]Identified Friction [Nm]Figure3—10:Identi edfrictionfromtheadaptivetermintheproposedcontroller. 3.4.2Experiment2 Inthesecondexperiment,anexternalfrictionloadwasinducedonthesystem. Anexternalmomentloadof 12 774 Nmwasappliedtothecirculardiskusingthe linearthruster(seeFigure2—21).Thedesi reddisktrajectoryof(3—38)wasagain utilized.Thecontrolschemesof(3—37),(3—39),and(3—40)wereimplementedwith thefollowinggainvalues: • PDcontroller =2600 =2600 • Model-basedcontroller 1=350 2=350 =0 0828 =0 0736 =0 104 • Proposedcontroller =8 9 =5 005 1=98 2=780 (3—44)

PAGE 57

45 ThecorrespondingpositiontrackingerrorfromeachcontrollerisshowninFigures 3—11-3—13,respectively.Acomparisonofthepositiontrackingerrorfromeach controllerisseeninFigure3—14.Figure3—15depictsanenlargedviewofthe comparisonofpositiontrackingerrorsfromthemodel-basedcontrollerandthe proposedcontroller.Thecontroltorqueinputbyeachcontrollerisshownin Figures3—16-3—18,respectively.Thefrictionidenti cationtermin(3—9)fromthe proposedcontrollerobtainedfromtheexperimentisgiveninFigure3—19. 0 5 10 15 20 25 30 35 40 5 4 3 2 1 0 1 2 3 4 5 Time [sec]Position Tracking Error [Degrees]Figure3—11:Positiontrackin gerrorfromthePDcontroller. 3.4.3Experiment3 Inthethirdexperiment,lubricationwasusedtoexaminechangesinthesystemperformance.Thecirculardiskwaslubricatedwithmotoroilandthecontrol schemein(3—37)wasimplementedwithsamecontrolgainsasinExperiment2 tominimizethepositiontrackingerror.Th eresultingpositiontrackingerroris showninFigure3—20,thecontroltorqueinputisshowninFigure3—21,andthe frictionidenti cationtermin(3—9)fromtheproposedcontrollerobtainedfromthe experimentisgiveninFigure3—22.

PAGE 58

46 0 5 10 15 20 25 30 35 40 0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 Time [sec]Position Tracking Error [Degrees]Figure3—12:Positiontrackingerrorfromthemodel-basedcontrollerwithfriction feedforwardtermsasdescribedin(3—40). 0 5 10 15 20 25 30 35 40 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 Time [sec]Position Tracking Error [Degrees]Figure3—13:Positiontrakingerrorfromtheproposedcontroller.

PAGE 59

47 0 5 10 15 20 25 30 35 40 5 4 3 2 1 0 1 2 3 4 5 Time [sec]Position Tracking Error [Degrees] PD controller Modelbased controller Proposed controller Figure3—14:Comparisonofpositiont rackingerrorsfromthethreecontrol schemes. 0 5 10 15 20 25 30 35 40 0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 Time [sec]Position Tracking Error [Degrees] Modelbased controller Proposed controller Figure3—15:Comparisonofpositiontra ckingerrorsfromthemodel-basedcontrollerandtheproposedcontroller.

PAGE 60

48 0 5 10 15 20 25 30 35 40 200 150 100 50 0 50 100 150 200 Time (sec)Torque (N)Figure3—16:TorqueinputbythePDcontroller. 0 5 10 15 20 25 30 35 40 250 200 150 100 50 0 50 100 150 Time [sec]Torque [Nm]Figure3—17:Torqueinputbythemodel-basedcontrollerwithfrictionfeedforward termsasdescribedin(3—40).

PAGE 61

49 0 5 10 15 20 25 30 35 40 250 200 150 100 50 0 50 100 150 200 Time [sec]Torque [Nm]Figure3—18:Torqueinputbytheproposedcontroller. 0 5 10 15 20 25 30 35 40 200 150 100 50 0 50 100 150 Time [sec]Identified Friction [Nm]Figure3—19:Identi edfrictionfromtheadaptivetermintheproposedcontroller.

PAGE 62

50 0 5 10 15 20 25 30 35 40 0.2 0.15 0.1 0.05 0 0.05 0.1 Time [sec]Position Tracking Error [Degrees]Figure3—20:Positiontrackingerrorwiththeproposedcontrollerwhenthecircular diskwaslubricated. 0 5 10 15 20 25 30 35 40 200 150 100 50 0 50 100 150 200 Time [sec]Torque [Nm]Figure3—21:Torqueinputbytheproposedcontrollerwhenthecirculardiskwas lubricated.

PAGE 63

51 0 5 10 15 20 25 30 35 40 150 100 50 0 50 100 150 Time [sec]Identified Friction [Nm]Figure3—22:Identi edfrictionfromtheadaptivetermintheproposedcontroller whenthecirculardiskwaslubricated. 3.4.4Experiment4 Inthefourthexperiment,thenetexternalfrictioninducedonthesystemas aresultofexternalloadappliedtothecirculardiskbythelinearthrusterwas identi ed.Thefrictioninthetestbedunderno-loadconditionswasidenti edasin Experiment1usingthecontrolgainsofExperiment2.Thisidenti edfrictionterm wassubtractedfromtheidenti edfrictiontermsobtainedfromExperiment2and Experiment3respectively.ThefrictionbetweenthecirculardiskandNylonblock whenthediskwasnotlubricatedcanbeseeninFigure3—23,andthefrictionwhen thediskwaslubricatedcanbeseeninFigure3—24. 3.4.5Experiment5 Inthe fthexperiment,theexperimentallyidenti edfrictiontorqueusingthe adaptivetermin(3—9)wascomparedwith thefrictiontorquemodelin(2—1).The frictionidenti edinFigure3—22wascomparedwiththemodelparametersin(2—1) thatwereadjustedtomatchtheexperimentaldata.Therotorvelocitysignalwas obtainedbyapplyingastandardbackwardsdi erencealgorithmtotheposition signal,andthefrictiontorquein(2—1)wascalculatedasafunctionofthisrotor

PAGE 64

52 0 5 10 15 20 25 30 35 40 120 100 80 60 40 20 0 20 40 60 80 Time [sec]Identified Friction [Nm]Figure3—23:Netexternalfrictioninducedwithnolubrication.Thenetfriction wascalculatedbysubractingtheidenti edfrictionterminExperiment1fromthe identi edfrictionterminExperiment2. 0 5 10 15 20 25 30 35 40 20 15 10 5 0 5 10 15 20 Time(sec)Identified Friction (N)Figure3—24:Netexternalfrictioninducedwithlubrication.Thenetfrictionwas calculatedbysubractingtheidenti edfrictionterminExperiment1fromthe identi edfrictionterminExperiment3.

PAGE 65

53 velocitywiththeconstantschosenas 1=34 8 2=650 3=1 4=26 5=200 6=19 5 ThematchingofthefrictiontorquewiththeexperimentaldataisplottedinFigure 3—25. 0 5 10 15 20 25 30 35 40 150 100 50 0 50 100 150 Time [sec]Friction Torque [Nm]ComparisonFigure3—25:Thefrictiontorquecalculatedfromthemodelin(2—1)approximates theexperimentallyidenti edfrictiontorquein(3—9). 3.5Discussion InExperiment1,itwasobservedthatthepositiontrackingerrorobtained fromthePDcontrollerin(3—39)deviatedaround 0 7337 degrees(asseeninFigure 3—2)comparedto 0 0506 degrees(asinFigure3—3)inthemodel-basedcontroller in(3—40).Thiserrorwasfurtherreducedtoabout 0 0116 degreeswhenthe proposedcontrollerin(3—37)wasimplemented(asseeninFigure3—4).Adetailed comparisonofthepositiontrackingerrors(indegrees)andcontroltorque(inNm) canbeseeninTableI. Similardi erenceintheorderofthemagnitudeofpositiontrackingerrors wasalsoobservedinExperiment2,whenanexternalfrictionwasappliedtothe

PAGE 66

54 Table3—1:Comparisonoftrackingresultswhennoexternalloadwasappliedtothe circulardisk. PD Model-based Proposed controller controller controller Standarddeviationinerror 0.7337 0.0506 0.0116 Rootmeansquarevalue 0.7442 0.0512 0.0116 Standarddeviationintorque 33.2926 36.7456 37.5658 Rootmeansquarevalue 33.7704 37.2826 32.1060 Table3—2:Comparisonoftrackingresultswhenanexternalloadwasappliedtothe circulardisk. PD Model-based Proposed controller controller controller Standarddeviationinerror 1.8225 0.1203 0.0186 Rootmeansquarevalue 1.8373 0.1212 0.0186 Standarddeviationintorque 82.7043 103.0262 72.9409 Rootmeansquarevalue 83.3753 103.8698 73.9593 circulardisk.Thepositiontrackinge rrorwhenthePDcontrollerin(3—39)was implementeddeviatedabout 1 8225 degrees(seeFigure3—11),whereasinthe model-basedcontrollerin(3—40)deviatedabout 0 1203 degreesasseeninFigure 3—12.Underthesameloadconditions,whentheproposedcontrollerin(3—37) wasimplemented,thepositiontrackingerrordeviatedaround 0 0186 degrees(see Figure3—13).Adetailedcomparisonofthemagnitudesofthepositiontracking errors(indegrees)andcontroltorque(inNm)canbeseeninTableII. Experiments1-3illustrateanapproximatefactorof60improvementoveraPD controller,andafactorofapproximately3overatypicalexactmodelknowledge controllerwithstaticandviscousfrictionfeedforwardterms(seeFigure3—5,Figure 3—14,Figure3—6,and3—15).Thesigni cantimprovementinthetrackingerror isalsoobservedfromTableIandTableII,wherethermsvalueoferrorfromthe proposedcontrollerisabouttwoordersof magnitudebettertha nthePDcontroller andaboutoneorderofmagnitudebetterthanthemodel-basedcontroller.This

PAGE 67

55 improvementinperformancefromtheproposedcontrollerwasobtainedwhileusing similarorlowerinputtorqueascanbeseeninTableIandTableII. Thisimprovementisbasedonthefactthattheproposedcontrollercontainsa feedforwardtermthatidenti esthefrictionasageneraltime-varyingdisturbance. Todevelopthefrictionidenti cationterm,thefrictionmodelisrequiredtobe continuouslydi erentiable.Experiments1-5illustratetheidenti edfrictiontorque, andExperiment5speci callyillustratesthatacontinuouslydi erentiablefriction modelproposedinMakkaretal.[42]andMakkarandDixonetal.[43]closely matchestheidenti edfriction.Mismatchesbetweentheidenti edfrictionandthe frictionmodelcanbeattributedtoanumberofe ectssuchaswear,nonuniformity ofthematerial,etc.However,thestructureofthefrictionidenti cationterm enablestheseanisotropice ectstobecaptured. Itisinterestingtocomparetheidenti edfrictioncomponentfromeachof theexperiments.Thedi erencebetweenExperiment1and2isthatanexternal frictionloadisappliedinExperiment2.Bycomparingtheidenti edfriction componentinFigure3—10andFigure3—19 ,asexpected,afactorof2-3increase inthestaticfrictioncomponentisobservedinFigure3—19withsomeincreasein theviscousfrictione ects.Thedi erencebetweenExperiment2and3isthat thecontactsurfaceislubricatedinExperiment3.Bycomparingtheidenti ed frictioncomponentinFigure3—19andFigure3—22,thestaticfrictioncomponent isunchangedandaslightreductionintheviscousfrictionisobserved.Thisis becausetheviscousfrictionofthemotorassemblyisthedominantviscouse ect,as isillustratedbytheresultsfromExperiment4. InExperiment5,theparametersofthenonlinearparameterizablefriction modelin(2—1)werevariedtomatchtheexperimentallyobtainedfrictiontorque inFigure3—22.Figure3—25showsthattheproposedfrictionmodelin(2—1)very closelyapproximatestheexperimentalfrictiontorque.However,sincefrictionis

PAGE 68

56 Figure3—26:WearingoftheNylonblockwhereitrubbedagainstthecirculardisk. anisotropicinnature,themagnitudeoffrictioninexperimentaldataisnotsymmetricalaboutthehorizontalaxiswhereasthefrictionmodelin(2—1)approximates itassymmetric.Hence,futureworkcanfocusonmathematicallydistortingthe modelproposedin(2—1)byadditionofmo retermstomakeitasymmetricormakingtheunknowncoe cientstime-varyinginordertocapturemorefrictione ects suchaswear.WearisevidentontheNylonblockasindicatedbygroves(seeFigure 3—26andFigure3—27)inthesurface. 3.6ConcludingRemarks Asemi-globalasymptotictrackingcont rollerisdevelopedbasedontheproposedcontinuouslydi erentiablefrictionmodelthatcontainsuncertainnonlinear parameterizableterms.Toachievethetrackingresult,anintegralfeedbackcompensationtermisusedtoidentifythesystemfrictione ects.ALyapunov-based stabilityanalysisisprovidedtoconcludethetrackingandfrictionidenti cationresults.Experimentalresultsshowtwoorde rsofmagnitudeimprovementintracking controloveraproportionalderivative(PD)controller,andaoneorderofmagnitudeimprovementoverthemodel-basedcontroller.Thiscontrollerassumesexact

PAGE 69

57 Figure3—27:Wearonthecirculardisk. modelknowledgeofthesystemdynamicsexceptfriction.Thecontroldevelopmentinthenextchapterismotivatedtoeliminatethisassumptionanddesigna controllerthatcansegregatethestructuredandunstructureduncertaintiesinthe dynamicswhileachievingasymptotictracking.

PAGE 70

CHAPTER4 TRACKINGCONTROLINTHEPRESENCEOFFRICTIONBYHIGHGAIN FEEDBACKANDAMODEL-BASEDFEEDFORWARDCOMPONENT-AN EXTENSION Theresultinthischapterismotivate dbythedesiretoincludesomeknowledgeofthedynamicsinthecontroldesignasameanstoimprovetheperformance andreducethecontrole ortwhileeliminatingtheassumptionthatthedynamics ofthesystemiscompletelyknown.Speci cally,forsystemsthatincludesome dynamicsthatcanbesegregatedintostructured(i.e.,linearparameterizable)and unstructureduncertainty,thisresultillustrateshowanewcontroller,errorsystem, andstabilityanalysiscanbecraftedtoincludeamodel-basedadaptivefeedforward terminconjunctionwiththehighgainintegralfeedbacktechniquetoyieldan asymptotictrackingresult. Thischapterpresentsthe rstresultthatillustrateshowtheamalgamation ofthesecompensationmethodscanbeusedtoyieldanasymptoticresult.Heuristically,theadditionofthemodel-basedadaptivefeedforwardtermshouldreduce theoverallcontrole ortbecausesomeofthedisturbancehasbeenisolatedand compensatedforbyanon-highgainfeedforwardelement.Moreover,theaddition oftheadaptivefeedforwardterminjectssomeknowledgeofthedynamicsinthe controlstructure,whichcouldleadtoimprovedperformance. Preliminaryexperimentalresultsarepresentedtoreinforcetheseheuristic notions.Speci cally,thepresentedcontrollerwasimplementedonasimpleonelinkrobottestbedanddemonstratedreducedtrackingerror.Forthistestbed, theonlydynamicsthatwereincludedinthefeedforwardtermincludedthe inertiaofthelinkagesystem.Thischapterisorganizedasfollows.Thedynamic 58

PAGE 71

59 modelandtheassociatedpropertiesareprovidedinSection4.1.Section4.2 describesthedevelopmentoferrorsystemfollowedbythestabilityanalysisin Section4.3.Section4.4describestheexp erimentalsetupandresultsthatindicate improvedperformanceobtainedbyimplementingtheproposedcontrollerfollowed bydiscussioninSection4.5andconclusioninSection4.6. 4.1DynamicModelandProperties TheclassofnonlineardynamicsystemsconsideredaredescribedinChapter1 andChapter2bythegeneralEuler-Lagrangeformulationin(1—1)wheretheterm ( ) Rdenotesageneralnonlineardisturbance(e.g.,frictionasinMakkaretal. [42]andMakkarandDixonetal.[43],unmodelede ects). Thesubsequentdevelopmentisbasedontheassumptionthat ( ) and ( ) aremeasurableandthat ( ) ( ) and ( ) areunknown.Moreover, thefollowingpropertiesandassumpti onswillbeexploitedinthesubsequent development: Property4.1: Theinertiamatrix ( ) asin(3—1)issymmetric,positivede nite, andsatis esthefollowinginequality ( ) R: 1k k2 ( ) ( ) k k2(4—1) where 1 R isaknownpositiveconstant, ( ) R isaknownpositivefunction, and kk denotesthestandardEuclideannorm. Property4.2: If ( ) ( ) L,then ( ) ( ) and ( ) arebounded. Moreover,if ( ) ( ) L,then ( ) 2 ( ) 2 ( ) 2( ) 2 ( ) 2( ) 2, ( ) 2 ( ) 2, ( ) ,and 2 ( ) 2existandarebounded. Property4.3: Thereexistpositivescalarconstant 1suchthat k ( ) k 1k k R (4—2)

PAGE 72

60 Property4.4: Thedynamicequationin(1—1)canbelinearparameterizedas ( ¨ ) ( )¨ + ( ) + ( ) (4—3) where Rcontainstheconstantunknownsystemparameters,andthedesired regressionmatrix ( ¨ ) R containsknownfunctionsofthedesired linkposition,velocity,andacceleration, ( ) ( ) ¨ ( ) R,respectively.The desiredtrajectoryisassumedtobedesignedsuchthat ( ) ( ) ¨ ( ) ... ( ) .... ( ) Rexistandarebounded. 4.2ErrorSystemDevelopment Thecontrolobjectiveistoensurethatthesystemtracksadesiredtimevaryingtrajectorydespitestructuredandunstructureduncertaintiesinthedynamic model.Toquantifythisobjective,apositiontrackingerrorasin(3—2),denotedby 1( ) R,isde nedas 1= (4—4) Tofacilitatethesubsequentanalysis, lteredtrackingerrorsasin(3—3and3—4), denotedby 2( ) ( ) R,arealsode nedas 2= 1+ 11(4—5) = 2+ 22(4—6) where 1, 2 R denotepositiveconstants.The lteredtrackingerror ( ) isnot measurablesincetheexpressionin(4—6)dependson ¨ ( ) Theopen-looptrackingerrorsystemcanbedevelopedbypremultiplying(4—6) by ( ) andutilizingtheexpressionsin(1—1),(4—3),(4—4),and(4—5)toobtain thefollowingexpression: ( ) = + ( ) (4—7) where ( ) Risde nedin(4—3),andtheauxiliarysignal ( 12 ) Ris

PAGE 73

61 de nedas = ( )(¨ + 1 1+ 22)+ ( ) + ( )+ ( ) (4—8) Basedontheexpressionin(4—7),thecontroltorqueinputisdesignedasfollows: = ˆ + (4—9) In(4—9), ( ) Rdenotesahigh-gainfeedbackcontroltermde nedas ( )=( +1) 2( ) ( +1) 2( 0) (4—10) + Z 0[( +1) 22( )+ ( 2( ))] where R arepositiveconstantcontrolgainsasin(3—9),and ˆ ( ) Rdenotesaparameterestimatevectorgene ratedon-lineaccordingtothefollowing updatelaw:ˆ = (4—11) with R beingaknown,constant,diagonal,positive-de nite,adaptationgain matrix.Since ( ) isonlyafunctionoftheknowntimevaryingtrajectory,(4—11) canbeintegratedbypartsasfollows: ˆ ( )= ˆ ( 0)+ 2( ) 0 Z 0n ¨ 2( ) 2 2( ) o (4—12) sothattheparameterestimatevector ˆ ( ) implementedin(4—9)doesnotdepend ontheunmeasurablesignal ( ) Remark4.1: Thecontroldesignin(4—9)issimilartotheresultsinXian etal.[57].However,previousdesignsbasedon[57]couldonlycompensatefor uncertaintyinthesystemthroughthehighgainfeedbackterm ( ) .Throughthe newerrorsystemdevelopmentandstabilityanalysismethodspresentedinthe currentresult,anadaptivefeedforwardtermcanalsobeusedtocompensatefor systemuncertainty.This exibilitypresentsasigni cantadvantagebecauseit

PAGE 74

62 allowsmoresystemdynamicstobeincorporatedinthecontroldesign.Speci cally, ifsomeofthesystemuncertaintycanbesegregatedintoalinearparameterizable form,thenthemodel-basedadaptivefeedforwardtermcanbeinjectedtocompensate fortheuncertaintyinsteadofjustrelyingonthenon-modelbasedhighgainfeedback term.Heuristically,thiscontributionshouldimprovethetrackingperformanceand reducethecontrole ort.Preliminaryexperimentalresultsonasimpleone-link robotmanipulatorprovidesomevalidationofthisheuristicidea. Theclosed-looptrackingerrorsystemcanbedevelopedbysubstituting(4—9) into(4—7)as ( ) = + ( ) (4—13) where ( ) Rrepresentstheparameterestimationerrorvectorde nedas = ˆ (4—14) Tofacilitatethesubsequentstabilityanalysis(andtoillustratesomeinsightinto thestructureofthedesignfor ( ) ),thetimederivativeof(4—13)isdeterminedas ( ) = ( ) + ˆ + ( ) (4—15) Thetimederivativeof(4—10)isgivenas ( )=( +1) + ( 2) (4—16) Aftersubstituting(4—11)and(4—16)into(4—15),thefollowingclosed-looperror systemcanbeobtained: ( ) = 1 2 ( ) + + ( ) 2(4—17) where ( 12 ) Rdenotesthefollowingunmeasurableauxiliaryterm: = + + 2 1 2 ( ) (4—18)

PAGE 75

63 Aftersubstitutingforthetimederivativeof ( 12 ) ,auxiliarytermin(4—18) canbeexpressedas = + ( ) £ 12 2 11+ 22¤ + ( ) 1 22(4—19) 1( 2 11)+ 2( 22)+ ( ) ¨ + ( ) ... + ( ) 2+ ( ) ¨ + ( )¨ + ( ) + ( ) ¨ + 2 1 2 ( ) Tofurtherfacilitatethesubsequentanalysis,anotherunmeasurableauxiliaryterm ( ¨ ) Risde nedas = ( ) ¨ + ( ) ¨ + ( ) ... + ( ) 2 (4—20) + ( ) ¨ + ( )¨ + ( ) Afteraddingandsubtracting(4—20),theclosed-looperrorsystemin(4—17)canbe expressedas ( ) = 1 2 ( ) + ( ) 2+ ( )+ ( ) (4—21) wheretheunmeasurableauxiliaryterm ( 12 ) Risde nedas ( )= ( ) ( ) (4—22) InasimilarmannerasinXianetal.[57],theMeanValueTheoremcanbeusedto developthefollowingupperbound ( ) ( k k ) k k (4—23) where ( ) R3 isde nedas ( ) [ 1 2] (4—24) Thefollowinginequalitiescanbedevelopedbasedontheexpressionin(4—20)and itstimederivative: k ( ) k ( ) 2(4—25)

PAGE 76

64 where 2 R areknownpositiveconstants. 4.3StabilityAnalysis Theorem4.1: Thecontrollergivenin(4—9),(4—10),and(4—12)ensuresthat allsystemsignalsareboundedunderclosed-loopoperationandthattheposition trackingerrorisregulatedinthesensethat 1( ) 0 providedthecontrolgain introducedin(4—10)isselectedsu cientlylarge,and isselectedaccordingtothefollowingsu cientcondition: + 1 22(4—26) where and 2areintroducedin(4—25). Proof: Let D R3 + +1beadomaincontaining ( )=0 ,where ( ) R3 + +1isde nedas ( )=[ ( ) ( ) p ( )](4—27) andtheauxiliaryfunction ( ) R isde nedas ( )= || 2( 0) || 2( 0)( 0) Z 0 ( ) (4—28) In(4—28),theauxiliaryfunction ( ) R isde nedas ( )= ( ( ) ( 2)) (4—29) Thederivative ( ) R canbeexpressedas ( )= ( )= ( ( ) ( 2)) (4—30) Providedthesu cientconditionintroducedin(4—26)issatis ed,thefollowing inequalitycanbeobtained(seeXianetal.[57]) Z 0 ( ) || 2( 0) || 2( 0)( 0) (4—31)

PAGE 77

65 Hence,(4—31)canbeusedtoconcludethat ( ) 0 Let ( ): D [0 ) R beacontinuouslydi erentiablepositivede nite functionde nedas ( )= 11+ 1 2 22+ 1 2 ( ) + + 1 2 1 (4—32) whichsatis esthefollowinginequalities: 1( ) ( ) 2( ) (4—33) providedthesu cientconditionintroducedin(4—26)issatis ed.In(4—33),the continuouspositivede nitefunctions 1( ) 2( ) R arede nedas 1( )= 1k k22( )= 2( ) k k2(4—34) where 1, 2( ) R arede nedas 1, 1 2 min { 1 1min 1 } 2( ) max { 1 2 ( ) 1 2 max 1 1 } where 1, ( ) areintroducedin(4—1)and min{} max{} denotetheminimum andmaximumeigenvalueoftheargumentrespectively.Aftertakingthetime derivativeof(4—32), ( ) canbeexpressedas ( )= ( ) + 1 2 ( ) + 2 2+2 1 1+ 1 ˆ Afterutilizing(4—5),(4—6),(4—1 1),(4—16),(4—21),and(4—30), ( ) canbe simpli edasfollows: ( )= ( ) ( +1) k k2 2k 2k2 2 1k 1k2+2 21 (4—35) Because 2( ) 1( ) canbeupperboundedas 21 1 2 k 1k2+ 1 2 k 2k2

PAGE 78

66 ( ) canbeupperboundedusingthesquaresofthecomponentsof ( ) as follows: ( ) ( ) ( +1) k k2 2k 2k2 2 1k 1k2+ k 1k2+ k 2k2 Byusing(4—23),theexpressionin(4—35)canberewrittenasfollows: ( ) 3k k2 ¡ k k2 ( || || ) k kk k ¢ (4—36) where 3, min { 2 1 1 2 1 1 } ,andtheboundingfunction ( || || ) R isa positivegloballyinvertiblenondecreasingfunction;hence, 12mustbechosen accordingtothefollowingconditions: 1 1 2 2 1 Aftercompletingthesquaresforthesecondandthirdtermin(4—36),thefollowing expressioncanbeobtained: ( ) 3k k2+ 2( ) k k2 4 (4—37) Thefollowingexpressioncanthenbeobtainedfrom(4—37) ( ) ( ) (4—38) where ( )= k k2,forsomepositiveconstant R ,isacontinuouspositive semi-de nitefunctionthatisde nedonthefollowingdomain: { R3 + +1|k k 1(2 p 3) } Theinequalitiesin(4—33)and(4—38)canbeusedtoshowthat ( ) Lin D ;hence, 1( ) 2( ) ( ) ,and ( ) Lin D .Giventhat 1( ) 2( ) ,and ( ) Lin D ,standardlinearanalysismethodscanbeusedtoprovethat 1( ) 2( ) Lin D from(4—5)and(4—6).Since Rcontainstheconstantunknown

PAGE 79

67 systemparametersand ( ) Lin D ,(4—14)canbeusedtoprovethat ˆ ( ) Lin D .Since 1( ) 2( ) ( ) Lin D ,theassumptionthat ( ) ( ) ¨ ( ) existandareboundedcanbeusedalongwith(4—4)-(4—6)toconcludethat ( ) ( ) ¨ ( ) Lin D .Theassumptionthat ( ) ( ) ¨ ( ) ... ( ) .... ( ) exist andareboundedalongwithProperty4canbeusedtoshowthat ( ¨ ) ( ¨ ) ,and ¨ ( ¨ ) Lin D .Since ( ) ( ) Lin D ,Property 2canbeusedtoconcludethat ( ) ( ) ( ) ,and ( ) Lin D .From (4—9)and(4—10),wecanshowthat ( ) ( ) Lin D .Giventhat ( ) Lin D ,(4—16)canbeusedtoshowthat ( ) Lin D .Since ( ) ¨ ( ) Lin D ,Property2canbeusedtoshowthat ( ) ( ) ( ) and ( ) Lin D ;hence,(4—15)canbeusedtoshowthat ( ) Lin D .Since 1( ) 2( ) ( ) Lin D ,thede nitionsfor ( ) and ( ) canbeusedtoprovethat ( ) is uniformlycontinuousin D Let S D denoteasetde nedasfollows: S ( ) D| 2( ( )) 1 1(2 p 3) 2 (4—39) Theregionofattractionin(4—39)canbemadearbitrarilylargetoincludeany initialconditionsbyincreasingthecontrolgain (i.e.,asemi-globaltypeof stabilityresult)asinXianetal.[57].Theorem8.4of[30]cannowbeinvokedto statethat k ( ) k2 0 ( 0) S (4—40) Basedonthede nitionof ( ) ,(4—40)canbeusedtoshowthat ( ) 0 ( 0) S (4—41) Hence,from(4—5)and(4—6),standardlinearanalysismethodscanbeusedto provethat 1( ) 0 ( 0) S

PAGE 80

68 Figure4—1:Theexperimentaltestbedconsistsofa1-linkrobotmountedonaNSK direct-driveswitchedreluctancemotor. 4.4ExperimentalResults ThetestbeddepictedinFigure4—1wasusedtoimplementthedeveloped controller.Thetestbedconsistsofaone-linkroboticarmofunknowninertia mountedonaNSKdirect-driveswitchedreluctancemotor(240.0NmModel YS5240-GN001).TheNSKmoto riscontrolledthroughpowerelectronicsoperating intorquecontrolmode.Themotorresolverprovidesrotorpositionmeasurements witharesolutionof614400pulses/revolutionataresolverandfeedbackresolutionof12bits.APentium2.8GHzPCoperatingunderQNXhoststhecontrol algorithm,whichwasimplementedviaQmotor3.0,agraphicaluser-interface,to facilitatereal-timegraphing,datalogging,andadjustmentofcontrolgainswithout recompilingtheprogram(forfurtherin formationonQmotor3.0,thereaderis referredtoLo eretal.[39]) Dataacquisitionandcontrolimplementationwere performedatafrequencyof1.0kHzusingtheServoToGoI/Oboard. Thedynamicsforthetestbedaregivenasfollows: ( )= ¨ + ( ) (4—42)

PAGE 81

69 where R denotestheinertiaofthelinkassemblyand ( ) R representsthe frictiontorque.Thecontroltorqueinput ( ) givenin(4—9)issimpli ed(i.e.,the centripetal-Coriolismatrixandgravitytermswereomitted)forthesimpletestbed as: ( )= ˆ ¨ + ( ) (4—43) where ˆ ( ) denotestheadaptiveestimatefortheinertiaofthelinkassembly,and ( ) istheadaptivecontroltermde nedin(4—10).Theadaptiveestimate ˆ ( ) is updatedaccordingtothefollowingupdatelaw: ˆ ( )= ˆ ( 0)+ Z 0 ... (4—44) where R isaconstantpositiveadaptivegainand ˆ ( 0)=0 179 kg-m2isan initialguessfortheinertiaofthelinkassembly.Thedesiredlinktrajectory(see Figure4—2)wasselectedasfollows(indegrees): ( )=50 0sin(1 5 ))(1 exp( 0 01 3)) (4—45) Forallexperiments,therotorvelocitysignalisobtainedbyapplyinga standardbackwardsdi erencealgorithmtothepositionsignal.Allstateswere initializedtozero.Inaddition,theintegralstructureoftheadaptivetermin(4—43) and(4—44)wascomputedon-lineviaastandardtrapezoidalalgorithm.Thesignum functionforthecontrolschemein(4—10)wasde nedas: ( 2( ))= 1 2 0 1 2 0 0 2=0

PAGE 82

70 0 5 10 15 20 25 30 35 40 50 40 30 20 10 0 10 20 30 40 50 Time [sec]Desired Trajectory [Degrees]Figure4—2:Desiredtrajectoryusedfortheexperiment. 4.4.1Experiment1 Inthe rstexperiment,thecontrollerin(4—9)wasimplementedwithout updatingtheadaptationterm(i.e., =0 ).Thusthecontroltorqueinputgivenin (4—43)takesthefollowingform: ( )= ˆ ( 0)¨ + ( ) Thegainsforthecontrollerthatyieldedthebeststeady-stateperformancewere determinedasfollows: =180 =20 1=40 2=15 (4—46) ThepositiontrackingerrorobtainedfromthecontrollerisplottedinFigure4—3. ThetorqueinputbythecontrollerisdepictedinFigure4—4. 4.4.2Experiment2 Inthesecondexperiment,theupdatelawde nedin(4—44)wasusedtoupdate theadaptivetermde nedin(4—43).Thefollowingcontrolgainswereusedto implementthecontrollerin(4—43):

PAGE 83

71 0 5 10 15 20 25 30 35 40 0.04 0.03 0.02 0.01 0 0.01 0.02 0.03 Time [sec]Position Tracking Error [Degrees]Figure4—3:Positiontrackingerrorwhentheadaptivegainiszero. 0 5 10 15 20 25 30 35 40 25 20 15 10 5 0 5 10 15 20 25 Time [sec]Torque [Nm]Figure4—4:Torqueinputwhentheadaptivegainiszero.

PAGE 84

72 0 5 10 15 20 25 30 35 40 0.04 0.03 0.02 0.01 0 0.01 0.02 0.03 Time [sec]Position Tracking Error [Degrees]Figure4—5:Positiontrackingerrorforthecontrolstructurethatincludestheadaptiveupdatelaw. =180 =20 1=40 2=15 =5 Theinitialguessfortheinertiaofthelinkassemblywassettothesamevalueas inthe rstexperiment.Thepositiontrackingerrorobtainedfromthecontrolleris plottedinFigure4—5.ThetorqueinputbythecontrollerisdepictedinFigure4—6. TheinertiaestimateisdepictedinFigure4—7. 4.5Discussion Thepositiontrackingerrorsforthetwocases(withoutadaptationandwith adaptation)areplottedinFigures4—3and4—5.Theresultsindicatethatthe RMSvalueofthepositiontrackingerrorwhentheadaptivefeedforwardtermis used,isabout5.62%lessthanthecasewhennoadaptationtermisused.This improvementinperformancebytheproposedcontrollerwasobtainedwhileusing similarinputtorqueascanbeseeninFigure4—4andFigure4—6. Inthesepreliminaryresults,theadaptivefeedforwardtermwaslimitedtoone parameter.Amulti-linkrobotictestbediscurrentlybeingusedtodevelopamore completesetofexperimentalresultsthatwouldallowadditionaldynamictermsto

PAGE 85

73 0 5 10 15 20 25 30 35 40 25 20 15 10 5 0 5 10 15 20 25 Time [sec]Torque [Nm]Figure4—6:Torqueinputforthecontrolstructurethatincludestheadaptiveupdatelaw. 0 5 10 15 20 25 30 35 40 1 0 1 2 3 4 5 6 Time [sec]Adaptive Inertia Estimater [kgm2]Figure4—7:Parameterestimateforthemassofthelinkassembly.

PAGE 86

74 beusedintheadaptivefeedforwardterm .Itisexpectedthatwithmoredynamics intheadaptivefeedforwardtermthatthecontrole ortandtrackingerrorwill decreaseevenfurtherincomparisonwithacontrollerthatonlycontainsahighgain feedbackterm. 4.6ConcludingRemarks Anewclassofasymptoticcontrollersisdevelopedthatcontainsanadaptive feedforwardtermtoaccountforlinearparameterizableuncertaintyandahighgain feedbacktermwhichaccountsforunstructureddisturbances.Themotivationfor injectingtheadaptivefeedforwardtermisthatimprovedtrackingperformanceand reducecontrole ortwouldresultfromincludingmoreknowledgeofthesystem dynamicsinthecontrolstructure.Thisideawasveri edbyourpreliminary experimentalresultsthatindicater educedRMStrackingerrorswhenonlyan additionalinertiaestimateisusedinthefeedforwardcontroller.

PAGE 87

CHAPTER5 CONCLUSIONANDRECOMMENDATIONS Motivatedbythefactthatdiscontinuousandpiecewisecontinuousfriction modelsareproblematicforthedevelopmen tofhigh-performancecontinuouscontrollers,anewmodelforfrictionisproposed.Thesimplecontinuouslydi erentiable modelwasshowntoexhibitviscous,Coulombic,static,andStribecke ects,and isinherentlypassive.Anumericalsimulationdemonstratedthemodularityofthe modelforuseindi erentfrictionregimes.Themodelwasalsoproventoclosely approximateexperimentaldata.Asigni cantadvantageoftheproposedmodelis thatthemodelsupportsthedevelopmentofdi erentiablemodel-basedcontrollers. Basedonthedevelopedmodel,semi-globalasymptotictrackingisprovenin thepresenceofaproposedcontinuouslydi erentiablefrictionmodelthatcontains uncertainnonlinearparameterizableterms.Toachievethetrackingresult,an integralfeedbackcompensationtermisusedtoidentifythesystemfrictione ects. ALyapunov-basedstabilityanalysisisprovidedtoconcludethetrackingand frictionidenti cationresults.Experimentalresultsshowtwoordersofmagnitude improvementintrackingcontroloveraproportionalderivative(PD)controller,and aoneorderofmagnitudeimprovementoverthemodel-basedcontroller. AnextensionofthedevelopedtrackingcontrollerispresentedinChapter4 whereanewclassofasymptoticcontrollersisdevelopedthatcontainsanadaptive feedforwardtermtoaccountforlinearparameterizableuncertaintyandahighgain feedbackterm(asinChapter3)whichaccountsforunstructureddisturbances.In comparisonwithpreviousresultsthatusedasimilarhighgainfeedbackcontrol structure,newcontroldevelopment,errorsystemsandstabilityanalysisarguments wererequiredtoincludetheadditionaladaptivefeedforwardterm.Themotivation 75

PAGE 88

76 forincludingtheadaptivefeedforwardtermisthatimprovedtrackingperformance andreducecontrole ortwouldresultfromincludingmoreknowledgeofthe systemdynamicsinthecontrolstructure.Thisideawasveri edbypreliminary experimentalresultsthatindicater educedRMStrackingerrorswhenonlyan additionalinertiaestimateisusedinthefeedforwardcontroller. Futuree ortscanfocusonexpanding/adjustingthedevelopedfrictionmodel toexhibitadditionale ects.E ectsofwear,temperature,humidity,etconthe performanceofthedevelopedmodeltoaccuratelypredictfrictioncanalsobe studiedtomakethemodelmorerobust.Thedevelopedmodelcanalsobemodi ed todevelopamoregenericdynamicmodelforfrictioninsteadofastaticmap betweenvelocityandfriction. Theidenti cationschemedevelopedinChapter3canbeusedtoidentify nonlinearuncertaintiesanddevelopcontrollerstocanceltheire ects.Forexample, thedevelopedschemecanbeusedtoidentifynonlineardisturbancesduringhighspeedmachining,inaircraftapplications,etc. Further,morecontrollerscanbedevelopedbasedontheideaofsegregating thestructuredandunstructureduncertaintiestofurtherimprovesystemperformance.Forexample,neuralnetworksandfuzzylogiccanbeexploredtoaccount forthestructureduncertaintiesalongwiththeintegralfeedbackcompensation strategydevelopedinChapter3toaccountforunstructureduncertaintiesto achieveasymptoticstabilityresults.

PAGE 89

APPENDIXA IDENTIFICATIONANDCOMPENSATIONFORFRICTIONBYHIGHGAIN FEEDBACK Lemma: Theterm ( ) Rcanbede nedasin(2—1)asfollows: ( )= 1(tanh( 2 ) tanh( 3 ))+ 4tanh( 5 )+ 6 (A—1) The rstandsecondtimederivativesof ( ) (i.e., ( ¨ ) and ¨ ( ¨ ... )existand arebounded,provided ( ) ( ) ¨ ( ) ... ( ) L. Proof: Aftertakingthetimederivativeof(A—1)thefollowingexpressioncan beobtained: ( ¨ )=¨ ( 12+ 45 13+ 6) 12¨ k tanh( 2 ) k2(A—2) + 13¨ k tanh( 3 ) k2 45¨ k tanh( 5 ) k2wherethefollowingpropertywasutilized: tanh( ( ))=(1 tanh2( ( ))) ( ) Thetimederivativeof(A—2)canberewrittenasfollows: ¨ ( ¨ ... )= ... ( 12+ 45 13+ 6) 12... k tanh( 2 ) k2(A—3) + 13... k tanh( 3 ) k2 45... k tanh( 5 ) k2 2 12 2¨ ¨ tanh( 2 ) (1 k tanh( 2 ) k2)+2 12 3¨ ¨ tanh( 3 )(1 k tanh( 3 ) k2) 2 42 5¨ ¨ tanh( 5 )(1 k tanh( 5 ) k2) AssumingasinProperty3.3that ( ) ¨ ( ) and ... ( ) Lin D ,itisstraightforwardfrom(A—1),(A—2),and(A—3)that ( ) ( ¨ ) and ¨ ( ¨ ... ) Rexistand arebounded. 77

PAGE 90

78 Lemma: Theauxiliaryerror ( ) de nedin(3—16)as ( )= ( ) ( ) canbeupperboundedasfollows: ( ) ( || || ) k k where ( || || ) issomepositivegloballyinvertiblenondecreasingfunction. Proof: Let ( ) Rbede nedasfollows: ( ¨ 12 ) ( ) ¨ 1 2 ( ) + 2= ( ) ¨ 1 2 ( ) + 2 Theunmeasurable ¨ ( ) canbeexpressedintermsofknownandmeasurablevalues 12 and as ¨ =¨ 2 11+( 1+ 2) 2 Expandingthede nitionof ( ) ,wehave ( ¨ 12 ) 1 2 ( ) + 2+ ( ) (¨ 2 11+( 1+ 2) 2) Recallthede nitionof ( ) : ( ) ( ¨ 0 0 0) = ( ) ¨ Theauxiliaryerror ( ) canthenbewrittenasthesumoferrorspertainingto eachofitsargumentsasfollows:

PAGE 91

79 ( )= ( ¨ 12 ) ( ¨ 0 0 0) = ( ¨ 0 0 0) ( ¨ 0 0 0) + ( ¨ 0 0 0) ( ¨ 0 0 0) + ( ¨ 0 0 0) ( ¨ 0 0 0) + ( ¨ 1 0 0) ( ¨ 0 0 0) + ( ¨ 12 0) ( ¨ 1 0 0) + ( ¨ 12 ) ( ¨ 12 0) ApplyingtheMeanValueTh eoremtofurtherdescribe ( ) ( )= ( 1 ¨ 0 0 0) 1|1= 1( ) (A—4) + ( 2 ¨ 0 0 0) 2|2= 2( ) + ( 3 0 0 0) 3|3= 3(¨ ¨ ) + ( ¨ 4 0 0) 4|4= 4( 1 0) + ( ¨ 15 0) 5|5= 5( 2 0) + ( ¨ 126) 6|6= 6( 0) where 1 ( ) 2 ( ) 3=¨ 4 (0 1) 5 (0 2) 6 (0 ) Fromequation(A—4), ( ) canbeupperboundedasfollows:

PAGE 92

80 || ( ) || ( 1 ¨ 0 0 0) 1|1= 1 || 1|| (A—5) + ( 2 ¨ 0 0 0) 2|2= 2 || 2 11|| + ( ¨ 4 0 0) 4|4= 4 || 1|| + ( ¨ 15 0) 5|5= 5 || 2|| + ( ¨ 126) 6|6= 6 || || ThepartialderivativesofNasexpressedin(A—5)canbeexpressedasfollows: ( 1 ¨ 0 0 0) 1=0 (A—6) ( 2 ¨ 0 0 0) 2= 2 ( ) 2¨ ( ¨ 4 0 0) 4= ( ) ( 2 1) ( ¨ 15 0) 5=1+( 1+ 2) ( ) ( ¨ 126) 6= 1 2 ( ) ( ) Bynotingthat 2= 2( ) (A—7) 4= 1 4( 1 0)= 1(1 4) =( )(1 4) (A—8) 5= 2 5( 2 0)= 2(1 5) =( 1+ 11)(1 5) (A—9) =( + 1( ))(1 5) 6= 6( 0)= (1 6)= (1 6) =(¨ ¨ + 1( )+ 2( + 1( )))(1 6) where (0 1) R =2,4,5,6areunknownconstants,wecanuseProperty3.2

PAGE 93

81 andProperty3.3toupperbound(A—6)asfollows: ( 2 ¨ 0 0 0) 2|2= 2 2( ) ( ¨ 4 0 0) 4|4= 4 4( ) ( ¨ 15 0) 5|5= 5 5( ) ( ¨ 126) 6|6= 6 6( ) where 2456 R aresomepositivefunctionsnondecreasingin ( ) and ( ) andwheretheirargumentscomefromthevariablesof(A—6)and(A—7).Because equations(3—2)and(3—3)giveusthefollowingexpressionsfor and = 1 = 2+ 11itisseenthattheupperboundscanberewrittenintermsof 1, 2,and : ( 2 ¨ 0 0 0) 2|2= 2 2( 12) ( ¨ 4 0 0) 4|4= 4 4( 12) ( ¨ 15 0) 5|5= 5 5( 12) ( ¨ 126) 6|6= 6 6( 12 ) Theboundon ( ) canbefurthersimpli ed: ( ) 2( 12) || 2 11|| + 4( 12) k 1k + 5( 12) k 2k + 6( 12 ) k k Usingtheupperbound || 2 11|| k 2k + 1k 1k

PAGE 94

82 ( ) canbefurtherupperboundedasfollows: ( ) ( 12( 12)+ 4( 12)) k 1k +( 2( 12) + 5( 12)) k 2k + 6( 12 ) k k Usingthede nitionof ( ) R3 in(3—21), ( ) canbeexpressedintermsof ( ) asfollows: ( ) ( 12( 12)+ 4( 12)) k ( ) k +( 2( 12) + 5( 12)) k ( ) k + 6( 12 ) k ( ) k ( 12( 12)+ 4( 12)+ 2( 12)+ 5( 12) + 6( 12 )) k k ( || || ) k k where ( || || ) issomepositivegloballyinvertiblenondecreasingfunction(seeXian etal.[57]). Lemma: De netheauxiliaryfunction ( ) R as ( ) ( ( ) ( 2)) (A—10) Then,if satis es + 1 22(A—11) then Z 0 ( ) (A—12) where isthefollowingpositiveconstant: = || 2( 0) || 2( 0)( 0)

PAGE 95

83 Proof: Integratingbothsidesof(A—10), Z 0 ( ) = Z 0 ( )( ( ) ( 2( ))) (A—13) Substituting(3—4)into(A—13), Z 0 ( ) = Z 0( 2( ) ( )+ 22( )( ) (A—14) 2( ) ( 2( )) 22( ) ( 2( ))) = Z 022( )( ( ) ( 2( ))) + Z 02( ) ( ) Z 02( ) ( 2( )) Afterintegratingthesecondintegralin(A—14)usingintegrationbyparts, Z 0 ( ) = Z 022( )( ( ) ( 2( ))) (A—15) + 2( )( ) | 0 Z 02( )( ) Z 02( ) ( 2( )) = Z 022( )( ( ) 1 2( ( ) ) ( 2( ))) + 2( )( ) 2( 0)( 0) || 2( ) || + || 2( 0) || Z 02|| 2( ) || ( || ( ) || + 1 2 ( ) ) + || 2( ) || ( || ( ) || )+ || 2( 0) || 2( 0)( 0) Thus,itisclearfrom(A—15)thatif satis es(A—11),then(A—12)holds[57].

PAGE 96

APPENDIXB TRACKINGCONTROLINTHEPRESENCEOFFRICTIONBYHIGHGAIN FEEDBACKANDAMODEL-BASEDFEEDFORWARDCOMPONENT-AN EXTENSION Lemma: Theauxiliaryerror ( ) de nedin(4)as ( )= ( ) ( ) canbeupperboundedasfollows: ( ) ( k k ) k k where ( k k ) issomepositivegloballyinvertiblenondecreasingfunction. Proof: ( ) Rin(419)canbeexpressedasfollows: ( ... 12 )= + ( ) 1( 2 11) + 22 + ( ) 1[ { 22 1( 2 11) } + 2( 22)] + ( ) + ( ) ... + ( ) 2+ ( ) ( + 22+ 12 2 11)+ ( )( + 22+ 12 2 11)+ ( ) + ( ) ( + 22+ 12 2 11) + 2 1 2 ( ) wherethefollowingareused: = ( 2 11) (B) = + 22+ 12 2 11 84

PAGE 97

85 Recallthede nitionof ( ) : ( ) ( 0 0 0) = ( ) + ( ) + ( ) ... + ( ) 2 + ( ) + ( ) + ( ) Theauxiliaryerror ( ) canbewrittenasthesumoferrorspertainingtoeach ofitsargumentsasfollows: ( )= ( ... 12 ) ( ... 0 0 0) = ( ... 0 0 0) ( ... 0 0 0) + ( ... 0 0 0) ( ... 0 0 0) + ( ... 0 0 0) ( ... 0 0 0) + ( ... 0 0 0) ( ... 0 0 0) + ( ... 1 0 0) ( ... 0 0 0) + ( ... 12 0) ( ... 1 0 0) + ( ... 12 ) ( ... 12 0) ApplyingtheMeanValueTheoremtofurtherdescribe ( ) ( )= ( 1 ... 0 0 0) 1|1= 1( ) (B) + ( 2 ... 0 0 0) 2|2= 2( ) + ( 3 ... 0 0 0) 3|3= 3( ) + ( 4 0 0 0) 4|4= 4( ... ... ) + ( ... 5 0 0) 5|5= 5( 1 0) + ( ... 16 0) 6|6= 6( 2 0) + ( ... 127) 7|7= 7( 0)

PAGE 98

86 where 1 ( ) 2 ( ) 3 (¨ ¨ ) 4 ( ... ... ) 5 (0 1) 6 (0 2) 7 (0 ) Fromequation(B—2), ( ) canbeupperboundedasfollows: ( ) ( 1 ¨ ... 0 0 0) 1|1= 1 k 1k (B—3) + ( 2 ¨ ... 0 0 0) 2|2= 2 k 2 11k + ( ¨ ... 5 0 0) 5|5= 5 k 1k + ( ¨ ... 16 0) 6|6= 6 k 2k + ( ¨ ... 127) 7|7= 7 k k ThepartialderivativesofNasexpressedin(B—3)canbeexpressedasfollows: ( 1 ¨ ... 0 0 0) 1= 2 ( 1) 2 1 ¨ + ( 1) 1... (B—4) + 2( 1 ) 2 1 2 + 2( 1 ) 1 ¨ + ( 1 ) 1¨ + 2 ( 1) 2 1 1

PAGE 99

87 ( 2 ¨ ... 0 0 0) 2= ( ) ¨ + 2( 2) 2 2 2+2 ( 2) 2+ 2( 2) 2 22¨ + ( 2) 2¨ + ( ) + 2 ( 2) 2 2¨ 2 ( ¨ ... 5 0 0) 5= 2 1( ( ) + ( ) + ( ¨ ) + ( ) )+ 3 1 ( ) 5 ( ¨ ... 16 0) 6= ( ) ( 1+ 2) ( ) ¡ 12+ 2 1+ 2 2¢ 6+1+( 1+ 2)( ( ) + ( ) + ( ) ) ( ¨ ... 127) 7= 77 + ( )( 1+ 2) ( ) ( ) ( ) 7 1 2 ( ) Bynotingthat 1= 1( ) 2= 2( ) 5= 1(1 5) 6= 2(1 6) 7= (1 7)

PAGE 100

88 where (0 1) R =1 2 5 6 7 areunknownconstants,wecanupper bound(B—4)asfollows: ( 1 ¨ ... 0 0 0) 1|1= 1 1( 1) ( 2 ¨ ... 0 0 0) 2|2= 2 2( 12) ( ¨ ... 5 0 0) 5|5= 5 5( 12) ( ¨ ... 16 0) 6|6= 6 6( 12) ( ¨ ... 127) 7|7= 7 7( 12 ) Theboundon ( ) canbefurthersimpli ed: ( ) 1( 1) k 1k + 2( 12) k 2 11k + 5( 12) k 1k + 6( 12) k 2k + 7( 12 ) k k Usingtheupperbound || 2 11|| k 2k + 1k 1k ( ) canbefurtherupperboundedasfollows: ( ) ( 1( 1)+ 12( 12)+ 5( 12)) k 1k +( 2( 12)+ 6( 12)) k 2k + 7( 12 ) k k

PAGE 101

89 Usingthede nitionof ( ) R3 in(4—24), ( ) canbeexpressedintermsof ( ) asfollows: ( ) ( 1( 1)+ 12( 12)+ 5( 12)) k ( ) k +( 2( 12)+ 6( 12)) k ( ) k + 7( 12 ) k ( ) k ( 1( 1)+ 12( 12)+ 5( 12) + 2( 12)+ 6( 12)+ 7( 12 )) k ( ) k Therefore, ( ) ( k k ) k k where ( k k ) issomepositivegloballyinvertiblenondecreasingfunction.

PAGE 102

REFERENCES [1]B.Armstrong-Helouvry, ControlofMachineswithFriction ,KluwerAcademic Publishers,Boston,MA,1991. [2]B.Armstrong-Helouvry,P.Dupont,andC.CanudasdeWit,“ASurvey ofModels,AnalysisToolsandCompensationMethodsfortheControlof MachineswithFriction,” Automatica ,Vol.30,No.7,pp.1083-1138,1994. [3]R.J.AntsaklisandK.M.Passino,eds., AnIntroductiontoIntelligentand AutonomousControl ,KluwerAcademicPublishers,Norwell,MA,1993. [4]N.BarahanovandR.Ortega,“NecessaryandSu cientConditionsfor PassivityoftheLuGreFrictionModel,” IEEETransactionsonAutomatic Control ,Vol.45,No.4,pp.830-832,April2000. [5]P.A.BlimanandM.Sorine,“Easy-to-useRealisticDryFrictionModelsfor AutomaticControl,”in Proceedingsofthe3rdEuropean.ControlConference Rome,Italy,pp.3788-3794,1995. [6]F.P.BowdenandD.Tabor, TheFrictionandLubricationofSolids, ClarendonPress,Oxford,UK,1950. [7]Z.Cai,M.S.deQueiroz,andD.M.Dawson,“RobustAdaptiveAsymptotic TrackingofNonlinearSystemswithAdditiveDisturbance,” Proceedingsof ConferenceonDecisionandControl,andtheEuropeanControlConference Seville,Spain,pp.2877-2882,2005. [8]C.CanudasdeWit,H.Olsson,K.J.Astrom,andP.Lischinsky,“ANew ModelforControlofSystemswithFriction,” IEEETransactionsAutomatic Control ,Vol.40,pp.419-425,March1995. [9]C.CanudasdeWitandP.Lichinsky,“AdaptiveFrictionCompensationwith DynamicFrictionModel,” ProceedingsoftheIFACWorldCongress ,San Francisco,CA,pp.197-202,1996. [10]C.CanudasdeWitandP.Lichinsky,“AdaptiveFrictionCompensationwith PartiallyKnownDynamicFrictionModel,” InternationalJournalonAdaptive ControlandSignalProcessing ,Vol.11,pp.65-80,1997. [11]C.CanudasdeWitandR.Kelly,“Passivity-BasedControlDesignforRobots withDynamicFrictionModel,” ProceedingsoftheIASTEDInternational ConferenceonRoboticsandManufacturing ,Cancun,Mexico,pp.84-87,1997. 90

PAGE 103

91 [12]Y.-Y.Chen,P.-Y.Huang,andJ. -Y.Yen,“Frequency-DomainIdenti cation AlgorithmsforServoSystemswithFriction,” IEEETransactionsonControl SystemsTechnology ,Vol.10,No.5,pp.654-665,September2002. [13]S.-I.ChoandI.-J.Ha,“ALearningApproachtoTrackinginMechanical SystemswithFriction,” IEEETransactionsonAutomaticControl ,Vol.45,No. 1,pp.111-116,January2000. [14]P.Dahl, ASolidFrictionModel ,TheAerospaceCorporation,ElSegundo,CA, Tech.Rep.TOR-0158(3107-18),1968. [15]D.M.Dawson,J.Hu,andT.C.Burg, NonlinearControlofElectricMachinery ,MarcelDekkerInc.,NewYork,NY,1998. [16]W.E.Dixon,A.Behal,D.M.Dawson,andS.Nagarkatti, NonlinearControl ofEngineeringSystems:ALyapunov-BasedApproach ,Birkhuser,Boston, MA,2003. [17]W.E.Dixon,Y.Fang,D.M.Dawson,andT.J.Flynn,“RangeIdenti cation forPerspectiveVisionSystems,” IEEETransactionsonAutomaticControl Vol.48,No.12,pp.2232-2238,2003. [18]P.Dupont,V.Hayward,B.Armstrong,andF.Altpeter,“SingleState ElastoplasticFrictionModels,” IEEETransactionsAutomaticControl ,Vol.47, pp.683-687,April2002. [19]M.R.ElhamiandD.J.Brook eld,“SequentialIdenti cationofCoulomband ViscousFrictioninRobotDrives,” Automatica ,Vol.33,No.3,pp.393-401, 1997. [20]M.Feemster,D.M.Dawson,A.Behal,andW.E.Dixon,“TrackingControl inthePresenceofNonlinearDynamicFrictionalE ects:RobotExtension,” AsianJournalofControl ,Vol.1,No.3,pp.153-168,1999. [21]G.Ferretti,G.Magnani,G.Martucci,P.Rocco,V.Stampacchia, Friction ModelValidationinSlidingandPreslidingRegimeswithHighResolution Encoders ,inExperimentalRoboticsVIII.,B.SicilianoandP.Dario,Eds., Springer-Verlag,NewYork,NY,pp.328-337,2003. [22]G.Ferretti,G.Magnani,andP.Rocco,“SingleandMultistateIntegral FrictionModels,” IEEETransactionsAutomaticControl ,Vol.49,pp.22922297,December2004. [23]B.FriedlandandY.J.Park,“OnAdaptiveFrictionCompensation,” IEEE TransactionsonAutomaticControl ,Vol.45,No.4,pp.1609-1612,April2000. [24]J.A.GreenwoodandJ.B.P.Williamson,“ContactofNominallyFlatSurfaces,” ProceedingsoftheRoyalSocietyofLondon.SeriesA,Mathematicaland PhysicalSciences ,Vol.295,pp.300-319,1966.

PAGE 104

92 [25]S.Gupta,D.Aiken,G.Hu,W.E.Dixon,“Lyapunov-BasedRangeandMotion Identi cationforaNona nePerspectiveDynamicSystem,” Proceedingsofthe IEEEAmericanControlConference ,Minneapolis,Minnesota,June2006,to appear. [26]R.H.A.Hensen,G.Z.Angelis,M.J.G.vandeMolengraft,A.G.deJager, andJ.J.Kok,“Grey-boxmodelingoffriction:Anexperimentalcasestudy,” EuropeanJournalofControl ,Vol.6,No.3,pp.258—267,2000. [27]R.H.A.Hensen,M.J.G.vandeMolengraft,andM.Steinbuch,“Frequency DomainIdenti cationofDynamicFrictionModelParameters,” IEEETransactionsonControlSystemsTechnology ,Vol.10,No.2,pp.191-196,March 2002. [28]R.M.HirschornandG.Miller,“ControlofNonlinearSystemswithFriction,” IEEETransactionsonControlSystemsTechnology ,Vol.7,No.5,pp.588-595, September1999. [29]S.Jain,F.Khorrami,N.Ahmad,S.Sa nkaranarayanan,“FrictionCompensationforDriveswithandwithoutTransmissionCompliance,” Proceedingsofthe IEEEAmericanControlConference ,Albuquerque,NM,pp.2925-2929,1997. [30]H.K.Khalil, NonlinearSystems ,3rded.,Prentice-HallInc.,UpperSaddle River,NJ,2002. [31]S.-J.KimandI.-J.Ha,“AFrequency-DomainApproachtoIdenti cationof MechanicalSystemswithFriction,” IEEETransactionsonAutomaticControl Vol.46,No.6,pp.888-893,June2001. [32]S.-J.Kim,S.-Y.Kim,andI.-J.Ha,“AnE cientIdenti cationMethodfor FrictioninSingle-DOFMotionControlSystems,” IEEETransactionson ControlSystemsTechnology ,Vol.12,No.4,pp.555-563,July2004. [33]M.Krstic,I.Kanellakopoulos,andP.Kokotovic, NonlinearandAdaptive ControlDesign ,Wiley-IntersciencePublications,NewYork,NY,1995. [34]V.Lampaert,J.Swevers,andF.Al-Bender,“Modi cationoftheLeuven IntegratedFrictionModelStructure,” IEEETransactionsonAutomatic Control ,Vol.47,No.4,pp.683-687,April2002. [35]F.Lewis,C.Abdallah,andD.Dawson, ControlofRobotManipulator MacMillanPublishingCo.,NewYork,NY,1993. [36]F.L.Lewis,J.Campos,andR.Selmic, Neuro-FuzzyControlofIndustrial SystemswithActuatorNonlinearities ,SIAM,Philadelphia,PA,2002. [37]T.-L.LiaoandT.-I.Chien,“AnExponentiallyStableAdaptiveFriction Compensator,” IEEETransactionsonAutomaticControl ,Vol.45,No.5,pp. 977-980,May2000.

PAGE 105

93 [38]W.LinandC.Qian,“Adaptivecontrolofnonlinearlyparameterizedsystems: anonsmoothfeedbackframework,” IEEETransactionsonAutomaticControl Vol.47,No.5,pp.757-774,2002. [39]M.Lo er,N.Costescu,andD.Dawson,“Qmotor3.0andtheQmotorRobotic Toolkit-AnAdvancedPC-BasedReal-TimeControlPlatform,” IEEEControl SystemsMagazine ,Vol.22,No.3,pp.12-26,June,2002. [40]C.Makkar,W.E.Dixon,W.G.Sawyer,G.Hu,“Lyapunov-BasedTracking ControlinthePresenceofUncertainNonlinearParameterizableFriction,” ProceedingsoftheIEEEAmericanControlConference ,Portland,Oregon,pp. 1975-1980,2005. [41]C.Makkar,W.E.Dixon,W.G.Sawyer,G.Hu,“Lyapunov-BasedTracking ControlinthePresenceofUncertainNonlinearParameterizableFriction,” IEEETransactionsonAutomaticControl ,submitted. [42]C.Makkar,W.E.Dixon,W.G.Sawyer,andG.Hu,“ANewContinuously Di erentiableFrictionModelforControlSystemsDesign,” IEEE/ASME InternationalConferenceonAdvancedIntelligentMechatronics ,Monterey,CA, pp.600-605,2005. [43]C.Makkar,W.E.Dixon,W.G.Sawyer,andG.Hu,“ANewContinuously Di erentiableFrictionModelforControlSystemsDesign,” SystemandControl Letters ,submitted. [44]M.McIntyre,W.E.Dixon,D.M.Dawson,andI.D.Walker,“FaultDetection andIdenti cationforRobotManipulators,” IEEETransactionsonRobotics Vol.21,No.5,pp.1028-1034,2005. [45]P.M.Patre,W.E.Dixon,C.Makkar,andW.MacKunis,“Asymptotic TrackingforSystemswithStructuredandUnstructuredUncertainties,” ProceedingsofConferenceonDecisionandControl ,SanDiego,CA,2006, submitted. [46]Z.Qu, RobustControlofNonlinearUncertainSystems ,Wiley-Interscience Publications,NewYork,NY,1998. [47]S.SastryandM.Bodson, AdaptiveControl:Stability,Convergence,and Robustness ,Prentice-Hall,UpperSaddleRiver,NJ,1989. [48]T.Schmitz,J.E.Action,J.C.Ziegert,andW.G.Sawyer,“TheDi culty ofMeasuringLowFriction:UncertaintyAnalysisforFrictionCoe cient Measurements,” JournalofTribology ,Vol.127,pp.673-678. [49]A.Serrani,A.Isidori,andL.Marconi,“Semiglobalnonlinearoutputregulation withadaptiveinternalmodel,” IEEETransactionsonAutomaticControl ,Vol. 46,pp.1178—1194,August2001.

PAGE 106

94 [50]S.SivakumarandF.Khorrami,“FrictionCompensationviaVariableStructure Control:RegulationandLowVelocityTracking,” ProceedingsoftheIEEE ConferenceonControlApplications ,Hartford,CT.,pp.645-650,1997. [51]J.J.SlotineandW.Li, AppliedNonlinearControl ,Prentice-Hall,UpperSadle River,NJ,1991. [52]J.Swevers,F.Al-Bender,C.G.Ganseman,andT.Prajogo,“AnIntegrated FrictionModelStructurewithImprovedPreslidingBehaviorforAccurate FrictionCompensation,” IEEETransactionsAutomaticControl ,Vol.45,pp. 675-686,April2000. [53]P.Tomei,“RobustAdaptiveControlofRobotswithArbitraryTransient PerformanceandDisturbanceAttenuation,” IEEETransactionsonAutomatic Control ,Vol.44,pp.654-658,1999. [54]P.Tomei,“RobustAdaptiveFrictionCompensationforTrackingControlof RobotManipulators,” IEEETransactionsonAutomaticControl ,Vol.45,No. 11,pp.2164-2169,November2000. [55]V.I.Utkin, SlidingModesinControlandOptimization ,Springer-Verlag,New York,NY,1992. [56]P.Vedagarbda,D.M.Dawson,andM.Feemster,“TrackingControlof MechanicalSystemsinthePresenceofNonlinearDynamicFrictionE ects,” IEEETransactionsonControlSystemsTechnology ,Vol.7,pp.446-456,July 1999. [57]B.Xian,D.M.Dawson,M.S.deQueiroz,andJ.Chen,“AContinuous AsymptoticTrackingControlStrategyforUncertainMulti-InputNonlinear Systems,” IEEETransactionsonAutomaticControl ,Vol.49,No.7,pp.12061206,July2004. [58]B.Xian,M.S.deQueiroz,D.M.Dawson,andM.McIntyre,“ADiscontinuous OutputFeedbackControllerandVelocityObserverforNonlinearMechanical Systems,” Automatica ,Vol.40,No.4,pp.695-700,April2004. [59]X.Zhang,A.Behal,D.M.Dawson,a ndB.Xian,“OutputFeedbackControl foraClassofUncertainMIMONonlinearSystemsWithNon-SymmetricInput GainMatrix,” ProceedingsofConferenceonDecisionandControl,andthe EuropeanControlConference ,Seville,Spain,pp.7762-7767,2005. [60]T.ZhangandM.Guay,“Commentson‘AnExponentiallyStableAdaptive FrictionCompensator’,” IEEETransactionsonAutomaticControl ,Vol.46, No.11,pp.1844-1845,November2001.

PAGE 107

BIOGRAPHICALSKETCH CharuMakkarwasborninNewDelhi,thecapitalcityofIndia.Shecompleted herBachelorinTechnologyintheyear2003fromGGSIndraprasthaUniversity, Delhi.SheworkedforayearinContinentalDeviceIndiaLtd.,theleadingsemiconductormanufacturing rminIndiaasaProcessRDEngineer.Shethenjoined thenonlinearcontrolsandroboticsresearchgroupintheUniversityofFloridafor MasterofScienceinmechanicalandaerospaceengineeringintheyear2004. ShewilljoinDeloitteConsultingasaSystemsAnalystinJuly2006. 95


Permanent Link: http://ufdc.ufl.edu/UFE0014262/00001

Material Information

Title: Nonlinear Modeling, Identification, and Compensation for Frictional Disturbances
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0014262:00001

Permanent Link: http://ufdc.ufl.edu/UFE0014262/00001

Material Information

Title: Nonlinear Modeling, Identification, and Compensation for Frictional Disturbances
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0014262:00001


This item has the following downloads:


Full Text











NONLINEAR MODELING, IDENTIFICATION, AND COMPENSATION FOR
FRICTIONAL DISTURBANCES
















By

CHARU MAKKAR


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2006


































Copyright 2006

by

Charu Makkar




















This work is dedicated to my parents for their unconditional love, unquestioned

support and unshaken belief in me.















ACKNOWLEDGMENTS

I would like to express sincere gratitude to my advisor, Dr. Warren E. Dixon,

who is a person with remarkable affability. As an advisor, he provided the neces-

sary guidance and allowed me to develop my own ideas. As a mentor, he helped

me understand the intricacies of working in a professional environment and helped

develop my professional skills. I feel fortunate in getting the opportunity to work

with him.

I would also like to extend my gratitude to Dr. W. G. Sawyer for his valuable

discussions. I also appreciate my committee members Dr. Carl D. Crane III, Dr. T.

F. Burks, and Dr. J. Hammer for the time and help they provided.

I would like to thank all my friends for their support and encouragement. I

especially thank Vikas for being my strength and a pillar of support for the last

two years. I would also like to thank my colleague Keith Dupree for helping me

out on those difficult days when I was doing my experiments, and otherwise

aquainting me with American culture almost everyday.

Finally I would like to thank my parents for their love and inspiration, my

sisters Sonia and Madhvi, my brothers-in-law Anil and Rohit and my darling nieces

Kovida, Adya and Hia for keeping up wih me and loving me unconditionally.















TABLE OF CONTENTS
page

ACKNOWLEDGMENTS .................... ......... iv

LIST OF TABLES .................... ............ vii

LIST OF FIGURES .................... ............ viii

ABSTRACT .................... ................ xi

CHAPTER

1 INTRODUCTION ............... .. ............ 1

2 MODELING OF NONLINEAR UNCERTAINTY-FRICTION ...... 9

2.1 Friction Model and Properties ................. ... 10
2.2 Stick-Slip Simulation ................... ...... 11
2.3 Experimental Results ................... ...... 24
2.4 Concluding Remarks ................... ...... 25

3 IDENTIFICATION AND COMPENSATION FOR FRICTION BY HIGH
GAIN FEEDBACK .............................. 28

3.1 Dynamic Model and Properties ................. .... 29
3.2 Error System Development ................... ... 30
3.3 Stability Analysis ................... ...... 32
3.4 Experimental Results ................. ......... 37
3.4.1 Experiment 1 ................... ...... 38
3.4.2 Experiment 2 ................... ...... 44
3.4.3 Experiment 3 ................... ...... 45
3.4.4 Experiment 4 ................... ...... 51
3.4.5 Experiment 5 ................... ...... 51
3.5 Discussion .................... ............ 53
3.6 Concluding Remarks ................... ...... 56

4 TRACKING CONTROL IN THE PRESENCE OF FRICTION BY HIGH
GAIN FEEDBACK AND A MODEL-BASED FEEDFORWARD COMPONENT-
AN EXTENSION .... ....... ....... ............ 58

4.1 Dynamic Model and Properties ................. .... 59
4.2 Error System Development ................... ... 60
4.3 Stability Analysis .................... ........ 64











4.4 Experimental Results
4.4.1 Experiment 1
4.4.2 Experiment 2
4.5 Discussion ......
4.6 Concluding Remarks


5 CONCLUSION AND RECO ENDATIONS . . . .. 75

APPENDIX

A IDENTIFICATION AND COMPENSATION FOR FRICTION BY HIGH
GAIN FEEDBACK ................. ............. 77

B TRACKING CONTROL IN THE PRESENCE OF FRICTION BY HIGH
GAIN FEEDBACK AND A MODEL-BASED FEEDFORWARD COMPONENT-
AN EXTENSION .... ....... ....... ............ 84

REFERENCES ....................... ............ 90

BIOGRAPHICAL SKETCH ................. ........... 95


..........................
..........................
..........................
..........................
..........................















LIST OF TABLES
Table page

3-1 Comparison of tracking results when no external load was applied to the
circular disk. ..................... ............. 54

3-2 Comparison of tracking results when an external load was applied to the
circular disk. . . . . . . . .. 54















LIST OF FIGURES
Figure page

2-1 Friction model as a composition of different effects including: a) Stribeck
effect, b) viscous dissipation, c) Coulomb effect, and d) the combined
m odel . . . . . .. . . .. 12

2-2 Characteristics of the friction model. . . . ... ... 12

2-3 Modular ability of the model to selectively model different friction regimes:
top plot-viscous regime (e.g., hydrodynamic lubrication), middle plot-
Coulombic friction regime (e.g., solid lubricant coatings at moderate slid-
ing speeds), and bottom plot-abrupt change from static to kinetic fric-
tion (e.g., non-lubricous polymers). . . . ... .... 13

2-4 Mass-spring system for demonstrating stick-slip friction. . ... 14

2-5 Friction coefficient vs slip velocity. . . . ..... .... 15

2-6 Block velocity vs time ............................. 16

2-7 Slip velocity vs time. .............................. 16

2-8 Friction coefficient vs time ................... ...... 17

2-9 Friction coefficient vs slip velocity. . . . .. .... 17

2-10 Block velocity vs time ............................. 18

2-11 Slip velocity vs time. .............................. 18

2-12 Friction coefficient vs time ........................... 19

2-13 Friction coefficient vs slip velocity. . . . .. .... 20

2-14 Block velocity vs time ................... ........ 20

2-15 Slip velocity vs time. .............................. 21

2-16 Friction coefficient vs time ........................... 21

2-17 Friction coefficient vs slip velocity. . . . .. .... 22

2-18 Block velocity vs time ................... ........ 22

2-19 Slip velocity vs time. .............................. 23









2-20 Friction coefficient vs time ................... ...... 23

2-21 Testbed for the experiment. ......................... 25

2-22 The proposed friction model very closely approximates the experimen-
tally identified friction term obtained from the adaptive controller devel-
oped in Makkar et al. [41]. Top plot depicts the experimentally obtained
friction torque, middle plot depitcs the friction plot obtained from the
proposed model and the bottom plot depicts a comparison of the two
with solid line indicating experimentally obtained friction torque and
dashed line indicating friction torque obtained from the proposed fric-
tion m odel. . . . . . . . .. ... 26

3-1 Desired disk trajectory.................... ......... 38

3-2 Position tracking error from the PD controller. . . . .... 40

3-3 Position tracking error from the model-based controller with friction feed-
forward terms as described in (3-40) . . . ... .. 40

3-4 Position tracking error from the proposed controller. . . ... 41

3-5 Comparison of position tracking errors from the three control schemes. .41

3-6 Comparison of position tracking errors from the model-based controller
and the proposed controller .......................... 42

3 7 Torque input by the PD controller. ................... .. 42

3-8 Torque input by the model-based controller with friction feedforward
terms as described in (3-40) .................. ..... 43

3-9 Torque input by the proposed controller. . . . ... 43

3-10 Identified friction from the adaptive term in the proposed controller. 44

3-11 Position tracking error from the PD controller. . . .... 45

3-12 Position tracking error from the model-based controller with friction feed-
forward terms as described in (3-40) . . . ... .. 46

3-13 Position taking error from the proposed controller. . . .... 46

3-14 Comparison of position tracking errors from the three control schemes. .47

3-15 Comparison of position tracking errors from the model-based controller
and the proposed controller .......................... 47

3-16 Torque input by the PD controller. . . . .. ... 48









3-17 Torque input by the model-based controller with friction feedforward
terms as described in (3-40)......................... 48

3-18 Torque input by the proposed controller. . . . ... 49

3-19 Identified friction from the adaptive term in the proposed controller 49

3-20 Position tracking error with the proposed controller when the circular
disk was lubricated................... ......... 50

3-21 Torque input by the proposed controller when the circular disk was lu-
bricated.... ................................. 50

3-22 Identified friction from the adaptive term in the proposed controller when
the circular disk was lubricated. . . . ... . 51

3-23 Net external friction induced with no lubrication. The net friction was
calculated by subracting the identified friction term in Experiment 1 from
the identified friction term in Experiment 2. . . . .... 52

3-24 Net external friction induced with lubrication. The net friction was cal-
culated by subracting the identified friction term in Experiment 1 from
the identified friction term in Experiment 3. . . . .... 52

3-25 The friction torque calculated from the model in (2-1) approximates the
experimentally identified friction torque in (3-9). . . ... 53

3-26 Wearing of the Nylon block where it rubbed against the circular disk. 56

3-27 Wear on the circular disk. ......................... .. 57

4-1 The experimental testbed consists of a 1-link robot mounted on a NSK
direct-drive switched reluctance motor. . . . .... 68

4-2 Desired trajectory used for the experiment. . . . .... 70

4-3 Position tracking error when the adaptive gain is zero. . ... 71

4-4 Torque input when the adaptive gain is zero. . . . ... 71

4-5 Position tracking error for the control structure that includes the adap-
tive update law.... ................. ........... 72

4-6 Torque input for the control structure that includes the adaptive update
law ................... ................ 73

4-7 Parameter estimate for the mass of the link assembly. . .... 73















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

NONLINEAR MODELING, IDENTIFICATION, AND COMPENSATION FOR
FRICTIONAL DISTURBANCES

By

Charu Makkar

May 2006

Chair: Warren E. Dixon
Major Department: Mechanical and Aerospace Engineering

For high-performance engineering systems, model-based controllers are

typically required to accommodate for the system nonlinearities. Unfortunately,

developing accurate models for friction has been historically challenging. Despite

open debates in Tribology regarding the continuity of friction, typical models

developed so far are piecewise continuous or discontinuous. Motivated by the fact

that discontinuous and piecewise continuous friction models can be problematic

for the development of high-performance controllers, a new model for friction is

proposed. This simple continuously differentiable model represents a foundation

that captures the major effects reported and discussed in friction modeling and

experimentation. The proposed model is generic enough that other subtleties such

as frictional anisotropy with sliding direction can be addressed by mathematically

distorting this model without compromising the continuous differentiability. From

literature, it is known that if the friction effects in the system can be accurately

modeled, there is an improved potential to design controllers that can cancel

the effects, whereas excessive steady-state tracking errors, oscillations, and limit

cycles can result from controllers that do not accurately compensate for friction.









A tracking controller is developed in Chapter 3 for a general Euler-Lagrange

system based on the developed continuously differentiable friction model with

uncertain nonlinear parameterizable terms. To achieve the semi-global asymptotic

tracking result, a recently developed integral feedback compensation strategy is

used to identify the friction effects on-line, assuming exact model knowledge of the

remaining dynamics. A Lyapunov-based stability analysis is provided to conclude

the tracking and friction identification results. Experimental results illustrate the

tracking and friction identification performance of the developed controller.

The tracking result in Chapter 3 is further extended to include systems with

unstructured uncertainties while eliminating the known dynamics assumption.

The general trend for previous control strategies developed for uncertain dynamics

in nonlinear systems is that the more unstructured the system uncertainty, the

more control effort (i.e., high gain or high frequency feedback) is required to reject

the uncertainty, and the resulting stability and performance of the system are

diminished (e.g., uniformly ultimately bounded stability). The result in Chapter

4 is the first result that illustrates how the amalgamation of an adaptive model-

based feedforward term with a high gain integral feedback term can be used to

yield an asymptotic tracking result for systems that have mixed unstructured and

structured uncertainty. Experimental results are provided that illustrate a reduced

root mean squared tracking error.















CHAPTER 1
INTRODUCTION

The class of Euler-Lagrange systems considered in this thesis are described by

the following nonlinear dynamic model:


M(q)q + V,(q, q)q + G(q) + f(q) = (t). (11)

In (1-1), M(q) E R WX" denotes the inertia matrix, V,(q, q) E R Wx" denotes the

centripetal-Coriolis matrix, G(q) E IR" denotes the gravity vector, f(q) E IR'

denotes friction vector, T(t) E IR represents the torque input control vector,

and q(t), q(t), q(t) E IR' denote the link position, velocity, and acceleration

vectors, respectively. For high-performance engineering systems, model-based

controllers (see Dixon et al. [16]) are typically used to accommodate for the system

nonlinearities. In general, either accurate models of the inertial effects can be

developed or numerous continuous adaptive and robust control methods can be

applied to mitigate the effects of any potential mismatch in the inertial parameters.

Unfortunately, developing accurate models for friction has been historically

problematic. In fact, after centuries of theoretical and experimental investigation,

a general model for friction has not been universally accepted, especially at low

speeds where friction effects are exaggerated. In fact, Armstrong-Helouvry [1]

examined the destabilizing effects of certain friction phenomena (i.e., the Stribeck

effect) at low speeds. To further complicate the development of model-based

controllers for high-performance systems, friction is often modeled as discontinuous;

thus, requiring model-based controllers to be discontinuous to compensate for the

effects.









Motivated by the desire to develop an accurate representation of friction in

systems, various control researchers have developed different analytical models,

estimation methods to identity friction effects, and adaptive and robust methods

to compensate for or reject the friction effects. In general, the dominant friction

components that have been modeled include: static friction (i.e., the torque

that opposes the motion at zero velocity), Coulomb friction (i.e., the constant

motion opposing torque at non-zero velocity), viscous friction (i.e., when full

fluid lubrication exists between the contact surfaces), ,ii%,yi- tr- ie (i.e., different

friction behavior for different directions of motion), Stribeck effect (i.e., at very low

speed, when partial fluid lubrication exists, contact between the surfaces decreases

and thus friction decreases exponentially from stiction), and position dependence

oscillatoryy behavior of the friction torque due to small imperfections on the motor

shaft and reductor centers, as well as the elastic deformation of ball bearings).

Classical friction models are derived from static maps between velocity and

friction force. From a comprehensive survey of friction models in control literature

(see Armstrong-Helouvry [1] and Armstrong-Helouvry et al. [2]), some researchers

believe that dynamic friction effects are necessary to complete the friction model.

Several dynamic friction models have been proposed (see Bliman and Sorine [5]

and Canudas de Wit et al. [8]). These models combine the Dahl model (see Dahl

[14]) with the arbitrary steady-state friction characteristics of the bristle-based

LuGre model proposed by Canudas de Wit et al. [8]. A recent modification to the

LuGre model is given in the Leuven model by Swevers et al. [52] that incorporates

a hysteresis function with non-local memory unlike the Lu-Gre model. The Leuven

model was later experimentally confirmed by Ferretti et al. [21]. However, a

modification to the Leuven model is provided by Lampaert et al. [34] that replaces

the stack mechanism used to implement the hysteresis by the more efficient

Maxwell slip model. Another criticism to the LuGre model has been recently









raised by Dupont et al. [18], who underline a nonphysical drift phenomenon that

arises when the applied force is characterized by small vibrations below the static

friction limit. Recently single and multistate integral friction models have been

developed by Ferretti and Magnani et al. [22] based on the integral solution of the

Dahl model. However, these friction models are based on the assumption that the

friction coefficient is constant with sliding speed and have a singularity at the onset

of slip. Unfortunately, each of the aforementioned models are discontinuous (i.e., a

signum function of the velocity is used to assign the direction of friction force such

as the results by Dupont et al. [18], Ferretti et al.[22], Lampaert et al. [34], and

Swevers et al. [52]), and many other models are only piecewise continuous (e.g., the

LuGre model in [8]). As stated previously, the use of discontinuous and piecewise

continuous friction models is problematic for the development of high-performance

continuous controllers.

Chapter 2 and the preliminary efforts by Makkar et al. [42] and Makkar

and Dixon et al. [43] provide a first step at creating a continuously differentiable

friction model that captures a number of essential aspects of friction without

involving discontinuous or piecewise continuous functions. The proposed model

is generic enough that other subtleties such as frictional anisotropy with sliding

direction can be addressed by mathematically distorting this model without

compromising the continuous differentiability.

If the friction effects in a system can be accurately modeled, there is an

improved potential to design controllers that can cancel the effects (e.g., model-

based controllers); whereas, excessive steady-state tracking errors, oscillations,

and limit cycles can result from controllers that do not accurately compensate for

friction. Given the past difficulty in accurately modeling and compensating for

friction effects, researchers have proposed a variety of (typically offline) friction

estimation schemes with the objective of identifying the friction effects. For









example, an offline maximum likelihood, frequency-based approach (differential

binary excitation) is proposed by Chen et al. [12] to estimate Coulomb friction

effects. Another frequency-based offline friction identification approach was

proposed by Kim and Ha [31]. Specifically, the approach by Kim and Ha [31] uses

a kind of frequency-domain linear regression model derived from Fourier analysis

of the periodic steady-state oscillations of the system. The approach by Kim

and Ha [31] requires a periodic excitation input with sufficiently large amplitude

and/or frequency content. A new offline friction identification tool is proposed

by Kim et al. [32] where the static-friction models are not required to be linear

parameterizable. However the offline optimization result by Kim et al. [32] is

limited to single degree-of-freedom systems where the initial and final velocity

are equal. Another frequency domain identification strategy developed to identify

dynamic model parameters for presliding behavior is given by Hensen and Angelis

et al. [26]. Additional identification methods include least-squares as developed by

Canudas de Wit and Lichinsky [10] and Kalman filtering by Hensen et al. [27].

In addition to friction identification schemes, researchers have developed

adaptive, robust, and learning controllers to achieve a control objective while

accommodating for the friction effects, but not necessarily identifying friction. For

example, given a desired trajectory that is periodic and not constant over some

interval of time, the development by Cho et al. [13] provides a learning control

approach to damp out periodic steady-state oscillations due to friction. As stated

by Cho et al. [13], a periodic signal is applied to the system and when the system

reaches a steady-state oscillation, the learning update law is applied. Liao et

al. [37] proposed a discontinuous linearizing controller along with an adaptive

estimator to achieve an exponentially stable tracking result that estimates the

unknown Coulomb friction coefficient. However, Zhang and Guay [60] describe

a technical error in the result presented by Liao et al. [37] that invalidates the






5


result. Additional development is provided by Zhang and Guay [60] that modifies

the result by Liao et al. [37] to achieve asymptotic Coulomb friction coefficient

estimation provided a persistence of excitation condition is satisfied. Tomei in [53]

proposed a robust adaptive controller where only instantaneous friction is taken

into account (dynamic friction effects are not included).

Motivated by the desire to include dynamic friction models in the control

design, numerous researchers have embraced the LuGre friction model proposed

by Canudas de Wit et al. [8]. For example, the result by Tomei [53] was extended

in [54] to include the LuGre friction model proposed by Canudas de Wit et al.

[8], resulting in an asymptotic tracking result for square integrable disturbances.

Robust adaptive controllers were also proposed by Jain et al. [29] and Sivakumar

and Khorrami [50] to account for the LuGre model. Canudas et al. investigated

the development of observer-based approaches for the LuGre model in [8]. Canudas

and Lichinsky in [9] proposed an adaptive friction compensation method, and

Canudas and Kelly in [11] proposed a passivity-based friction compensation term

to achieve global asymptotic tracking using the LuGre model. Barabanov and

Ortega in [4] developed necessary and sufficient conditions for the passivity of the

LuGre model. Three observer-based control schemes were proposed by Vedagarbda

et al. [56] assuming exact model knowledge of the system dynamics. The results

by Vedagarbda et al. [56] were later extended to include two adaptive observers

to account for selected uncertainty in the model. The observer-based design in

Vedagarbda et al. [56] was further extended by Feemester et al. [20]. Specifically,

a partial-state feedback exact model knowledge controller was developed to achieve

global exponential link position tracking of a robot manipulator by Feemster et al.

[20]. Two adaptive, partial-state feedback global asymptotic controllers were also

proposed in [20] that compensate for selected uncertainty in the system model. In

addition, a new adaptive control technique was proposed by Feemster et al. [20] to









compensate for the nonlinear parameterizable Stribeck effect, where the average

square integral of the position tracking errors were forced to an arbitrarily small

value.

In Chapter 3 and in the preliminary results by Makkar et al. [40] and Makkar

and Dixon et al. [41], a tracking controller is developed for a general Euler-

Lagrange system that contains the new continuously differentiable friction model

with uncertain nonlinear parameterizable terms that was developed in Chapter

2. The continuously differentiable property of the proposed model enabled the

development of a new identification scheme based on a new integral feedback

compensation term. A semi-global tracking result is achieved while identifying the

friction on-line, assuming exact model knowledge of the remaining dynamics.

The control development in Chapter 3 is based on the assumption of exact

model knowledge of the system dynamics except friction. The control of systems

with uncertain nonlinear dynamics, however, is still a much researched area of

focus. For systems with dynamic uncertainties that can be linear parameterized,

a variety of adaptive (e.g., see Krstic [33], Sastry and Bodson [47], and Slotine et

al. [51]) feedforward controllers can be utilized to achieve an asymptotic result.

Some recent adaptive control results have also targeted the application of adaptive

controllers for nonlinearly parameterized systems (see Lin and Qian [38]). Learning

controllers have been developed for systems with periodic disturbances (see

Antsaklis et al. [3]), and recent research has focused on the use of exosystems

by Serrani et al. [49] to compensate for disturbances that are the solution of a

linear time-invariant system with unknown coefficients. A variety of methods

have also been proposed to compensate for systems with unstructured uncertainty

including: various sliding mode controllers (e.g., see Slotine and Li [51], and Utkin

[55]), robust control schemes (see Qu [46]), and neural network and fuzzy logic

controllers (see Lewis et al. [36]). From a review of these approaches a general









trend that can be determined is that controllers developed for systems with more

unstructured uncertainty will require more control effort (i.e., high gain or high

frequency feedback) and yield reduced performance (e.g., uniformly ultimately

bounded stability).

A significant outcome of the new control structure developed by Xian and

Dawson et al. [57] is that asymptotic stability is obtained despite a fairly general

uncertain disturbance. This technique was used by Cai et al. [7] to develop a

tracking controller for nonlinear systems in the presence of additive disturbances

and parametric uncertainties under the assumption that the disturbances are C2

with bounded time derivatives. Xian et al. [58] utilized this strategy to propose

a new output feedback discontinuous tracking controller for a general class of

nonlinear mechanical (i.e., second-order) systems whose uncertain dynamics

are first-order differentiable. Zhang et al. [59] combined the high gain feedback

structure with a high gain observer at the sacrifice of yielding a semi-global

uniformly ultimately bounded result. This particular high gain feedback method

has also been used as an identification technique. For example, the method has

been applied to identify friction (see Makkar et al. [40] and Makkar and Dixon et

al. [41]), for range identification in perspective and paracatadioptric vision systems

(e.g., see Dixon et al. [17], and Gupta et al. [25]), and for fault detection and

identification (e.g., see McIntyre et al. [44]).

The result in Chapter 4 and the preliminary results in Patre et al. [ 15] is

motivated by the desire to include some knowledge of the dynamics in the control

design as a means to improve the performance and reduce the control effort while

eliminating the assumption that the dynamics of the system is completely known.

For systems that include some dynamics that can be segregated into structured

(i.e., linear parameterizable) and unstructured uncertainty, this result illustrates

how a new controller, error system, and stability analysis can be crafted to include






8


a model-based adaptive feedforward term in conjunction with the high gain integral

feedback technique to yield an asymptotic tracking result. This chapter presents

the first result that illustrates how the amalgamation of these compensation

methods can be used to yield an asymptotic result. Experimental results are

presented to reinforce these heuristic notions.















CHAPTER 2
MODELING OF NONLINEAR UNCERTAINTY-FRICTION

Friction force discontinuities have been debated for at least three centuries

dating back to the published works by Amonton in 1699 (e.g., see the classic text

by Bowden and Tabor [6]). Modern publications assume the sliding motion between

solids occurs at a large number of very small and discrete contacts. In contrast

to popular and simple models that assume a structural interaction across regular

and repeating surface features, the contact across engineering surfaces is known

to occur on the tops of asperities or surface protuberances, which, like fractals,

are distributed across all length scales. The number of contacts for engineering

systems is enormous, which led to the seminal work by Greenwood and Williamson

[24], who treated the distributions of these contacts using statistical distributions.

These functions were integrated to give a continuous expression for the relationship

between contact area and pressure. During sliding, and in particular at the

initiation of gross motion (i.e., pre-sliding), the dynamics of individual asperity

contacts breaking and forming is of great theoretical interest; however, due to

the large number of contacts in engineering systems, the dynamics are treated as

continuous following classical statistical methods.

This chapter provides a first step at creating a continuously differentiable

friction model that captures a number of essential aspects of friction without

involving discontinuous or piecewise continuous functions. In unlubricated or

boundary lubricated sliding, wear is inevitable. The proposed friction model

contains time-varying coefficients that can be developed (e.g., modeled by a

differential equation) to capture spatially and temporally varying effects due to

wear. This continuously differentiable model represents a foundation that captures









the major effects reported and discussed in friction modeling and experimentation.

The proposed model is generic enough that other subtleties such as frictional

anisotropy with sliding direction can be addressed by mathematically distorting

this model without compromising the continuous differentiability.

This chapter is organized as follows. The friction model and the associated

properties are provided in Section 2.1. The generality of the model is demonstrated

through a numerical simulation in Section 2.2. Specifically, numerical simulations

are provided for different friction model parameters to illustrate the different

effects that the model captures. Section 2.3 describes that the developed model

approximates the experimental results obtained in Makkar et al. [41].

2.1 Friction Model and Properties

The proposed model for the friction term f(q) in (1 1):


f() = 7l(tanh(720) tanh(73)) + 74 tanh(y ) + y76 (2-1)

where 7i ER Vi = 1, 2, ...6 denote unknown positive constants' The friction model

in (2-1) has the following properties.

It is continuously differentiable and not linear parameterizable.

It is symmetric about the origin.

The static coefficient of friction can be approximated by the term y7 + 74.

The term tanh(72y ) tanh(73q ) captures the Stribeck effect where the friction

coefficient decreases from the static coefficient of friction with increasing slip

velocity near the origin.

A viscous dissipation term is given by 76q.

The Coulombic friction is present in the absence of viscous dissipation and is

modeled by the term 74 tanh(yg7).


1 These parameters could also be time-varying.









The friction model is dissipative in the sense that a passive operator q(t) -

f (.) satisfies the following integral inequality [4]
It
tj(T)f( ((T))dT > -c2


where c is a positive constant, provided q(t) is bounded.

Figures 2-1 and 2-2 illustrate the sum of the different effects and characteris-

tics of the friction model. Figure 2-3 shows the flexibility of such a model.

2.2 Stick-Slip Simulation

The qualitative mechanisms of friction are well-understood. To illustrate

how the friction model presented in (2-1) exhibits these effects, various numerical

simulations are presented in this section. The system considered in Figure 2-4

is a simple mass-spring system, in which a unit mass M is attached to a spring

with stiffness k resting on a plate moving with a velocity xp(t) in the positive

X direction, which causes the block to move with a velocity xbb(t) in the same

direction. The modeled system can be compared to a mass attached to a fixed

spring moving on a conveyor belt. The plate is moving with a velocity that slowly

increases and saturates, given by the following relation:


x = 1 e-t.



The system described by Figure 2-4 is modelled as follows:


MXb(t) + kxb(t) Mgf(xP(t) Xb(t)) = 0


where the term Xp(t) Xb(t) represents the slip velocity, (i.e., the difference between

the plate velocity and block velocity at any instant of time). To demonstrate the

flexibility of the model, model parameters were varied in order to capture the

Stribeck effect, Coulombic friction effect and viscous dissipation. For example,




















0.6





S0.2

-10 -8 -6 -4 -2


- - - - - - - - -


a


2 4 6
slip speed (m/s)


Figure 2-1: Friction model as a composition of different effects including: a)

Stribeck effect, b) viscous dissipation, c) Coulomb effect, and d) the combined

model.


symmetric about the origin


static
Stribeck effect


viscous
Coulomb


slip speed (m/s)


Figure 2-2: Characteristics of the friction model.


8 10


1 _













0.6-


0.4-



0.2-


C

0.6-


0.4-



0.2-


(


0.6-


0.4-



0.2-


2 4 6 8 1
slip speed


slip 4 6 8 speed
slip speed


0.4 0.6
slip speed


Figure 2-3: Modular ability of the model to selectively model different friction
regimes: top plot-viscous regime (e.g., hydrodynamic lubrication), middle plot-
Coulombic friction regime (e.g., solid lubricant coatings at moderate sliding
speeds), and bottom plot-abrupt change from static to kinetic friction (e.g., non-
lubricous polymers).


















VVp





x axis


Figure 2-4: Mass-spring system for demonstrating stick-slip friction.


hydrodynamic lubrication in many operating regimes is viscous, lacking the other

effects, which are easily set to zero in the model. Simple Coulombic friction models

are often good for solid lubricant coatings at moderate sliding speeds. To capture

this effect, the static and viscous terms can be set to zero. For some sticky or non-

lubricous polymers, there exists an abrupt change from static to kinetic friction,

which is captured by making the Stribeck decay very rapid.

A Coulombic friction regime is displayed in Figure 2-5 where the friction

model parameters in (2-1) were set as follows: 71 = 0, 72 = 0, 73 = 0, 74 = 0.1,

75 = 100, 76 = 0. The Coulombic friction coefficient is a constant, opposing

the motion of the block as seen in Figures 2-5 and 2-8. The block velocity, slip

velocity, and the friction force as a function of time are depicted in Figures. 2-6 -

2-8. These figures indicate that the block velocity slowly rises, reaches a maximum

and then begins to oscillate. The slip velocity also rises and then oscillates after

reaching a maximum value. These figures indicate that the friction force causes

the block to move along with the plate until the spring force overcomes the friction

force; hence, the block begins to slip in an opposite direction of the plate velocity













0.1


0.05


0-


-0.05


-0.1

-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
[m/sec]


Figure 2-5: Friction coefficient vs slip velocity.


causing the spring to compress. As the spring releases energy back into the system,

the block velocity exceeds the plate velocity. The magnitude of the constant

friction coefficient results in a constant oscillation between the friction force and

the spring force.

The viscous friction plot in Figure (2-9) is obtained by adjusting the parame-

ters as follows: 71 = 0, 72 = 0, 73 = 0, 74 = 0, 75 = 0, 76 = 0.01. The block

velocity, slip velocity and the friction force are given in Figures 2-10 2-12. The

block velocity in Figure 2-10 slowly decreases as the viscous friction increases as

displayed in Figure 2-9. The viscous coefficient of friction is an order of magnitude

smaller in comparison to the Coulombic friction coefficient, as a result the friction

force is not sufficient enough to sustain the oscillations of the block. The block

eventually comes to rest and constantly slips on the moving plate.

The Stribeck effect in (2-13) is modeled using the following friction model

parameter values: 1i = 0.25, 72 = 100, 73 = 10, 74 = 0, 75 = 0, 76 = 0. The

block velocity, slip velocity and the friction force are plotted in Figures 2-14 2-16.

As seen in Figure 2-13, the Stribeck effect is seen as the high breakaway force at


I I I I I I I I









16










05


04


03




0 1
02-




0


-01 -


-02 -


-03 -


-04 -


0 10 20 30 40 50 60 70 80 90 100
[sec]



Figure 2-6: Block velocity vs time.




















1 2 -
16










12 -
08


06


04


02






0 10 20 30 40 50 60 70 80 90 100
[sec]



Figure 2 7: Slip velocity vs time.








































0 20 40 60 80 100
[sec]


Figure 2-8: Friction coefficient vs time.


x 10-3


-0.2 0 0.2 0.4 0.6 0.8 1
[m/sec]


Figure 2-9: Friction coefficient vs slip velocity.


0.05 k


-0.05









18












01-

008

006

0 04


-002

0-







-0 06




0- 1
0 10 20 30 40 50 60 70 80 90 100 110
[sec]



Figure 2 10: Block velocity vs time.













12







08



06



04



02 -



0 -



02
0 10 20 30 40 50 60 70 80 90 100
[sec]


Figure 2-11: Slip velocity vs time.


























20 40 60 80 100
[sec]


Figure 2-12: Friction coefficient vs time.


the beginning of the motion of the block, which then exponentially decreases.

The block moves with the plate due to the initial friction force. Eventually

enough energy is stored in the spring so that the spring force overcomes the

breakaway friction. The friction force exponentially decays after the breakaway

force is reached. After the block overcomes the breakaway force, the spring force

becomes dominant and causes the block to reverse its direction and move towards

the spring. Since there is no force opposing this motion, a large slip velocity is

exhibited while the spring compresses and releases, pushing the block faster than

the plate.

Figure 2-17 illustrates the sum of the different effects in the friction model in a

stick-slip regime with the following friction model parameters 71 = 0.25, 72 = 100,

73 = 10, 74 = 0.1, 75 = 100, 76 = 0.01. Figures 2-18 2-20 show the velocity of

the block, slip velocity and friction force as a function of time. The block velocity

slowly increases and stores enough energy to reverse the direction of the block.

This storage and release of energy causes the system to oscillate. Figure 2-20 shows

the stick-slip phenomenon.




















0.25

0.2

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

-0.2

-0.25
-3 -2 -1 0 1 2 3 4 5
[m/sec]



Figure 2-13: Friction coefficient vs slip velocity.















2






0


Figure 2-14: Block velocity vs time.









































0





-21
-2 m _
0 10 20 30 40 50 60 70
[sec]



Figure 2-15: Slip velocity vs time.













02



015-



01 -



0 05-


0



-005


-01



-015


Figure 2-16: Friction coefficient vs time.





















0.3


0.2


0.1


0


-0.1


-0.2


-0.3


-0.4
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4
[m/sec]



Figure 2-17: Friction coefficient vs slip velocity.












15


1 -


05-





-05 -


10 20 30 40 50 60 70 80 90
[sec]



Figure 2-18: Block velocity vs time.
































E
15








05-


-05 5---
0 10 20 30 40 50 60 70
[sec]


Figure 2-19: Slip velocity vs time.


0 10 20 30 40 50 60 70 80 90 100
[sec]


Figure 2-20: Friction coefficient vs time.









2.3 Experimental Results

The experimental testbed shown in Figure 2-21 consists of a circular disk

made of Aluminium, mounted on a NSK direct-drive switched reluctance motor

(240.0 Nm Model YS5240-GN001). The NSK motor is controlled through power

electronics operating in torque control mode. The motor resolver provides rotor

position measurements with a resolution of 153600 pulses/revolution at a resolver

and feedback resolution of 10 bits. A Pentium 2.8 GHz PC operating under

QNX hosts the control algorithm, which was implemented via Qmotor 3.0, a

graphical user-interface, to facilitate real-time graphing, data logging, and adjust

control gains without recompiling the program (for further information on Qmotor

3.0, the reader is referred to Loffler et al. [39]). Data acquisition and control

implementation were performed at a frequency of 1.0 kHz using the ServoToGo I/O

board. A 0.315 m X 0.108 m X 0.03175 m rectangular Nylon block was mounted

on a pneumatic linear thruster to apply an external friction load to the rotating

disk. A pneumatic regulator maintained a constant pressure of 15 pounds per

square inch on the circular disk. This testbed was used to implement a tracking

controller with adaptive friction identification developed by the authors in Makkar

et al. [40].

The aim of this experiment was to match the experimentally identified friction

torque using the adaptive term in Makkar et al. [40] and Makkar and Dixon et al.

[41] with the friction torque calculated from the new proposed friction model in

(2-1). The experimentally identified friction is depicted Figure 2-22.

The coefficients in (2-1) were varied to match the experimentally identified

friction torque. The friction torque in (2-1) was calculated as a function of the

rotor velocity with the coefficients chosen as


71 = 34.8 72 = 650 73 = 1 74 = 26 75 = 200 76 = 19.5


(2-2)





























Figure 2-21: Testbed for the experiment.


Figure 2-22 shows a plot of the experimentally determined friction torque, a plot

of friction torque developed from (2-1) and a plot comparing the experimentally

determined friction torque with the developed friction model overlaid.

The experimentally obtained friction torque in Figure 2-22 has viscous and

static friction components and exhibits the Stribeck effect. These facts were taken

into consideration while choosing the constants in (2-1). Values for 7i and 74 were

chosen to account for the static friction, and 76 was chosen to capture the viscous

friction component. As seen in Figure 2-22, the analytical model approximates the

experimental data with the exception of some overshoot. The experimental origin

of the directional frictional anisotropy is discussed in detail in Schmitz et al. [48]

and is attributed to small misalignment between the loading axis and the motor

axis.

2.4 Concluding Remarks

A new continuously differentiable friction model with nonlinear parameter-

izable terms is proposed. This model captures a number of essential aspects of

friction without involving discontinuous or piecewise continuous functions and can























Experimentally identified friction


Time [sec]
Identified friction from model


Time [sec]
Comparison


Time [sec]


Figure 2-22: The proposed friction model very closely approximates the experi-
mentally identified friction term obtained from the adaptive controller developed
in Makkar et al. [41]. Top plot depicts the experimentally obtained friction torque,
middle plot depitcs the friction plot obtained from the proposed model and the
bottom plot depicts a comparison of the two with solid line indicating experimen-
tally obtained friction torque and dashed line indicating friction torque obtained
from the proposed friction model.






27


be modified to include additional effects. The continuously differentiable property

of the proposed model provides a foundation to develop continuous controllers that

can identify and compensate for nonlinear frictional effects. The development of

one such controller is discussed in next chapter.















CHAPTER 3
IDENTIFICATION AND COMPENSATION FOR FRICTION BY HIGH GAIN
FEEDBACK

Motivated by the desire to include dynamics friction models in the control

design, a new tracking controller is developed in this chapter that contains the new

continuously differentiable friction model with uncertain nonlinear parameterizable

terms that was developed in Chapter 2. Friction models are often based on the

assumption that the friction coefficient is constant with sliding speed and have a

singularity at the onset of slip. Such models typically include a signum function of

the velocity to assign the direction of friction force (e.g., see Lampaert et al. [34],

and Swevers et al. [52]), and many other models are only piecewise continuous

(e.g., the LuGre model in [8]). In Makkar et al. [42], Makkar and Dixon et al.

[43], and Chapter 2, a new friction model is proposed that captures a number of

essential aspects of friction without involving discontinuous or piecewise continuous

functions. The simple continuously differentiable model represents a foundation

that captures the major effects reported and discussed in friction modeling and

experimentation and the model is generic enough that other subtleties such as

frictional anisotropy with sliding direction can be addressed by mathematically

distorting this model without compromising the continuous differentiability.

Based on the fact that the developed model is continuously differentiable, a

new integral feedback compensation term originally proposed by Xian et al.

[57] is exploited to enable a semi-global tracking result while identifying the

friction on-line, assuming exact model knowledge of the remaining dynamics.

A Lyapunov-based stability analysis is provided to conclude the tracking and

friction identification results. Experimental results show two orders of magnitude









improvement in tracking control over a proportional derivative (PD) controller,

and a one order of magnitude improvement over the model-based controller.

Experimental results are also used to illustrate that the experimentally identified

friction can be approximated by the model in Makkar et al. [42] and Makkar and

Dixon et al. [43].

This chapter is organized as follows. The dynamic model and the associated

properties are provided in Section 3.1. Section 3.2 describes the development of

errorsystem followed by the stability analysis in Section 3.3. Section 3.4 describes

the experimental set up and results that indicate improved performance obtained

by implementing the proposed controller followed by discussion in Section 3.5 and

conclusion in Section 3.6.

3.1 Dynamic Model and Properties

The class of nonlinear dynamic systems considered are assumed to be modeled

by the general Euler-Lagrange formulation in (1-1) where the friction term f(q) is

assumed to have the form in (2-1) as in Makkar et al. [42] and Makkar et al. [43].

The subsequent development is based on the assumption that q(t) and q(t)

are measurable and that M(q), Vm(q, q), G(q) are known. Moreover, the following

properties and assumptions will be exploited in the subsequent development:

Property 3.1: The inertia matrix M(q) is symmetric, positive definite, and

satisfies the following inequality V y(t) E IR":


1i yll2 < yTM(q)y < m(q) y2 (3 1)

where mT IR is a known positive constant, m(q) E IR is a known positive function,

and I| | denotes the standard Euclidean norm.

OM(q) 02M(q)
Property 3.2: If q(t) E C,, then q, and q 2 exist and are bounded.
aq q)
Moreover, if q(t), q(t) C then Vm,(q, q) and G(q) are bounded.









Property 3.3: Based on the structure of f((q) given in (2-1), f(q), f(q, ), and

f(q, q, q) exist and are bounded provided q(t), q(t), q(t), V (t) E L".
3.2 Error System Development

The control objective is to ensure that the system tracks a desired trajectory,

denoted by qd(t), that is assumed to be designed such that qd(t), qd(t), qd(t),

qd(t) E RI exist and are bounded. A position tracking error, denoted by eC(t)
R", is defined as follows to quantify the control objective:

e = qd q. (32)

The following filtered tracking errors, denoted by e2 (t), r(t) IR, are defined to

facilitate the subsequent design and analysis:

e2 A i + ale (3 3)

r A2 + a262 (3-4)

where al, a2 e IR denote positive constants. The filtered tracking error r(t) is not

measurable since the expression in (3-4) depends on q(t).

After premultiplying (3-4) by M(q), the following expression can be obtained:

M(q)r = M(q)qd + Vm(q, q)q + G(q) (3-5)

+ f(q) 7(t) + M(q)aite + M(q)a2C2

where (1-1), (3-2), and (3-3) were utilized. Based on the expression in (3-5) the

control torque input is designed as follows:

T(t)= M(q)qd + V,(q, q)q + G(q) + M(q)alel + M(q)a22 + p1(t) (3-6)

where p(t) E IR denotes a subsequently designed control term. By substituting
(3-6) into (3-5), the following expression can be obtained:


(3-7)


M(q)r = f () I-t(t).









From (3-7), it is evident that if r(t) -- 0, then p(t) will identify the friction
dynamics; therefore, the objective is to design the control term p(t) to ensure that
r(t) 0. To facilitate the design of p(t), we differentiate (3-7) as follows:

M(q)r = f (q)- M(t) M(q)r. (3-8)

Based on (3-8) and the subsequent stability analysis, p/(t) is designed as follows:

I-(t) = (k + 1)e2(t) (k, + 1)2(to) (3-9)

+ [(k, + 1)a2e2() + sgn(Ce(2())]dT

where k, E R and / IR are positive constants. The time derivative of (3-9) is
given as

/i(t) = (k, + 1)r + psgn(e2). (3-10)

The expression in (3-9) for /(t) does not depend on the unmeasurable filtered
tracking error term r(t). However, the time derivative of f(t) (which is not
implemented) can be expressed as a function of r(t). After substituting (3-10) into

(3-8), the following closed-loop error system can be obtained:

M(q)r = -M(q)r (k, + 1)r e9 3sgn(e) + N(t) (3 11)

where N(t) E IR" denotes the following unmeasurable auxiliary term:

N(q, t) f(q) (q)r + e2. (3-12)
2

To facilitate the subsequent analysis, another unmeasurable auxiliary term Nd(t)
IR' is defined as follows:

Nd(t) A Of (d) = 71729d 7172qd I|tanh(729d) 2 7173qd (3 13)
oqd
+ ,173qd Iltanh(731d)12 + 74754d 7475qd Iltanh('75d) 12 + 6 qd.








The time derivative of (3-13) is given as follows:
O2f(d)..2 Of(d)...
Nd(t) = ~ + q = qd(772 1Y3 + 7475 + 6) (3-14)

-7'172 d Itanh(7'2d) 12 + 7y173 d tanh('73) 2 7475 'd tanh(7s'i ) 12

-2-'y7211 q 2tanh(y72d)[1 ||tanh(72 d) i1] + 27 I 2 tanh(73d)

[1- ||tanh(73)d)2 ] 2] y4-7 qid 2tanh(Q5 d)[1- |tanh(75d) |2].

After adding and subtracting (3-13), the closed-loop error system in (3-11) can be
expressed as follows:

M(q)i = --M(q)r (k, + 1)r e2 sgn(e2) + N(t) + Nd(t) (3-15)
2

where the unmeasurable auxiliary term N(t) E IR' is defined as

N(t) N(t)- Nd(t). (3-16)

Based on the expressions in (3-13) and (3-14), the following inequalities can
be developed:

INd(t) < qd| |772 + 7475 + 6 71731 < Nd (3 17)

Nd(t) < l -7172 +7475 + 76 73 + II 12 (2722 + 271732 + 274752) (318)

< (Nd2

where (Nd, (Nd2 E R are known positive constants.
3.3 Stability Analysis

Theorem 3.1: The controller given in (3-6) and (3-9) ensures that the
position tracking error is regulated in the sense that

el(t) -- 0 as t oc

provided / is selected according to the following sufficient condition:








1
P > (N + -(Nd (3-19)
a2

where (N, and (Nd2 are introduced in (3-17) and (3-18), respectively, and ks is
selected sufficiently large. The control system represented by (3-6) and (3-9) also
ensures that all system signals are bounded under closed-loop operation and that
the friction in the system can be identified in the sense that

f(q) p(t) 0 as t oo.

Proof: Let D C R3'+1 be a domain containing y(t) = 0, where y(t) R3n+1 is
defined as

y(t) A [z(t) v(t)]T (3-20)

where z(t) E IR3n is defined as

z(t) A [eT e rT]T, (3-21)

and the auxiliary function P(t) E IR is defined as

P(t) A e2(to) -e2to)Nd (o) L()dT (3-22)

where/3 IR is nonnegative by design. In (3-22), the auxiliary function L(t) E IR is
defined as

L(t) A rT(Nd(t) sgn(e2)). (3-23)

The derivative P(t) E IR can be expressed as

P(t) = -L(t) = -rT(Nd(t) psgn(e2)). (3-24)

Provided the sufficient condition introduced in (3-19) is satisfied, the following
inequality can be obtained as

L() d < e2 (t) C2(to) Nd(to). (3-25)









Hence, (3-25) can be used to conclude that P(t) > 0. Let V(y, t) : D x [0, oo) --+ R

be a continuously differentiable positive definite function defined as

V(y, t) A eCe + -ee2 + rTM(q)r + P (3-26)
2 2

that can be bounded as

W1i(y) < V(y,t) < W2(y) (3-27)

provided the sufficient condition introduced in (3 19) is satisfied. In (3-27), the

continuous positive definite functions W1(y), W2(y) E IR are defined as

Wi(y) = A, |Iyl2 W2(y) = A2(q) ly 2 (3-28)

where A1, A2(q) E R are defined as

A A -min{l, Ti}
2
A2(q) A max{ m(q), 1}

where mi, m(q) are introduced in (3-1). After taking the time derivative of (3-26),

V(y, t) can be expressed as

V(, t) = rTM(q) + Tr (q)r + ej 2 + 2ET + P.
2

After utilizing (3-3), (3-4), (3-15), and (3-24), V(y, t) can simplified as follows:

V(y,t) = rTN(t) (k, + 1) ||r|2 I 2e22 2a( -e2)2 + 2e~re. (3-29)

Because eT(t)ei(t) can be upper bounded as
1 2 1 2
eCe < 1112 + 112,
2 2

V(y, t) can be upper bounded using the squares of the components of z(t) as
follows:

V(y, t) < r N(t) (k, + 1) |r|12 a2 1Je21 2a,1 Ie11 + Ie 11 + IJ e1 2









By using the fact IIN(t)| < p(|| z) I ||, the expression in (3-29) can be rewritten
as follows:

V(y,t)< -A3 I (k, Ir12 ) r l ) |) (3-30)

where A3 min{2a, 1, a2 1,1} and the bounding function p(I|I|I) E R is a
positive globally invertible nondecreasing function; hence, a1, a2 must be chosen
according to the following conditions:

1
al > a2 > 1.
2

After completing the squares for the second and third term in (3-30), the following
expression can be obtained:

V(y,t) < -A3 l + p(z) (3-31)
4ks

The following expression can then be obtained from (3-31):


V(y,t) < -W(Y) (3-32)

where W(y) = c | I||, for some positive constant c E IR, is a continuous positive
semi-definite function that is defined on the following domain:

D {y E RI3n+I I < p-1(2 3k)}.

The inequalities in (3-27) and (3-32) can be used to show that V(y, t) e C in D;
hence, eC(t),e2(t), and r(t) E Co in D. Given that el(t), e2(t), and r(t) E Co in
D, standard linear analysis methods (e.g., Lemma 1.4 of [15]) can be used to prove
that e(t), 2 (t) Co in D from (3-3) and (3-4). Since el(t), e2(t), r(t) Co in
D, the assumption that qd(t), qd(t), qd(t) exist and are bounded can be used along
with (3-2)-(3-4) to conclude that q(t), q(t), q(t) e Co in D. Since q(t), q(t) E Co
in D, Property 2 can be used to conclude that M(q), Vm(q, q), G(q), and f(q) E LC
in D. From (3-6) and (3-9), we can show that p(t), T(t) E Co in D. Given that








r(t) E Co in D, (3-10) can be used to show that f(t) E Co in D. Property 2 and
Property 3 can be used to show that f(q) and M(q) E Co in D; hence, (3-8) can
be used to show that r(t) E Co in D. Given that r(t) E Co in D, then (3-2)-(3-4)
can be used to conclude that 9 (t) E Co in D. Since e(t), e2(t), i (t) Coo in
D, the definitions for W(y) and z(t) can be used to prove that W(y) is uniformly
continuous in D.

Let S C D denote a set defined as follows:

S y(t)C D V W2(y(t)) < Ai (-(2 )2 (3 33)

The region of attraction in (3-33) can be made arbitrarily large to include any
initial conditions by increasing the control gain ks (i.e., a semi-global type of
stability result) as in Xian et al. [57]. Theorem 8.4 of [30] can now be invoked to
state that

c (t)2 -+0 as t-oo Vy(to) S. (3-34)

Based on the definition of z(t), (3-34) can be used to show that

r(t) 0 as t o Vy(to) E S. (3-35)

Hence, from (3-3) and (3-4), standard linear analysis methods (e.g., Lemma 1.6 of

[15]) can be used to prove that

ei(t) 0 as t o Vy(to) E S.

The result in (3-35) can also be used to conclude from (3-7) that


as t oc Vy(to) E S.


I-(t) f MOt) 0









3.4 Experimental Results

The experimental testbed used for implementing the controller is described in

Chapter 2. The dynamics for the testbed are given as follows:


[Im + 0.5ma2] [q
(t) = -- + f(q) (3-36)
M(q)q
where Im(rotor moment of inertia) = 0.255 kg-m2, m (mass of the circular disk)

= 3.175 kg, a (radius of the disk) = 0.25527 m, and the friction torque f(q) E IR

is defined in (2-1). The control torque input 7(t) given in (3-6) is simplified (i.e.,

the centripetal-Coriolis matrix and gravity terms were omitted) as follows for the

simple testbed:

T(t) = M(q)qd + M(q)ai e + M(q)a22 + (t) (3-37)

where p(t) is the adaptive friction identification term defined in (3-9). The desired

disk trajectory (see Figure 3-1) was selected similar to the one used in Feemster et

al. [20] as follows (in degrees):


qd(t) = 11.25 tan-1(3.0 sin(0.5t))(1 exp(-0.01t3)). (338)



This soft-sinusoidal trajectory was proposed in [20] to emphasize a low-speed

transition from forward to reverse directions. For all experiments, the rotor velocity

signal is obtained by applying a standard backwards difference algorithm to

the position signal. All states were initialized to zero. In addition, the integral

structure of the adaptive term in (3-37) was computed on-line via a standard

trapezoidal algorithm.

















5-

0


-5-

-10


-15
0 5 10 15 20 25 30 35 40
Time[sec]


Figure 3-1: Desired disk trajectory.


3.4.1 Experiment 1

In the first experiment, no external load from the thruster was applied to

the circular disk. In addition to the controller given in (3-36) and (3-37), a PD

controller and a model-based controller were also implemented for comparison. The

PD controller was implemented as:


r(t) = kdl + kpei (3-39)


where kd E IR is the derivative gain and kp IR is the proportional gain. The

model-based controller was implemented with standard friction feedforward terms

as:

r(t) = M(q)qd + M(q)alil + M(q)a2e2 + k sgn (q) + kq + k,q (3-40)


where kc E R is the Coulomb friction coefficient, k, IR is the viscous friction

coefficient, and k, E IR is the static friction coefficient.

The gains for each controller that yielded the best steady-state performance

were determined as follows for the PD controller, model-based controller, and

proposed controller, respectively:









PD controller

kd = 2600 k, = 2600 (3-41)

Model-based controller


a, = 315 02 = 315 kc = 0.0828 (3-42)

k, = 0.0736 k, = 0.104

Proposed controller


k, = 10 = 5 al = 100 a2 = 600. (3-43)

The coefficient k,, k,, and ks in (3-40) were calculated from the friction

identification plot in Fig. 3-10. The coefficient ks is calculated from the peak

friction torque at the onset of a cycle, kc is calculated from the flat portion of the

curve after the initial peak, and k, is calculated from the high peak after the flat

portion of the curve.

The position tracking error from each controller is plotted in Figures 3-2-3-4

respectively. A comparison of the position tracking error from each controller

is seen in Figure 3-5. Figure 3-6 depicts an enlarged view of the comparison

of position tracking errors from the model-based controller and the proposed

controller. The torque input by each controller is depicted in Figures 3-7-3-9

respectively. The friction identification term in (3-9) from the proposed controller

obtained from the experiment is given in Figure 3-10.

The signum function for the control scheme in (3-9) was defined as

1 e2 > 0
sgn(e2t) -1 2 < 0

0 e2 = 0


























aS 1
0)



-0.5
0
0) 0


F- -0.5


0
-1


20
Time [sec]


Figure 3-2: Position tracking error from the PD controller.














0.15


0.1


0.05-


0


S-0.05


-0.1


-0.15


-0.2
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3-3: Position tracking error from the model-based controller with friction

feedforward terms as described in (3 40).








41












0.06


0.04


0.02


0


-0.02


-0.04
a
F--





-0.06


-0.08
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3-4: Position tracking error from the proposed controller.














-- PDController
2 -- Model-based controller
Proposed controller
1.5 -




0.5


0


-0.5 -


a

-1.5 -


-2
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-5: Comparison of position tracking errors from the three control schemes.






















0.05

0
LU

-0.05

0
-0.1
a_


-0 15


5 10 15 20
Time [sec]


25 30 35 40


Figure 3-6: Comparison of position tracking errors from the model-based controller
and the proposed controller.











80

60

40

20

S0

o -20
'*F A-


Figure 3-7: Torque input by the PD controller.


- Model-based controller
Proposed controller



































0 5 10 15 20 25 30 35 40
Time (sec)


Figure 3-8: Torque input by the model-based controller with friction feedforward
terms as described in (3-40).


0 5


Figure 3-9:


10 15 20 25 30 35 40
Time [sec]


Torque input by the proposed controller.

















20

L 0

-20 -

-40

-60

-80
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-10: Identified friction from the adaptive term in the proposed controller.


3.4.2 Experiment 2

In the second experiment, an external friction load was induced on the system.

An external moment load of 12.774 Nm was applied to the circular disk using the

linear thruster (see Figure 2-21). The desired disk trajectory of (3-38) was again

utilized. The control schemes of (3-37),(3-39), and (3-40) were implemented with

the following gain values:

PD controller

kd = 2600 kp = 2600

Model-based controller


al = 350 a2 = 350 kc = 0.0828


k, = 0.0736 k, = 0.104


Proposed controller


k, =8.9 3 = 5.005 a1 =98 a2 = 780.


(3-44)











The corresponding position tracking error from each controller is shown in Figures

3-11-3-13, respectively. A comparison of the position tracking error from each

controller is seen in Figure 3-14. Figure 3-15 depicts an enlarged view of the

comparison of position tracking errors from the model-based controller and the

proposed controller. The control torque input by each controller is shown in

Figures 3-16-3-18, respectively. The friction identification term in (3-9) from the

proposed controller obtained from the experiment is given in Figure 3-19.


5


3-
26





0-2

-3

-4
-5
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-11: Position tracking error from the PD controller.


3.4.3 Experiment 3

In the third experiment, lubrication was used to examine changes in the sys-

tem performance. The circular disk was lubricated with motor oil and the control

scheme in (3-37) was implemented with same control gains as in Experiment 2

to minimize the position tracking error. The resulting position tracking error is

shown in Figure 3-20, the control torque input is shown in Figure 3-21, and the

friction identification term in (3-9) from the proposed controller obtained from the

experiment is given in Figure 3-22.








46





















-06 -
0.2


0.1
0




H-0.





-0.5 -

-0.6
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3-12: Position tracking error from the model-based controller with friction

feedforward terms as described in (3 40).














0.15


0.1 -


0.05-


0


-0.05


-0.1 -
a_
I 005|
H-I


20
Time [sec]


Figure 3 13: Position taking error from the proposed controller.








47









PD controller
5 ---- Model-based controller
Proposed controller





4-


3








-4 -

5
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3 14: Comparison of position tracking errors from the three control

schemes.












0.4 Model-based controller
Proposed controller
0.3


0.2

p 0
LU



-0.21
-0.3
H-


-0.4


-0.5

0.6
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3-15: Comparison of position tracking errors from the model-based con-

troller and the proposed controller.


































20
Time (sec)


Figure 3 16: Torque input by the PD controller.


0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-17: Torque input by the
terms as described in (3 40).


model-based controller with friction feedforward





































0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3 18: Torque input by the proposed controller.


0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-19: Identified friction from the adaptive term in the proposed controller.





















0.1 -


0.05





-0.05-


-0.1


0 5 10 15 20
Time [sec]


25 30 35


Figure 3-20: Position tracking error with the proposed controller when the circular

disk was lubricated.













200

150-

100

50

0

-50-

-100

-150

-200
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 3-21: Torque input by the proposed controller when the circular disk was

lubricated.
















S50 -rIK w r

LL 0

S-50

-100

-150
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-22: Identified friction from the adaptive term in the proposed controller
when the circular disk was lubricated.


3.4.4 Experiment 4

In the fourth experiment, the net external friction induced on the system as

a result of external load applied to the circular disk by the linear thruster was

identified. The friction in the testbed under no-load conditions was identified as in

Experiment 1 using the control gains of Experiment 2. This identified friction term

was subtracted from the identified friction terms obtained from Experiment 2 and

Experiment 3 respectively. The friction between the circular disk and Nylon block

when the disk was not lubricated can be seen in Figure 3-23, and the friction when

the disk was lubricated can be seen in Figure 3-24.

3.4.5 Experiment 5

In the fifth experiment, the experimentally identified friction torque using the

adaptive term in (3-9) was compared with the friction torque model in (2-1). The

friction identified in Figure 3-22 was compared with the model parameters in (2-1)

that were adjusted to match the experimental data. The rotor velocity signal was

obtained by applying a standard backwards difference algorithm to the position

signal, and the friction torque in (2-1) was calculated as a function of this rotor




































20
Time [sec]


Figure 3-23: Net external friction induced with no lubrication. The net friction
was calculated by subracting the identified friction term in Experiment 1 from the
identified friction term in Experiment 2.









20

15

10


to



-5

-10

-15

-20
0 5 10 15 20 25 30 35 40
Time(sec)


Figure 3-24: Net external friction induced with lubrication. The net friction was
calculated by subracting the identified friction term in Experiment 1 from the
identified friction term in Experiment 3.






53



velocity with the constants chosen as


71 = 34.8 72 = 650 73 = 1

74 = 26 7 = 200 76 = 19.5.


The matching of the friction torque with the experimental data is plotted in Figure

3-25.

Comparison
150

100

50

o 0

1 -50

-100

-150
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 3-25: The friction torque calculated from the model in (2-1) approximates
the experimentally identified friction torque in (3-9).


3.5 Discussion

In Experiment 1, it was observed that the position tracking error obtained

from the PD controller in (3-39) deviated around 0.7337 degrees (as seen in Figure

3-2) compared to 0.0506 degrees (as in Figure 3-3) in the model-based controller

in (3-40). This error was further reduced to about 0.0116 degrees when the

proposed controller in (3-37) was implemented (as seen in Figure 3-4). A detailed

comparison of the position tracking errors (in degrees) and control torque (in Nm)

can be seen in Table I.

Similar difference in the order of the magnitude of position tracking errors

was also observed in Experiment 2, when an external friction was applied to the









Table 3-1: Comparison of tracking results when no external load was applied to the
circular disk.


PD Model-based Proposed
controller controller controller
Standard deviation in error 0.7337 0.0506 0.0116
Root mean square value 0.7442 0.0512 0.0116
Standard deviation in torque 33.2926 36.7456 37.5658
Root mean square value 33.7704 37.2826 32.1060

Table 3-2: Comparison of tracking results when an external load was applied to the
circular disk.


PD Model-based Proposed
controller controller controller
Standard deviation in error 1.8225 0.1203 0.0186
Root mean square value 1.8373 0.1212 0.0186
Standard deviation in torque 82.7043 103.0262 72.9409
Root mean square value 83.3753 103.8698 73.9593


circular disk. The position tracking error when the PD controller in (3-39) was

implemented deviated about 1.8225 degrees (see Figure 3-11), whereas in the

model-based controller in (3-40) deviated about 0.1203 degrees as seen in Figure

3-12. Under the same load conditions, when the proposed controller in (3-37)

was implemented, the position tracking error deviated around 0.0186 degrees (see

Figure 3-13). A detailed comparison of the magnitudes of the position tracking

errors (in degrees) and control torque (in Nm) can be seen in Table II.

Experiments 1-3 illustrate an approximate factor of 60 improvement over a PD

controller, and a factor of approximately 3 over a typical exact model knowledge

controller with static and viscous friction feedforward terms (see Figure 3-5, Figure

3-14, Figure 3-6, and 3-15). The significant improvement in the tracking error

is also observed from Table I and Table II, where the rms value of error from the

proposed controller is about two orders of magnitude better than the PD controller

and about one order of magnitude better than the model-based controller. This









improvement in performance from the proposed controller was obtained while using

similar or lower input torque as can be seen in Table I and Table II.

This improvement is based on the fact that the proposed controller contains a

feedforward term that identifies the friction as a general time-varying disturbance.

To develop the friction identification term, the friction model is required to be

continuously differentiable. Experiments 1-5 illustrate the identified friction torque,

and Experiment 5 specifically illustrates that a continuously differentiable friction

model proposed in Makkar et al. [42] and Makkar and Dixon et al. [43] closely

matches the identified friction. Mismatches between the identified friction and the

friction model can be attributed to a number of effects such as wear, nonuniformity

of the material, etc. However, the structure of the friction identification term

enables these anisotropic effects to be captured.

It is interesting to compare the identified friction component from each of

the experiments. The difference between Experiment 1 and 2 is that an external

friction load is applied in Experiment 2. By comparing the identified friction

component in Figure 3-10 and Figure 3-19, as expected, a factor of 2-3 increase

in the static friction component is observed in Figure 3-19 with some increase in

the viscous friction effects. The difference between Experiment 2 and 3 is that

the contact surface is lubricated in Experiment 3. By comparing the identified

friction component in Figure 3-19 and Figure 3-22, the static friction component

is unchanged and a slight reduction in the viscous friction is observed. This is

because the viscous friction of the motor assembly is the dominant viscous effect, as

is illustrated by the results from Experiment 4.

In Experiment 5, the parameters of the nonlinear parameterizable friction

model in (2-1) were varied to match the experimentally obtained friction torque

in Figure 3-22. Figure 3-25 shows that the proposed friction model in (2-1) very

closely approximates the experimental friction torque. However, since friction is




























Figure 3-26: Wearing of the Nylon block where it rubbed against the circular disk.


anisotropic in nature, the magnitude of friction in experimental data is not sym-

metrical about the horizontal axis whereas the friction model in (2-1) approximates

it as symmetric. Hence, future work can focus on mathematically distorting the

model proposed in (2-1) by addition of more terms to make it asymmetric or mak-

ing the unknown coefficients time-varying in order to capture more friction effects

such as wear. Wear is evident on the Nylon block as indicated by groves (see Figure

3-26 and Figure 3-27) in the surface.

3.6 Concluding Remarks

A semi-global asymptotic tracking controller is developed based on the pro-

posed continuously differentiable friction model that contains uncertain nonlinear

parameterizable terms. To achieve the tracking result, an integral feedback com-

pensation term is used to identify the system friction effects. A Lyapunov-based

stability analysis is provided to conclude the tracking and friction identification re-

sults. Experimental results show two orders of magnitude improvement in tracking

control over a proportional derivative (PD) controller, and a one order of magni-

tude improvement over the model-based controller. This controller assumes exact



























Figure 3-27: Wear on the circular disk.


model knowledge of the system dynamics except friction. The control develop-

ment in the next chapter is motivated to eliminate this assumption and design a

controller that can segregate the structured and unstructured uncertainties in the

dynamics while achieving asymptotic tracking.















CHAPTER 4
TRACKING CONTROL IN THE PRESENCE OF FRICTION BY HIGH GAIN
FEEDBACK AND A MODEL-BASED FEEDFORWARD COMPONENT-AN
EXTENSION

The result in this chapter is motivated by the desire to include some knowl-

edge of the dynamics in the control design as a means to improve the performance

and reduce the control effort while eliminating the assumption that the dynamics

of the system is completely known. Specifically, for systems that include some

dynamics that can be segregated into structured (i.e., linear parameterizable) and

unstructured uncertainty, this result illustrates how a new controller, error system,

and stability analysis can be crafted to include a model-based adaptive feedforward

term in conjunction with the high gain integral feedback technique to yield an

asymptotic tracking result.

This chapter presents the first result that illustrates how the amalgamation

of these compensation methods can be used to yield an asymptotic result. Heuris-

tically, the addition of the model-based adaptive feedforward term should reduce

the overall control effort because some of the disturbance has been isolated and

compensated for by a non-high gain feedforward element. Moreover, the addition

of the adaptive feedforward term injects some knowledge of the dynamics in the

control structure, which could lead to improved performance.

Preliminary experimental results are presented to reinforce these heuristic

notions. Specifically, the presented controller was implemented on a simple one-

link robot testbed and demonstrated reduced tracking error. For this testbed,

the only dynamics that were included in the feedforward term included the

inertia of the linkage system. This chapter is organized as follows. The dynamic









model and the associated properties are provided in Section 4.1. Section 4.2

describes the development of errorsystem followed by the stability analysis in

Section 4.3. Section 4.4 describes the experimental set up and results that indicate

improved performance obtained by implementing the proposed controller followed

by discussion in Section 4.5 and conclusion in Section 4.6.

4.1 Dynamic Model and Properties

The class of nonlinear dynamic systems considered are described in Chapter 1

and Chapter 2 by the general Euler-Lagrange formulation in (1-1) where the term

f(q) E RI denotes a general nonlinear disturbance (e.g., friction as in Makkar et al.
[42] and Makkar and Dixon et al. [43], unmodeled effects).

The subsequent development is based on the assumption that q(t) and

q(t) are measurable and that M(q), Vm(q, q) and G(q) are unknown. Moreover,
the following properties and assumptions will be exploited in the subsequent

development:

Property 4.1: The inertia matrix M(q) as in (3-1) is symmetric, positive definite,

and satisfies the following inequality V y(t) E R":


mi Iyll 2< yTM(q)y < m(q)I y 11 (4 1)

where mT IR is a known positive constant, m(q) E IR is a known positive function,

and I I I denotes the standard Euclidean norm.

Property 4.2: If q(t), q(t) oo, then V,(q, ), f() and G(q) are bounded.
8M(q) 22M(q) 8V,(q,q) o2V,(q,4)
Moreover, if q(t), q(t) E ,, then, M(q) 2
9q q 2 9q q2
OVm(q, ) &2Vm(q,4) af(4) 02f() 2G(q) 92G(q)
,9 ) 2,) 0 ) 02 q ) and exist and are bounded.
P q q2 4q e q2 e s c q2
Property 4.3: There exist positive scalar constant (ci such that


(4-2)


JIV.(q, 4) I < (,, 1111, Vq, 4ER .









Property 4.4: The dynamic equation in (1-1) can be linear parameterized as


Yd (qd, qd, qd)8 0 M(qd)qd + Vm(qd, qd)d + G(qd) (4-3)

where 0 E IR contains the constant unknown system parameters, and the desired

regression matrix Yd (qd, qd, qd) E 7- contains known functions of the desired

link position, velocity, and acceleration, qd (t) qd (t) qd (t) E IR', respectively. The

desired trajectory is assumed to be designed such that qd(t), qd(t), qd(t), qd(t),

qd(t) E IR" exist and are bounded.

4.2 Error System Development

The control objective is to ensure that the system tracks a desired time-

varying trajectory despite structured and unstructured uncertainties in the dynamic

model. To quantify this objective, a position tracking error as in (3-2), denoted by

el(t) E IR", is defined as
e = qd q. (4-4)

To facilitate the subsequent analysis, filtered tracking errors as in (3-3 and 3-4),

denoted by e2(t), r(t) E R, are also defined as

e2 = 1 + ale1 (4-5)

r = e2 + a22 (4-6)

where a1, a2 E R denote positive constants. The filtered tracking error r(t) is not

measurable since the expression in (4-6) depends on q(t).

The open-loop tracking error system can be developed by premultiplying (4-6)

by M(q) and utilizing the expressions in (1-1), (4-3), (4-4), and (4-5) to obtain
the following expression:

M(q)r = YdO + W T(t) (4-7)

where Yd(t)O E IR" is defined in (4-3), and the auxiliary signal W(el,e2,t) E IR" is









defined as

W = M(q) (4d + alel + a2e2) + V,(q, q)q + G(q) + f(q) Yd. (4-8)

Based on the expression in (4-7), the control torque input is designed as follows:

7 = YdO + p. (4-9)

In (4-9), i(t) e IR' denotes a high-gain feedback control term defined as

pI(t) = (k + 1)e2(t) (k, + 1)e2(to) (4-10)

+ [(ks + 1)a22 () + 3sgn(C2())]d

where ks,/3 IR are positive constant control gains as in (3-9), and 0(t) E IR

denotes a parameter estimate vector generated on-line according to the following

update law:

0 = FdTr (4-11)

with F e IRpx being a known, constant, diagonal, positive-definite, adaptation gain

matrix. Since Yd(t) is only a function of the known time varying trajectory, (4-11)

can be integrated by parts as follows:


0(t = 0to) + FYTe2 () dTe2 () a2dTe2 (a) da (4-12)

so that the parameter estimate vector 0(t) implemented in (4-9) does not depend

on the unmeasurable signal r(t).

Remark 4.1: The control design in (4 9) is similar to the results in Xian

et al. [57]. However, previous designs based on [57] could only compensate for

uncertoifit/i in the system through the high gain feedback term [t(t). Through the

new error system development and ,stbilitfi analysis methods presented in the

current result, an adaptive f/, ,lf. ..',i,l term can also be used to compensate for

system uncertainty. This fl-.ribilitfi presents a significant advantage because it









allows more system dynamics to be incorporated in the control design. Sp .:I' 'ill;,

if some of the system uncertainty can be segregated into a linear parameterizable

form, then the model-based adaptive f ,If.., it,,11, term can be injected to compensate

for the uncertainty instead of just relying on the non-model based high gain feedback

term. Heuristically, this contribution should improve the tracking performance and

reduce the control effort. Preliminary experimental results on a simple one-link

robot manipulator provide some validation of this heuristic idea.

The closed-loop tracking error system can be developed by substituting (4-9)

into (4-7) as

M(q)r = YdO + W p(t) (4-13)

where 0(t) E IR represents the parameter estimation error vector defined as

0 = 0 0. (4-14)

To facilitate the subsequent stability analysis (and to illustrate some insight into

the structure of the design for I(t)), the time derivative of (4-13) is determined as


M(q)r = -M(q)r + YdO YdO + W f(t). (4-15)

The time derivative of (4-10) is given as


/(t) = (ks + 1)r + psgn(e2). (4-16)

After substituting (4-11) and (4-16) into (4-15), the following closed-loop error

system can be obtained:


M(q)i = -M(q)r + YdO + N f(t) e2 (4-17)

where N(ei, e2, r, t) E IR denotes the following unmeasurable auxiliary term:

N =-YdFYr + W + e2 M(q)r. (4-18)
2









After substituting for the time derivative of W(eC, e2, t), auxiliary term in (4-18)

can be expressed as
T 9M a(q) 2


v = -YdlYd r + q [aLe2 aC a262]2 + M (q)air a2e2 (4-1

a\q9M(q) ... m(q, q)
a a (e2 e) + a (r 2+2) + g d + + q 2
VmG(q, q) (. Gf(q (q) .. 1 M(q)
+ qq + ,(q, q) + q + q 2- O + "e qr.
8,q Oq 89q 2 Oq

To further facilitate the subsequent analysis, another unmeasurable auxiliary term

Nd(qd, qd, qd, t) E RI is defined as


N f (fd) M+ (qd) avmq+(qd, d)
ad qd + qdqd + M(qd)qd + qd
O6(d )qd )qd
v, (qd, qd) G(qd)
+ V(qdqdd + Vm(qd, qd)qd + 9a(qd) YdO-
+ 9qd aqd


(4-20)


After adding and subtracting (4-20), the closed-loop error system in (4-17) can be

expressed as


1-
M(q)r = -M(q)r + YdO f(t) e2+ N(t) + Nd(t)
2


(4-21)


where the unmeasurable auxiliary term N(eC, e2, r, t) E IR" is defined as


N(t) = N(t)- Nd(t).


(4-22)


In a similar manner as in Xian et al. [57], the Mean Value Theorem can be used to

develop the following upper bound


N(t)



(4-23)


where z(t) E IR3n is defined as


(4-24)


(t) A [eT e rT]T.


The following inequalities can be developed based on the expression in (4-20) and
its time derivative:

Nd(t)| < (Nd d(t) < (Nd2 (4-25)


f


9)









where (N, (Nd2 IR are known positive constants.
4.3 Stability Analysis

Theorem 4.1: The controller given in (4-9), (4-10), and (4-12) ensures that
all system signals are bounded under closed-loop operation and that the position
tracking error is regulated in the sense that

e(t) -- 0 as t oc

provided the control gain ks introduced in (4-10) is selected sufficiently large, and
p is selected according to the following sufficient condition:

1
S> (Nd + -(Nd2 (4-26)
a2

where (Nd and (Nd2 are introduced in (4-25).
Proof: Let D C IR3+p+1 be a domain containing y(t) = 0, where y(t)
R3n+p+l is defined as
y(t) = [zT( t) T(]) P ]T (4-27)

and the auxiliary function P(t) E IR is defined as

P(t) = 3 |e2(tO) etoN -2Nd0) L(7)d. (4-28)

In (4-28), the auxiliary function L(t) E IR is defined as

L(t) = rT(Nd(t) psgn(e2)). (4-29)

The derivative P(t) E IR can be expressed as

P(t) = -L(t) = -rT(Nd(t) psgn(e2)). (4-30)

Provided the sufficient condition introduced in (4-26) is satisfied, the following
inequality can be obtained (see Xian et al. [57])

SL()d < 2 t) ( 2 (lo)Nd(to). (4-31)









Hence, (4-31) can be used to conclude that P(t) > 0.

Let V(y, t) : D x [0, oc) --+ R be a continuously differentiable positive definite

function defined as

V(y, t) = e"e + -e e2 + + -rT+ -10 (4-32)
e2 2 2

which satisfies the following inequalities:

Ui(y) < V(y,t) < U2(y) (4-33)

provided the sufficient condition introduced in (4-26) is satisfied. In (4-33), the

continuous positive definite functions U (y), U2(y) E R are defined as

UI(y) = 71 ily 2 U2(y) = 92(q) I Y12 (4 34)

where ll, r2(q) E IR are defined as

171 A min{l,Tl,Amin {F-}}
11
R2(q) A max 2m(q), 2Amax F 1}1}

where mi, m(q) are introduced in (4-1) and Amin {} Amax {'} denote the minimum

and maximum eigenvalue of the argument respectively. After taking the time

derivative of (4-32), 1V(y, t) can be expressed as

V(y t)= rM(q)r + -rM(q)r + ee2 + 2ef1 + P 0-
2

After utilizing (4-5), (4-6), (4-11), (4-16), (4-21), and (4-30), V(y, t) can be

simplified as follows:


V(y, t) = rN(t) (k, + 1) |r| 2 Ie21 2a- Ie1112 + 2e i. (4 35)

Because et(t)eC(t) can be upper bounded as
1 1 2
e e < 1 1112 + I J211 ,
2 2









V(y, t) can be upper bounded using the squares of the components of z(t) as
follows:

V(y,t) < rTN(t) (k, + 1) r 2- a 2 1 2 2a ||112 + IICI2 + eIC 2

By using (4-23), the expression in (4-35) can be rewritten as follows:

V(y,t) < -A3 11-(k, Ir1-2 p(z)Ir I.1||) (4-36)

where A3 A min{2a, 1, a2 1,1}, and the bounding function p( llz ) E R is a

positive globally invertible nondecreasing function; hence, al, a2 must be chosen
according to the following conditions:

1
a, > a, a2 > 1.
2

After completing the squares for the second and third term in (4-36), the following

expression can be obtained:


'(y,t) < -A3I + 4 + (4-37)

The following expression can then be obtained from (4-37)


V(y,t) < -U(y) (4-38)

where U(y) = c I. ||, for some positive constant c R, is a continuous positive

semi-definite function that is defined on the following domain:

D A {y Re3n++l I Ill < p-1(2 A,}.

The inequalities in (4-33) and (4-38) can be used to show that V(y, t) e L

in D; hence, ei(t), e2(t), r(t), and 0(t) E Lm in D. Given that ei(t), e2(t), and

r(t) E L in D, standard linear analysis methods can be used to prove that e(t),

2 (t) E c in D from (4-5) and (4-6). Since 0 e RP contains the constant unknown








system parameters and O(t) E C in D, (4-14) can be used to prove that O(t) E Co
in D. Since el(t), 2 (t), r(t) Coo in D, the assumption that qd(t), qd(t), qd(t)
exist and are bounded can be used along with (4-4)-(4-6) to conclude that q(t),

q(t), q(t) E Co in D. The assumption that qd(t), qd(t), qd(t) qd(t), q d(t) exist
and are bounded along with Property 4 can be used to show that Yd (qd, qd, qd),

Yd (qd, qd, qd), and Yd (qd, qd, qd) Coo in D. Since q(t), q(t) E C in D, Property
2 can be used to conclude that M(q), V,(q, q), G(q), and f(q) E Co in D. From
(4-9) and (4-10), we can show that /(t), 7(t) E Co in D. Given that r(t) E Co
in D, (4-16) can be used to show that /(t) E Co in D. Since q(t), q(t) E Co
in D, Property 2 can be used to show that Vm,(q, q), G(q), f(q) and M(q) E Coo
in D; hence, (4-15) can be used to show that r(t) E Co in D. Since el(t), e2(t),
r(t) E Co in D, the definitions for U(y) and z(t) can be used to prove that U(y) is
uniformly continuous in D.
Let S C D denote a set defined as follows:

A y(t)C D U2(y(t)) < rl (p-1(2V/ ) (4-39)

The region of attraction in (4-39) can be made arbitrarily large to include any
initial conditions by increasing the control gain k, (i.e., a semi-global type of
stability result) as in Xian et al. [57]. Theorem 8.4 of [30] can now be invoked to
state that
cI(t) 2- 0 as t-oo Vy (to)E S. (4-40)

Based on the definition of z(t), (4-40) can be used to show that

r(t) 0 as t o Vy(to) E S. (4-41)

Hence, from (4-5) and (4-6), standard linear analysis methods can be used to
prove that


ei(t) 0 as t o Vy(to) E S.



























Figure 4-1: The experimental testbed consists of a 1-link robot mounted on a NSK
direct-drive switched reluctance motor.

4.4 Experimental Results

The testbed depicted in Figure 4-1 was used to implement the developed

controller. The testbed consists of a one-link robotic arm of unknown inertia

mounted on a NSK direct-drive switched reluctance motor (240.0 Nm Model

YS5240-GN001). The NSK motor is controlled through power electronics operating

in torque control mode. The motor resolver provides rotor position measurements

with a resolution of 614400 pulses/revolution at a resolver and feedback resolu-

tion of 12 bits. A Pentium 2.8 GHz PC operating under QNX hosts the control

algorithm, which was implemented via Qmotor 3.0, a graphical user-interface, to

facilitate real-time graphing, data logging, and adjustment of control gains without

recompiling the program (for further information on Qmotor 3.0, the reader is

referred to Loffler et al. [39]). Data acquisition and control implementation were

performed at a frequency of 1.0 kHz using the ServoToGo I/O board.

The dynamics for the testbed are given as follows:


(4-42)


T (t) = Jq + f (q)









where J E R denotes the inertia of the link assembly and f(q) E R represents the

friction torque. The control torque input 7(t) given in (4-9) is simplified (i.e., the

centripetal-Coriolis matrix and gravity terms were omitted) for the simple testbed

as:

T(t) = Jd + 1(t) (4 43)

where J(t) denotes the adaptive estimate for the inertia of the link assembly, and

[p(t) is the adaptive control term defined in (4-10). The adaptive estimate J(t) is

updated according to the following update law:


J(t)= (to) + f rqr (4-44)


where F IR is a constant positive adaptive gain and J (to) = 0.179 kg-m2 is an

initial guess for the inertia of the link assembly. The desired link trajectory (see

Figure 4-2) was selected as follows (in degrees):


qd(t) = 50.0 sin(1.5t))(1 exp(-0.01t3)). (4-45)



For all experiments, the rotor velocity signal is obtained by applying a

standard backwards difference algorithm to the position signal. All states were

initialized to zero. In addition, the integral structure of the adaptive term in (4-43)

and (4-44) was computed on-line via a standard trapezoidal algorithm. The signum

function for the control scheme in (4-10) was defined as:


1 : 2 > 0
sgn(e2(t)) -1 2 < 0

0 e2 = 0












50
40
30

20
10
0
(t 0
I- -10-

-20
-30
-40
-50
0 5 10 15 20 25 30 35 40
Time [sec]


Figure 4-2: Desired trajectory used for the experiment.


4.4.1 Experiment 1

In the first experiment, the controller in (4-9) was implemented without

updating the adaptation term (i.e., F = 0). Thus the control torque input given in

(4-43) takes the following form:


r(t) = (to) 4d + (t)


The gains for the controller that yielded the best steady-state performance were

determined as follows:



k, =180 = 20 a1 =40 a2 = 15. (4-46)


The position tracking error obtained from the controller is plotted in Figure 4-3.

The torque input by the controller is depicted in Figure 4-4.

4.4.2 Experiment 2

In the second experiment, the update law defined in (4-44) was used to update

the adaptive term defined in (4-43). The following control gains were used to

implement the controller in (4-43):
































0.01 T' r'rr

0.02 -


0.03


n 04


1 5 10 15 20
Time [sec]


25 30 35


Figure 4 3: Position tracking error when the adaptive gain is zero.













25

20

15

10

5
ET
0 0
F-5


Figure 4-4: Torque input when the adaptive gain is zero.















0.01

0

-0.01

-0.02

-0.03

0 5 10 15 20 25 30 35 40
Time [sec]


Figure 4-5: Position tracking error for the control structure that includes the adap-
tive update law.




k, =180 = 20 a1 = 40 02 = 15 F = 5.

The initial guess for the inertia of the link assembly was set to the same value as

in the first experiment. The position tracking error obtained from the controller is

plotted in Figure 4-5. The torque input by the controller is depicted in Figure 4-6.

The inertia estimate is depicted in Figure 4-7.

4.5 Discussion

The position tracking errors for the two cases (without adaptation and with

adaptation) are plotted in Figures 4-3 and 4-5. The results indicate that the

RMS value of the position tracking error when the adaptive feedforward term is

used, is about 5.62% less than the case when no adaptation term is used. This

improvement in performance by the proposed controller was obtained while using

similar input torque as can be seen in Figure 4-4 and Figure 4-6.

In these preliminary results, the adaptive feedforward term was limited to one

parameter. A multi-link robotic testbed is currently being used to develop a more

complete set of experimental results that would allow additional dynamic terms to






























5

0
F 5-

-10 -

-15-

-20

-25
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 4-6: Torque input for the control structure that includes the adaptive up-

date law.

















5

E
4-


E 3-









0-


-1
0 5 10 15 20 25 30 35 40
Time [sec]



Figure 4-7: Parameter estimate for the mass of the link assembly.









be used in the adaptive feedforward term. It is expected that with more dynamics

in the adaptive feedforward term that the control effort and tracking error will

decrease even further in comparison with a controller that only contains a high gain

feedback term.

4.6 Concluding Remarks

A new class of asymptotic controllers is developed that contains an adaptive

feedforward term to account for linear parameterizable uncertainty and a high gain

feedback term which accounts for unstructured disturbances. The motivation for

injecting the adaptive feedforward term is that improved tracking performance and

reduce control effort would result from including more knowledge of the system

dynamics in the control structure. This idea was verified by our preliminary

experimental results that indicate reduced RMS tracking errors when only an

additional inertia estimate is used in the feedforward controller.















CHAPTER 5
CONCLUSION AND RECO1l i\ [ENDATIONS

Motivated by the fact that discontinuous and piecewise continuous friction

models are problematic for the development of high-performance continuous con-

trollers, a new model for friction is proposed. The simple continuously differentiable

model was shown to exhibit viscous, Coulombic, static, and Stribeck effects, and

is inherently passive. A numerical simulation demonstrated the modularity of the

model for use in different friction regimes. The model was also proven to closely

approximate experimental data. A significant advantage of the proposed model is

that the model supports the development of differentiable model-based controllers.

Based on the developed model, semi-global asymptotic tracking is proven in

the presence of a proposed continuously differentiable friction model that contains

uncertain nonlinear parameterizable terms. To achieve the tracking result, an

integral feedback compensation term is used to identify the system friction effects.

A Lyapunov-based stability analysis is provided to conclude the tracking and

friction identification results. Experimental results show two orders of magnitude

improvement in tracking control over a proportional derivative (PD) controller, and

a one order of magnitude improvement over the model-based controller.

An extension of the developed tracking controller is presented in Chapter 4

where a new class of asymptotic controllers is developed that contains an adaptive

feedforward term to account for linear parameterizable uncertainty and a high gain

feedback term (as in Chapter 3) which accounts for unstructured disturbances. In

comparison with previous results that used a similar high gain feedback control

structure, new control development, error systems and stability analysis arguments

were required to include the additional adaptive feedforward term. The motivation









for including the adaptive feedforward term is that improved tracking performance

and reduce control effort would result from including more knowledge of the

system dynamics in the control structure. This idea was verified by preliminary

experimental results that indicate reduced RMS tracking errors when only an

additional inertia estimate is used in the feedforward controller.

Future efforts can focus on expanding/adjusting the developed friction model

to exhibit additional effects. Effects of wear, temperature, humidity, etc on the

performance of the developed model to accurately predict friction can also be

studied to make the model more robust. The developed model can also be modified

to develop a more generic dynamic model for friction instead of a static map

between velocity and friction.

The identification scheme developed in Chapter 3 can be used to identify

nonlinear uncertainties and develop controllers to cancel their effects. For example,

the developed scheme can be used to identify nonlinear disturbances during high-

speed machining, in aircraft applications, etc.

Further, more controllers can be developed based on the idea of segregating

the structured and unstructured uncertainties to further improve system perfor-

mance. For example, neural networks and fuzzy logic can be explored to account

for the structured uncertainties along with the integral feedback compensation

strategy developed in Chapter 3 to account for unstructured uncertainties to

achieve asymptotic stability results.













APPENDIX A
IDENTIFICATION AND COMPENSATION FOR FRICTION BY HIGH GAIN
FEEDBACK
Lemma: The term f(q) E IR" can be defined as in (2-1) as follows:

f(q) = 7y(tanh(y72) tanh(73')) + 74 tanh(yq) + y6q. (A-1)

The first and second time derivatives of f(q) (i.e., f(q, q) and f(qV, q, ') exist and
are bounded, provided q(t), q(t), q(t), V(t) E LC.-
Proof: After taking the time derivative of (A-1) the following expression can
be obtained:

f(q, q) = (772 + 7475 7'73 + 76) 7172q tanh(72) ||2 (A 2)

+ 7Y73i ||tanh(73' ) I2 74753 ||tanh(75y) ||2

where the following property was utilized:

d tanh(u(t)) = (1- tanh2(u(t)))()
ct dt

The time derivative of (A-2) can be rewritten as follows:

f (, q, 9') = 9 (172 + 7475 7173 + 76) 7172q |Itanh(y72) l2 (A 3)

+ 7173 q tanh(73i)| 2 7475 I |tanh(75j) |2 271 2qq tanh(72)

(1- ||tanh(72q) |2) + 2l7q q tanh(73q)(1 |tanh(73i) 12)

27475q tanh(75)(1- |Itanh(75q) |2).

Assuming as in Property 3.3 that q(t), q(t), and 4 (t) E Lm in D, it is straightfor-
ward from (A 1), (A-2), and (A 3) that f(q), f(q, ), and f(q, q, q) E R exist and
are bounded.









Lemma: The auxiliary error N(t) defined in (3-16) as

N(t) = N(t) Nd(t)

can be upper bounded as follows:

N (t) < p(|l|z|) 1 ||

where p(I z| ) is some positive globally invertible nondecreasing function.
Proof: Let N(t) E R' be defined as follows:

N(q, q, q e2,r) A M(q)r + e2
aq 2
Of (q) .. 1 r M+(q)
q qr +e2.
Oq 2 Oq

The unmeasurable q(t) can be expressed in terms of known and measurable values

eI, e2, and r as

q = qd r a1ce + (ai + a2)e2.

Expanding the definition of N(t), we have

A 19M(q)
N(q, q, qd, e,e2,r) O(q + e
2 aq
af(q') 9
+ (-- -r ae + (a + a)).

Recall the definition of Nd(t):

Nd(t) N(qd, Od, 0, 0, 0)
Of(qd) ..
qd .

The auxiliary error N(t) can then be written as the sum of errors pertaining to
each of its arguments as follows:










N(t) = N(q, q,qd, c, e2, r)

= N(q, qd, qd, 0, 0, 0)

+ N(q, q, d, 0, 0)

+ N(q, d, 0, 0, 0)

+ N(q, q,qd, e,0, 0)

+ N(q, q, qd, ei, e2, 0)

+ N(q, q, d, el, e2, r)


- N(qd, qd, d, 0, 0, 0)

- N(qd, d, qd, O, 0, 0)

N(q, gd, gd, 0, 0, 0)

N(q, q, g 0, 0)

- N(q, q, 0, 0)

- N(q, q, e,0, 0)

- N(q, q, ld, e1, 02, 0).


Applying the Mean Value Theorem to further describe N(t),

(t) = N(l, d, qd, 0, 0, 0) (- qd)
O(Ta
ON(q, 2, d,, 0, 0)
+ a72^2 (q qd)

8N(q, q, (T3, 0, 0,0) ..
+ ----3 a3 3 = -3d)

SON(q, q,,, 0a, 0 0)
+--4 4a v4 (ei 0)

SN(q, q, q, ,d a4, ( 0 0)
+aV (e2 0)

SN(q, q, qd, el, e2, )
+ 6 1.g ^(r 0)


where


Uv 6



U3

V4

V5

V6


(qd, q)

(d, )

qd

(0,ei)

(0,e2)

(0,r).


From equation (A-4), N(t) can be upper bounded as follows:


(A-4)









O(N11 N ( 9, aq d, 0, 0, 0)
IN(t)| < O
Bo-1


|aI I C111


+ ON(q, o2, 4d, 0, 0,0)
dQ2 '+ |2 |e2 -ae181
O(T2
+ ON(q, q, qd, a4, 0, 0)

S N (q, q 4, qd, (1, a5 0)1 II

+ ON(q, 4qd,, ,)2, (a) 2 e26
+ ^ |^ ^11


The partial derivatives of N as expressed in (A-5) can be expressed as follows:


ON(aT, qd, 4d, 0, 0)
0a1
ON(q, U02, d, 0, 0, 0)
0o-2
ON(q, d, d, 4, 0, 0)
06a4
ON(q, q4, qd, e1, a5, 0)
0a-5
ON(q, q 1, qd, e,2, a6)
0a6


(A-6)


O2f()..
qd

d9M
Of( )
OO (-a(

1+ (a + a2)
IO q). a/()
2 8q q


By noting that

2 = C2(q qd)

U4 = e1 C4(ei 0) = ei(l C4)

= (qd q)(1 C4)


VU = e2 C5(e2 0)


(ei + aiei)(1 cs)

(qd q + al(qd q))(1


e2(1 C5)


(A-9)


V6 = r c6(r 0) = r(1


c6) = r(1 c6)


q)))(1 -C6)


where ci E (0, 1) E R, i =2, 4, 5, 6 are unknown constants, we can use Property 3.2


(A-5)


(A-7)


(A-8)


(id q + al(qd q) + a2(qd q + al(qd









and Property 3.3 to upper bound (A-6) as follows:

ON(q, a2, d, 0, 0, 0)

ON(q, q, d, 4, 0, 0),
O0-4
&N(q, q, q el, ae, 0)
ON-(q,-4d),1,-(T5) 0) =v < ps(q,q)

ON(q, 4, q, e, e, s) 2 6) p6(q,qr)


where P2, P4, P5, P6 e IR are some positive functions nondecreasing in q(t) and q(t),

and where their arguments come from the variables of (A-6) and (A-7). Because

equations (3-2) and (3-3) give us the following expressions for q and q


q = qd el

4 = qd e2 + ale1


it is seen that the upper bounds can be rewritten

ON(q, 02, 4d, 0, 0, 0)
802 I72 v2
ON(q, q, qd, 0a4, 0, 0)
00 4
ON(q, q, q, e e si5, 0)

ON(q, b, oq, Nc u 2, ( s 6


The bound on N(t) can be further simplified:


in terms of e, e2, and r:


< P2(e1,e2)

< p4(ei,e2)

< P5(e61,2)

< p6(e,e2, r).


N(t) < p2(e,i,2) Ie2 a11 + 4(1i,e2) Ie11

+ P5(ei, e2) e21 + P6(e,2, ,r) )r

Using the upper bound


I |e2 1 I< |e2 + a Ie11








N(t) can be further upper bounded as follows:

N(t) < (aip2(m1,C2) +4(ei, 62)) l1 + (p2(e,,e2)

+ 5(ei,e )) Ie211 + p6(ei,e2, 2 r) I .

Using the definition of z(t) R3" in (3-21), N(t) can be expressed in terms of z(t)
as follows:

N(t) < (aip2(e1, 2) +p4(ee2)) .(t) + (p(ei, 2)

+p5(ei,e2)) I .(t) + p6(e,e2,r) I .(t)ll

< (alp(i, 62) + p4(e,e 2) + p2(e,e2) + p5(e,e2)

+ p6(e,e2,r))

< p( ll ) I 11

where p( Iz |) is some positive globally invertible nondecreasing function (see Xian
et al. [57]).


Lemma: Define the auxiliary function L(t) E IR as

L(t) A rT(Nd(t) psgn(e2)). (A-10)

Then, if P satisfies
1
S> Nd + -(Nd2 (A11)
a2
then

SL((T)dT < (b (A-12)

where (b is the following positive constant:


b = I2 (t) 2(to)TNd(t).








Proof: Integrating both sides of (A-10),


L(T)dT= fr(T)T(Nd(

Substituting (3-4) into (A-13),
t L()dT = OT Nd(7) +
J r T
T


7) psgn(e2(T)))dr.




a22 () TNd(r)


(A-13)




(A-14)


T) j3psgn(e2(7)) a2pe2(T)Tsgn(e2(7)))d
07
= a2 e2(T) T (Nd(T)- _3gn(C2((T)))dT

+ jd(T)d -sgn(C2(T))dT.
to 00

After integrating the second integral in (A-14) using integration by parts,

L(,F)d( T 2C2 ( T(Nd(T)- QSgn( 2())))dT

C Lrdt f() C2 -3gn sgn(e2()))dT

ato 07
+ II (t)l (I INd(t) + II (to)l C (t )Nd(t).
a= OT2 () sgn2e(7)) d7

+e2(t)TNd(t) e2(to)N(t 3 e(t) + 3 \e (to)

/< a1 62 N(N()(7 +- ( 3)dr

+ 2 ( Nd(t) -P) + P e2(to) e(to)TNd (to).


(A-15)


Thus, it is clear from (A-15) that if P satisfies (A-11), then (A-12) holds [57].














APPENDIX B
TRACKING CONTROL IN THE PRESENCE OF FRICTION BY HIGH GAIN
FEEDBACK AND A MODEL-BASED FEEDFORWARD COMPONENT-AN
EXTENSION

Lemma: The auxiliary error N(t) defined in (4-22) as

N(t) = N(t) Nd(t)

can be upper bounded as follows:

N(t) < p(lll) ll|

where p ( |. |) is some positive globally invertible nondecreasing function.

Proof: N(t) E IR' in (4-19) can be expressed as follows:


N (q, q,, ed, e2, r)


T M(q)
Yd F 9d


aQ (e2 all)
+a2C2


+ M(q)al[{r a22 a (e2 ae)} + a2 (r


a2e2)]


OM(q) .V. mV(q, q) 2 V,(q, q).
+ cdq + M(q)Vd + q2 + 1(4d
8q 8q 8q
- r + a2e2 + ae2 ae) + V(q, )(qd + a22

G(q) 8f(q)
+ ale2 aeiC) + +q+ (qd + a2e2
Oq aq
2 1 1M(q)
+ a1e2 aCle) YdO + e2 (q)r
2 Oq


where the following are used:


(B-1)


q = rd (e2 ale ),

q = 4d r + 02C2 + ale2


ale1.









Recall the definition of Nd(t):

Nd(t) N(qd, qd, 4d, 0,0, 0)
Sf (qd) 9M(qd) d (qd) V (qd, d) 2
= n d + 4 + Md + (qd)q d + -qd
dqd qqd d qd
&,Vm(qd,qjd) .dVq, d G(qd)
+ -qdqd + V,(qd, qd)qd + qd YdO.
+ dqd dqd.

The auxiliary error N(t) can be written as the sum of errors pertaining to each

of its arguments as follows:

N(t) = N (q, q, qd, 'd, el, e2,r) N (qd, d, 4d, qd, 0, 0, 0)

= N (q, gd, qd, qd, 0, 0, 0) N (qd, gd, qd, "d, 0, 0, 0)

+ N (q, q, qd, 0, 0, 0) N (q, d, 4d, qd, 0, 0, 0)

+ N (q, q, qd, 00, 0) N (q, q, d0, 0, 0)

+ N (q, q, qd, 00, 0) N (q, q, d0, 0, 0)

+ N (q, q, qd, e, 0, 0) N (q, q4d, qV, 0, 0, 0)

+ N (q, qd, q'V, e, 02,0) N (q,q, qd,, C1, 0, 0)

+ N (q, q, d, qd, 1e, e~, ) N (q, qd, qd e, 0,) .

Applying the Mean Value Theorem to further describe N(t),

9N(eT, ga, qd, 9'd, 0, 0, 0)
N(t) = N(a1,'djqdO O ([q qd) (B-2)

+N(q, U2, 4d, d, 0, 0, 0)
+ T C2=-v2 q qd)
Ba2
+ N(qq, a3, 'd, 0, 0, 0) .. .

+ N(q, qd, 4 0, 0, 0O') V4 .Vd)

+ ON(q, q, qd, (45,0, 0, 0) (- 0)
+ ----o4=r)4 [qs d q0d)


+ N(q, q,, q'd, C5, 6, ,0)
+ ----- =\ (e 2 0)
+ ON(q, qd, 'qd, eC, C,2, (
+ ---\ e7 0)
+ 1\=r| (r 0)
B~a7









where



v2 (d, q)

v3 (d, 4d)

v4 E ('d, 'd)

v5 E (0,e i)

ve e (, e2)

7 E (0, r).

From equation (B-2), N(t) can be upper bounded as follows:




+ a2------ -------- \7 =v2 e'2 ale1
+ ON(q, a2, 4d, Od, 0, 0, 0)1 -l1


+ON(q, q, qd, C "d, 0, 0, 0)

+ N(qq, q, qd, C 6, 06, 0)

+ N(q, q, q d, l, e,2,(7) 7V7 r


(B-3)


The partial derivatives of N as expressed in (B-3) can be expressed as follows:


N(TiV, gdo qd, 9d, 0, 0)


2Mf(ai) M(ai) ...
2 qltd + Id

+ v0(l d) 2 Vm (1T, Kd)
1+ ad 4
a(7 I08 q(7T


al


+ (&V (i ) d 2 &G( i).
+ 4d + 2 d
(70- (70


(B-4)











8M(q).. 2Vm(
S qd + -
02Vm (q, a2)
+ 2 (2qd +
80(2


q, (2) V2 +2 2(q,
20q 2 Oq


9Vmq, a2
g2 d
-- U2- d


OG(q)
Oq


Oao2
0a2


02f(2) ..
+ qd
O(T2


2 OM(q)
q


+ O ) + aM(q)
+ q


OVm(q, )
,9 q
q
oc)


+ Vm(q, qd)


OYdO


ON(q, q, qd, qd, el, a6, 0)
a976








ON(q, q, ld. qd, elC, e2, a-7)


9M(q)
a q (al + a2)
Oq
aoo
+ 1 + (a+ +
Oa6
+ vm(9, )
+ 9 ),




a (T7- YdFYd

vm(q,) (,
O q V(q,
Nq


- M(q) (aa2 + a + )

) ( ) + V(q )
02 +V(99


r + M(q) (ai + a2)

a 1 OM(q)
1 ar 2 Oq


By noting that


vi = q ci(q qd)

V2 = ( C2 (q d)

V5 = e (1 c5)

V6 = e2 (1 6)


V7 = r ( 7)


ON(q, 2d, qd, "d, 0, 0)
aO 2


f( )
9q


ON (q, q, qd, Vd), as, 0, 0)









where ci E (0, 1) IR, i = 1, 2, 5, 6, 7 are unknown constants, we can upper

bound (B-4) as follows:

ON(i, qd, 4d, d, 0, 0, 0)
-----10 1 pil-e Pl 1i)

ON(q, 2d, qd, Qd, O,0, 0) 2
O(Ta2
ON(q, q, d, (5' ,0, 0)
a05v5 P5(ei,e2)
ON(q, q, qd, 9 d, ae, 6, 0)
O c6 p 6(,T 6e 2
ON(q, q, qd, d, Ce, 2, oT) 7


The bound on N(t) can be further simplified:

N(t) < pi(ei) ||eI|| + P2(6, C2)e 2 a \11

+ p5(ei, e2) I1 e + p6(e, 62) 11e2

+ p7(ei, e2,r) Ir

Using the upper bound


I |e2 ale < lI|e21 +a 1 |ei| 1

N(t) can be further upper bounded as follows:

N(t) < (pi(ei) + a012(61,e2) + 5(i,e2)) e111

+ (P2(ei,e2) + p6(ei,e2)) Ie21 + p7( e,e2,r) |rI