<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2014-12-31.

DARK ITEM
Permanent Link: http://ufdc.ufl.edu/UFE0043776/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2014-12-31.
Physical Description: Book
Language: english
Creator: You, Kyusuk
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Kyusuk You.
Thesis: Thesis (M.E.)--University of Florida, 2011.
Local: Adviser: Burks, Thomas F.
Electronic Access: INACCESSIBLE UNTIL 2014-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043776:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043776/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2014-12-31.
Physical Description: Book
Language: english
Creator: You, Kyusuk
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Kyusuk You.
Thesis: Thesis (M.E.)--University of Florida, 2011.
Local: Adviser: Burks, Thomas F.
Electronic Access: INACCESSIBLE UNTIL 2014-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043776:00001


This item has the following downloads:


Full Text

PAGE 1

1 ADAPTABLE CATCHING SYSTEMS FOR CONTINUOUS CITRUS HARVESTING By KYU SUK YOU A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 201 1

PAGE 2

2 2011 K yu S uk Y ou

PAGE 3

3 To my family and advisor Dr. Burks

PAGE 4

4 ACKNOWLEDGMENTS I would like to give my sincere thanks to my advisor Dr. Thomas Burks for his It was a wonderful experience and I am deeply indebted for giving me opportunity to work with him. I would like to specially thank my supervisory committee members Dr. W. S. Lee and Dr. John Schueller for their support, technical insight and inputs I am very thankful to Mr. Gregory Pugh and Mr. Mike Zingaro for their constant support and guidance throughout my project. I would like to express my deepest appreciation to my wife. Her devotion and l ove made this project possible.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURE S ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 Citrus Industry in Florida ................................ ................................ ......................... 12 Mechanical Harvesting System in Citrus Industry ................................ ................... 12 Impo rtance of Developing Catching Systems ................................ ................... 13 Traditional harvesting machines ................................ ................................ 13 Traditional catching systems ................................ ................................ ...... 14 Present harvesting machines ................................ ................................ ..... 14 Present catching systems ................................ ................................ .......... 15 Motivation of Thesis ................................ ................................ ................................ 16 2 OBJECTIVES ................................ ................................ ................................ ......... 20 Brush Sealing Me thod around Tree Trunks ................................ ............................ 20 Specifying Catching Modes by Range Sensing ................................ ...................... 20 Synchronizing Moving of Catching Units ................................ ................................ 21 Skipping the Catching and Collecting ................................ ................................ ..... 21 3 LITERATURE REVIEW ................................ ................................ .......................... 22 Development of Low Profile Catching System ................................ ........................ 22 PLC Control ................................ ................................ ................................ ............ 24 4 EXPERIMENTAL M ETHOD ................................ ................................ .................... 26 Algorithms for Catching Operation ................................ ................................ .......... 26 System Modeling ................................ ................................ ................................ .... 27 Brush Catching Units ................................ ................................ ........................ 27 Lateral Moving Units ................................ ................................ ......................... 28 Base Frames ................................ ................................ ................................ .... 28 System Building ................................ ................................ ................................ ...... 29 PLC Programming ................................ ................................ ................................ .. 29 Preparations for programming ................................ ................................ .......... 30

PAGE 6

6 Case Programming ................................ ................................ .......................... 32 Static Catching Mode ................................ ................................ ....................... 33 Proximity Sensing based Catching Mode ................................ ......................... 34 Range Sensing Based Catching Mode ................................ ............................. 34 5 EXPERIMENTS ................................ ................................ ................................ ...... 52 Static Catching Test ................................ ................................ ................................ 52 Dynamic Test Based on Proximity Sensors ................................ ............................ 53 Dynamic Test Based on Range sensors ................................ ................................ 53 6 CONCLUSION ................................ ................................ ................................ ........ 60 APPENDIX: 3D PART MODELING AND DRAWING ................................ .................... 64 LIST OF REFERENCES ................................ ................................ ............................... 66 BIOGRAPHIC AL SKETCH ................................ ................................ ............................ 68

PAGE 7

7 LIST OF TABLES Table page 4 1 I/O a ddresses with a ssigned a ctuation ................................ ............................... 50 4 2 Calibrating d istance t ime d ata of a ctuators ................................ ........................ 51 5 1 Recovery r ate from 3 c atching m odes ................................ ................................ 59 5 2 Skipping r ate from 3 c atching m odes ................................ ................................ 59

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Drop loss on t he d riving p ath ................................ ................................ .............. 18 1 2 Trunk s haker ................................ ................................ ................................ ....... 18 1 3 Canvas r oll out c ollector ................................ ................................ ..................... 18 1 4 Inverted u mbrella t ype c atcher ................................ ................................ ........... 18 1 5 Two u nit c atching s ystem ................................ ................................ ................... 18 1 6 Self p ropelled c anopy s haker ................................ ................................ ............. 18 1 7 Pulled b ehind c anopy s haker ................................ ................................ .............. 19 1 8 Plate t ype c atching s ystems ................................ ................................ ............... 19 1 9 Oxbo u ltra l ow l oss c atcher ................................ ................................ ................ 19 1 10 Disc t ype c atching s ystem ................................ ................................ .................. 19 1 11 BEI Centipede s cale c atcher ................................ ................................ .............. 19 4 1 Output d ata of r ange s ensor for t ree p ositions ................................ .................... 38 4 2 Algorithm of p roximity s ensing b ased c atching o peration ................................ ... 38 4 3 Algorithm of r ange s ensing b ased c atching o peration ................................ ........ 39 4 4 Revolving b rush t ype c atching s yst em ................................ ............................... 39 4 5 Self r otating b rush t ype c atching s ystem ................................ ............................ 40 4 6 Strip b rush t ype c atching s ystem ................................ ................................ ........ 40 4 7 Components of a c atching s ystem ................................ ................................ ...... 41 4 8 Bracket p arts in BCU ................................ ................................ .......................... 41 4 9 Strip b rush b rooms ................................ ................................ ............................. 41 4 10 Subparts in LMU ................................ ................................ ................................ 42 4 11 Actuating s ystem c ontrolled by PLC ................................ ................................ ... 42 4 12 Time p lot for a ctuating d istance ................................ ................................ .......... 43

PAGE 9

9 4 13 Tilt a ctuating s ystem with 3 r evolute j oints and 1 p rismatic j oint ......................... 43 4 14 Time p lot for t ilting a ngle ................................ ................................ .................... 44 4 15 Initially p ositioning m ode ................................ ................................ .................... 44 4 16 Normal c atching m ode ................................ ................................ ........................ 45 4 17 Left m isaligned c atching m ode ................................ ................................ ........... 45 4 1 8 Right m isaligned c atching m ode ................................ ................................ ......... 45 4 1 9 Skip c atching m ode with o pening a ngle of 45 d egrees ................................ ....... 46 4 20 Location of p roximity s ensors ................................ ................................ ............. 46 4 21 Range s ensor o utput s ignals for d ifferent m aterials ................................ ............ 46 4 22 Algorithm of d ata f iltering p rocess ................................ ................................ ...... 47 4 23 Range after f iltering and s tepping ................................ ................................ ....... 48 4 24 Location and s ensing of r ear p roximity s ensor ................................ ................... 48 4 25 Tracking t ree during c atching and d etecting i mmature t ree ................................ 49 4 26 Tracking t ree during c atching for t he f irst t ree ................................ ................... 49 4 27 Modified s kip c atching m ode with o pening a ngle of 55 d egrees ......................... 49 5 1 Complementary t ree and c itrus s et ................................ ................................ ..... 55 5 2 Step s ignal from t he f ront and r ear p roximity s ensors ................................ ........ 55 5 3 o peration for t he c enter t ree ................................ ............... 56 5 4 o peration for t he c enter t ree ................................ ............... 56 5 5 o peration for t he l eft o ffset t ree ................................ .......... 57 5 6 o peration for t he r ight o ffset t ree ................................ ........ 57 5 7 o peration for s kipping c atching c itrus ................................ 58 6 1 V s pace among t ree b undles ................................ ................................ .............. 63

PAGE 10

10 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 Engineering ADAPTABLE CATCHING SYSTEMS FOR CONTINUOUS CITRUS HARVESTING By K yu S uk Y ou December 201 1 Chair: Thomas F. Burks Major: Agricultur al and Biological Engineering In order to continuously harvest citrus fruit using on going harvesting machines such as canopy shakers, the catching system should catch and collect fruits continuously. I n this thesis, three concepts were explored for a catching system appropriate for the canopy harvesting system. The first concept is a brush catching concept applied to low profile catching systems. S trip brush brooms were mounted at the end of collecting beds to make seals around tree trunk and along the narrow space between trees. T he strip bru sh brooms used in this project were originally manufactured for street sweeping vehicles. E ach brush could tilt from 30 degree s up to 45 degree s using an electrical linear actuator. T he second concept is to adjust catching unit to an irregular tree row using tree position measured by range sensor B rush catching units are move d laterally using actuators controlled by the programmable logic controller (PLC). As the tree approach es the front of catching system, its distance is measured by range sensor. An o utput signal from range sensor is filtered and modulated as continuous step signal. B ased on this data, brush catching units can move to accommodate tree misalignment within the row T he last concept is a kind of selective catching concept T his concept was not fully completed in this project. It required two

PAGE 11

11 procedures to execute ; 1) skip catching mode using actuators ; and 2) identifying sensing for immature or diseased trees. T he f irst required procedure could be solved by making enough space between brush catching units for the rejected tree to pass through without catching fruits A ctuating times for skipping mode were studied and preset to PLC. O therwise, the maturity of trees can be determined using a DGPS based tree fruit harvest selection map. A timer based method was used to order the trees for selective rejection of im mature trees in this study T he efficiency of this catching system was measured in terms of collecting rate and skipping rate under the simulated catching environment M ean percentage in catching tennis balls ( substituted for fruits ) was reported as 98.55 % while the effectiveness for skipping immature fruits was 68%.

PAGE 12

12 CHAPTER 1 INT RODUCTION Citrus Industry in Florida Today, there are more than 11,000 citrus growers cultivating almost 82 million citrus trees on more than 620,000 acres of land in Florida. Nearly 76,000 other people also work in the citrus industry or in related businesses. The state produces more oranges than any other region of the world, except Brazil, and leads the world in grapefruit producti on (Florida Citrus Industry Website 2010 ) All told, the citrus industry generates more than $9.3 billion in economic activity in Florida. As such, the citrus industry plays an important role in the life of every Floridian (Florida Citrus Industry Website 2010 ) Mechanical Harvesting System in Citrus Industry I ncreasing orange consumption and rising labor cost made inevitable the develop ment of mechanical harvesting system s Introducing mechan ical harvesting systems has contributed to the citrus industry not only to reduce produc tion cost but also to enhance harvesting efficiency However other problems are curren tly causing drop in yield. N ot only natural disasters like hurricane, cold snaps, etc. but also new diseases, decreasing yield and shortening g rove life Harvesting systems ha ve been developed in order to prevent or mitigate these factors by applying precision tech nologies In this section, the histories of catching systems developed for tree harvesting are described. In the 1950 s, citrus tree p runing studies led to the development of the first mechanical hedging machine built at the University of Florida C itrus R esearch and education center in Lake Alfred, FL ( CREC ) Now, mechanical hedging is a standard

PAGE 13

13 practice in Florida groves. Research on production practices has enabled growers to manage and maintain the high density plantings common in Florida groves. Importance of Developing Catching Systems M ost industrialized catching system were developed based on stop and go harvesting machines like trunk shakers creating uncovered space along the tree row. T h ese catching system s have a high loss reported up to 10% of total products D ropped fruits shown in Figure 1 1 are generally manual picked up or left on the ground. The recovery rate is impor tant to decide the product cost. T he ultimate objective of this study is to increase recovery rate. Traditional h arvesting m achines Four types of harvesting systems have been attempted for tradi tional citrus harvesting; Air blast, water cannon trunk shakers and canopy shakers A n air blower type harvester detaches fruits from branches by strong wind generated from air blower like a fan. Even though this system is very simple and economical it is difficult to catch and collect crops due to fruit traje ctories and bouncing T he water cannon harvester shoots water stream to detach fruits instead of air blasting. However there are problems to carry heavy capacity water tank during harvesting. A trunk shaking type system shown in Figure1 2 has been commonly used not only in the citrus industry but also in many other agriculture industries. Fruits c an be detached from its branch by shaking the tree trunk The early Canopy shakers were developed to harvest berries. T he harvesting unit consisted of two f inger s tudded vibrating panels. And the purpose of this design was to achieve maximum selectivity of berries. This concept provided a uniform shake to all parts of the cane.

PAGE 14

14 Traditional c atching s ystems M any kinds of catching frames had been developed for citrus harvesters. A t the beginning, harvesters picked fruit and just dropped citrus on the ground, and later labor picked fruit Due to rapidly increasing labor costs, development of portable catching and collecting systems ha s been demande d. O ne concept develo ped is canvas roll out frames shown in Figure 1 3 which spread both side s of a tree trunk for catching and collected crops. A nother type is an inverted umbrella type catching system shown in Figure 1 4 which wrap s its shields around the tree trunk becoming shaped like an inverted umbrella T he last concept is a low profile catching system using two unit s in tandem named a two unit systems shown in Figure1 5 T his catching system was developed for the trunk shaking harvesting machines and has been ve ry popular because of the outstanding efficiency of catching and collecting B y using elevating conveyors two unit systems could achieve a low profile, which boom shakers could reach over easily and which could reduce the interference with limbs. Present h arvesting m achines T oday, the most commonly used harvesting system is a canopy shaking system. In spite of the high harvesting efficien c y of the trunk shaking system with the two unit catching system it was less productive for harvesting time requiring more labor due to stop and go type harvesting process The canopy shaking systems are able to harvest citrus continuously which can reduce harvesting time and labor. T here are two kinds of canopy shaking systems commonly used in the citrus industry. T he f irst is a self propelled canopy shaking systems described in F igure 1 6 and another is a pulled behind canopy shaking system shown in Figure 1 7

PAGE 15

15 Present c atching s ystems T wo unit system s developed based on stop and go harvesting machines have negatives for use with canopy shaking system. These catching systems cannot cover space between trees along the tree row I n this case, the drop loss is recorded up to 10% from total harvested. Continuous catching systems have been developed by the commercial citrus industry in two ways. O ne is a plate type catching systems shown in Figure 1 8 which seals b etween canopy shakers by using overlapped plates mounted at the end of catching frames. This type catching system i s the first generation applied to continu ous harvesting systems O ne of compan ies using the plate type catching frames is Oxbo International Corporation. Recently they have developed the Ultra Low Loss catcher F i gure 1 9 and put it into the Korvan 8000 blueberry harvesting system. The ultra low loss catcher had 9.4% fruit drop rate and it reduced up to 50% of drop loss when compared to disc type catchers. (Oxbo International Corporation Website) However the flap type catching systems receives bigger impacts from citrus tree s and falling citrus than from blueberry trees When catching misaligned tree s the impact can occur not at the end of flaps but near the holding axle s of the flaps. Some flaps are broken when colliding with the tree trunks, or opposing flaps. A nother catching system is the centipede scale catching system sho w n in Figure 1 10 using self rotating discs and rectangular plates in tandem. B y the driving force, each side of disc sealers can be pushed and it returns to original position after passing of a tree I n fact, this catchi ng system has never been ap plied to citrus harvesting machines due to its design equitable to over the top harvesters B lueberry Equipment, Inc. developed and commercialize d the Centipede Scale System (CSS) shown in F i gure

PAGE 16

16 1 1 1 for blueberry and jatropha harvesting in 2008 T he centipede scale catching system use s number of disc s for sealing and collecting small crops like berries and jatropha. Two sealing unit consist s of a self rotating rubber disc and a rotating plate linked by joints. T he contact dama ge to tree trunks during catching can be reduced by using rubber material for rotating disc and the self rotating motion of disc In addition, a spring mounted at the end of rotating plate absorb s shock occurred when the disc met the trunk. The catching ef ficiency of the CSS is extraordinary; the dropped rate in total yield is under 18% for blueberry ( B lueberry Equipment, Inc Website) However, it is unclear how well this system will perform in citrus d ue to higher vertical load from falling citrus and lateral load from bigger offset of citrus trees in the tree row. D isc and plates are custom built caus ing potential high part cost and maintenance cost. Motivation of Thesis C urrent catching systems have been reviewed and their weaknesses applying to continuous harvesting system are listed in following: 1. Passive sealing operation around tree trunk 2. H ard and costly maintenance 3. U niform ed catching pattern ignoring tree positions 4. Lack of catching solution for immature trees N ew catching system is required incorporating features to improve current systems ; 1. F lexible material sealing units 2. Enough strength to hold dropped citrus 3. E asy to fix and maintain, cheap sealing material 4. Real time adjustable catching systems according to tree positions 5. E nable to avoid ca tching citrus for immature trees

PAGE 17

17 Brush materials are chosen for sealing units in order to improve the first and second features. S trip brush brooms used in street sweeping vehicles can return to its original position after passing trunk. A lternatively, com mercialized strip brush brooms in the sweeping industrial market can be easily and cost effectively replaced I n order to solve the third and fourth features, the catching operations ha ve to be controlled according to tree position and maturity

PAGE 18

18 Figure 1 1 Drop loss on T he Driving Path Figure 1 2 Trunk Shaker Figure 1 3 Canvas Roll o ut Collector Figure 1 4 Inverted Umbrella Type Catcher Figure 1 5 Two Unit Catching S ystem Figure 1 6 Self Propelled Canopy Shaker Photos courtesy of Citrus Research and Education Center

PAGE 19

19 Figure 1 7 Pulled Behind Canopy Sha k er Figure 1 8 Plate T ype C atching S ystems Figure 1 9 Oxbo Ultra Low Loss Catcher Figure 1 10 Disc Type Catching System Figure 1 11 BEI Centipede Scale Catcher Photos courtesy of Kyusuk You

PAGE 20

20 CHAPTER 2 OBJECTIVES The ultimate goal of this study is to develop new type s of catching systems to reduce yield loss from dropping through uncovered space among irregularly aligned trees and to skip trees selectively when marked by maturity or disease T he specific objectives are following : 1. To develop a method to catch citrus around tree trunks 2. T o adapt catching modes by sensing displacement of trees from center of row 3. T o develop a method to adj ust catching units by the catching mode 4. To skip catching f o r immature or diseased trees Brush Sealing Method around Tree Trunks T he first objective will be accomplished by mounting strip brush brooms at the end of each catching unit s. I n order to simulate sealing continuously around tree trunks while moving it will be necessary to have contact the seal and trunk B rush materials will be applied to the catching system not only to reduc e damages to trunk during catching but also to capture the dropp ed citrus on the frames It is expected that brush materials will be eas ier and cheap er to repair and replace than any other commercialized sealing materials. Specifying Catching Modes by Range Sensing A feedback loop model using sensors and transducers wi ll be applied to the system to control the catching frames U nder limited experimental conditions only IR proximity and range sensors are used in this project. Using feedback from the transducers, t h e second objective was solved by processi ng signals from the range

PAGE 21

21 sensor which measured tree position T his information was then used to shift the catch from the appropriate position to catch in incoming tree fruits. Synchronizing Moving of Catching Units Synchronizing catching frame motions wil l be accomplished by pairing individual catching frames, and controlling their motion in tandem Four linear electrical actuators will be mounted under four c ollecting beds; two at front, two at rear. T he front pair will be controlled in a synchronized man ner. T hrough entire text, Between Frame Gap (BFG) is defined as the long and narrow space between both sides of brushes T he lateral moving units (LMU) define the paired motion of the catching frames in a lateral actuator control approach. T he n the BFG c an be adjust ed to the current tree position by retracting or ext ending actuators. LMU motion is controlled by PLC using acquired data from sensors. Skipping the Catching and Collecting T he catching system should avoid collecting fruit from diseased or immat ure trees to eliminate bad fruit and protect young tree trunks. T o achieve this goal simultaneous activity will control the system First, t he catching system should make BFG as wide as possible by moving collecting bed backward by controlling LMU Additional actuators are mounted o n the catching system which can tilt upward the catching brushes to give more free space to avoid contact between brushes and the immature trunk.

PAGE 22

22 CHAPTER 3 LITERATURE REVIEW T his section of the thesis discusses t he studies done on development of low profile catching system, filtering of sensed data and actuating system using PLC. D evelopment of Low Profile Catching System L ow profile fruit catching is commonly applied to fruit harvest ers to reduce labor costs of fruit pick up and improve efficiency of harvesting. H arvesting systems are uniquely developed for each specific species of fruit. Most of citrus catching system s do not have adequate seal between the independent sides of harvesting par resulting in pro duct loss occurred here cannot be ignored. Coppock (1967) applied a fruit conveyor to both sides of catching frame, and a crossover conveyor was equipped at end of catching frame. Mentioning that an effective fruit seal must be formed between the frames at the base of tree, he built catching frames symmetrically about tree row line with conveyor on each side. Applying crossover conveyors to sealing materials this catching system achieved a recovery rate of 97.2% for Hamlin orange. E ven though this result is outstanding, this catching system was developed suitable for trunk shakers. Besides, using conveyors to seal the catch ing frames does not insure a make seal around tree trunk, particularly on continuous harvesting. Churchill (1980) designed a drip line pickup system. The p urposes for developing the system were to increase fruit handling capacity and the efficiency of trash and fruit separation ; and to reduce damage using trash eliminator in pickup conveyors. T he system was designed focused on picking up fruits fallen on the ground not on direct

PAGE 23

23 catching of frui t s Recovery efficiency of 98% was reported on citrus harvesting with 1.36km/h (0.85mi/h) driving speed. Groz et al. (1989) designed a canvas roll out catching system for citr us operated by means of self propelled hydraulic powered carriage using a telescopic trunk shaker, a roll out fruit collector, and a fruit handling system. Unlike other canvas roll out catching systems designed suitable to collect deciduous fruits like app les, this canvas roll out catching system was developed for citrus fruit which ha s dense foliage and high hanging branches loaded with fruits. Collecting rate of 90% and average harvesting capacity of 35 trees/hr were obtained while testing under differen t grove conditions and citrus varieties. Peterson (1997) applied a collecting conveyor to catch falling citrus on a continuous mechanical harvesting system. Double spike drum canopy shakers were used for removing oranges W ith a conveyor as long as the harvesters collect ing falling oranges. Oranges dropped on collecting conveyors could be gathered in storage bins by using the cross conveyor located at rear of the catching system. While this harvesting system reported removal rate of 95% recovery ra te of the removed fruits averaged only 73% Development of an over the row harvesting machines for berries led to the advent of high density catching system s using variable catching materials. T he catching system studied in this thesis is also designed fo cusing on conceived over the row citrus harvesting system. According to the annual research progress report of IFAS Citrus Initiative (200 6 0 7 ), problems in developing the two unit catching system for continuous citrus harvesting were driving synchronizati on of left and right vehicle s and vertical

PAGE 24

24 misalignment of catching frame seals due to grove terrain. T o solve these problems, it was considered to change design concept of catching system from two unit type to over the row type in annual research progress report (2009 10) BEI International Co. introduced Centipede Scale System (2008) applying self rotating disc covered with rubber materials to reduce falling damage. A rotating disc is join ed to the collecting conveyor by a revolute plate. T his catching s ystem was designed for over the ro w type harvest ers The field test using BEI 3000 model with centipede scale system reported drop rate of 18% Ultra Low Less catchers developed by Oxbo International Co reduced drop rate by up to 9.4% using a plate type catching units (2010) C atching plates are arranged with 120 mm of catching space which makes high density catching fruits on blueberry harvesting. S ince it was also developed for the over the ro w type harvest ers it is applied to small fruits such as blueberry and jatropha. PLC Control A programmable logic controller (PLC) is essentially a digital like system typically used in factories and process plants, to connect input device such as switches to output devices such as actuators. T o control moving and tilting components of the catching system, Micrologix 1200 PLC was used in this study due to its ability to properly sequenc e complex task, consisting of many discrete operations and involving several devices which need to be carried out in a sequential manner. Reviewing other studies using PLC interfaced with variable devices will help design the working scheme of the catching system. Thomas et al (2003) construct ed a mobile supplement cattle feeder for animals using an Aromat model FP1 C40C. Seven sensors for identifying animals were

PAGE 25

25 connected to input ports of PLC, and 7 solid state relays connected to motors and solenoid v alves were wired to input ports. W hen an animal comes to the supplement feeder, PLC identifies this animal sensing R FID tag on the right ear. A s the animal approaches the feed bowl, IR sensors sense range to its head. A ctuating solenoid valves, each supple ment is mixed in the bowl with proper rates to the animal. PLC transfers amount of rest to the main computer. Schmilovitch et al. (1995) developed a device sorting dry dates on the basis of firmness. According to data from a sensing system measuring its f irmness the system classifies the date into four group s using air blowing triggers. Based on input data from t he sensing system consisting of feeding actuators and a linear encoder, PLC (Sysmac C40P model) controlled output operations of the classifying s ystem composed with air discharger and a step motor. So et al. (2010) analyzed design considerations for the mechanical harvesting of black raspberries and manufactured a prototype over the row mechanical harvester operating with a run stop shaking procedure. A nalyzing the presence of tree in front of the system using a photoelectric sensor, PLC (LG, K80S series) controlled hydraulic motors for harvesting black raspberries and hydraulic cylinder for moving collecting plates. A s the sensor detects tree in front of the harvester, shakers starts to operate and collecting plates come to center automatically. A fter harvesting, shakers and catching plate go back to initial position. T his operating mecha nism is similar with the operating logic used in this thesis in positioning catching frame s automatically according to sensed data. However, their s haker and collecting plates operated on a discontinuous harvesting procedure while the one proposed in this report is continuous.

PAGE 26

26 CHAPTER 4 EXPERIMENTAL METHOD This chapter presents the experimental approach of the catching system and illustrates the conceived design using a 3 dimensional CAD program, Solidworks ( 2008 ) which simplifies building the prototype T he prototype consists of the mechanical components, eight linear electrical actuators and a PLC to simulate catching operations. Tennis balls are used as a substitute for real citrus in the experiments. Three operating modes measure recover y rates for mature fruits and the skipping rates for immature fruits: static operation mode, proximity sensing operation mode, and combined sensing operation mode. The static operation mode catches falling citrus in stop catch go order and its operation i s controlled manually. The proximity sensing mode simulates an automatic catching system controlled by PLC using sensor input to predict trunk position. The combined sensing operation mode uses a range sensor and two proximity sensors together to control the catching system by tree positions. Algorithms for Catching Operation Catching operations are classified according to lateral positions and matur ity of tree; center located, left misaligned right misaligned, and immature. In the combined sensing ope ration mode, range sensing will determine the position of tree as a measurement of a tree displacement from a center row. Figure 4 1 shows output voltages from the range sensor for each position of tree. Figure 4 2 presents a flow chart describing the proximity sensing operation. Two count up operators are applied in a PLC ladder logic in order to discriminate tree positions: center located tree if n 1 =1, left misaligned tree if n 1 =2, right misaligned tree if n 1 =3, and immature tree if n 1 =4.

PAGE 27

27 CTU counts one up whenever a tree passes sensing zone. C TU#1 is mounted in front of the catching system to detect presence of tree and CTU#2 is mounted between front and rear catching units to track tree during catching. I n the combined sensing operation mode, tree positions are determined by comparing output value acquired from the range sensor shown in Figure 4 3. Immature tree is identifie d using proximity data n 1 =4. System Modeling Three concepts of the catching system using brush materials were considered. The first concept was a revolving brush type catching system shown in Figure 4 4. T he speed of revolving brushes is synchronized with the driving speed of catching system which can reduce damage from contacts between brush materials and tree trunks. However, fruits on the left and right corner of the revolving brush are expected to drop due to enlarged space among brushes. T he second concept was a sweeping brush type catching s ystem shown in Figure 4 5. Rotating brush core in the catching system catches fruits and then moves them backward. However, overall diameter of brush core exceeds the maximum clearance of the low profile catching zone. S weeping brushes were also applied in the last concepts U nlike the second concept, sweeping brush system consist s o f two front and two rear sub catching frames shown in Figure 4 6. E ach catching frame has 2 units, brush catching unit (BCU) and lateral moving unit (LMU) shown in Figure 4 7. Base frames guides was the lateral movement of the catching system and supports directional loads from tree. I t fixed on the ground for this experiment Brush C atching U nits A major concern in designing the brush catching units is to make a n appropr iate bracket to hold the strip brush. T his bracket should be designed not only to be easy to

PAGE 28

28 replace the strip brush but also to support the brush which receive s vertical loads from falling fruits and lateral loads from contacting trunks. B oth sides of a bracket open for inserting the strip brush shown in Figure 4 8. T he inserted strip brush is fixed in bracket by using bolts. The s trip brush us ed in this study is a commercialized product named, R unway S weeping C assette S ection manufactured by the United Rotary Brush Company shown in Figure 4 9. T his strip brush is composed with sixteen bundles of polypropylene filaments which have 300 mm of length and 25 mm of diameter The brush and bracket sets are mounted to the lateral moving unit by two rotating hinges which allow the brush assembly to tilt from 20 degree s to 60 degree s I n case of mature trees, 20 degree s tilted strip brushes encourage fallen fruits to roll to elevating conveyor s and 60 degree s tilted strip brushes avoid catching immat ure fruits. T he dimension of bracket is 580 mm ( L ) x70 mm ( H ) x155 mm ( W). The length of the bracket determine s the length of a strip brush. Lateral M oving U nits In order to adjust BFG according to the tree position, LMU should change its moving direction rapidly and smoothly. F our guiding rollers are mounted under both sides of catching bed shown in Figure 4 10. T his guiding roller reduces friction between LMU and base frame during i ts moving. In addition, it supports the directional load from BCU. A linear actuator mounted underneath the catching bed moves catching units laterally and LMU has an elevating conveyor to carry fruits away from BFG T he dimension of a catching bed is 5 80 mm ( L ) x 130 mm ( H ) x1 75 mm ( W). Base F rames The b ase frame assembly supports the movement of front and rear catching units. The b ase frame assembly consists of four base frames in which each frame has

PAGE 29

29 dimension with 580 mm ( L ) x 125 mm ( H ) x 880 mm ( W). Lateral moving actuators are connected at the center of outer beams in base frames. I beams located on both sides of the base frame guide s lateral mov ement of catching units shown in Figure 4 9. System Building The c atching system discussed above ha s been created using Solidworks 2008 for each part and assemblies. The prototype catching system wa s built based on these drawings but it has some modifications due to limited experimental conditions; declined collecting boards replaced the elevating co nveyors and the position s of the tilting actuators were moved E levating conveyors are already available through commercial suppliers Since this study i s interested only in catching the fruits rather than transporting, the conveyors were omitted and replaced with co llecting boards. Tilting actuators were relocated from underneath BCU and bottom of LMU to avoid disturbance of falling fruits. PLC Programming Control actuators and sensors consist of Allen Bradley 1200 PLC unit six points isolated h igh current output module (1762 OX6I) and four channels analog I/O expansion module (1762 IF2OF2). The ladder logic for catching operations is designed by using RS Logix program T wo output ports are required to extend and to retract a n actuator T here a re two controllers each controlling operations of four actuators using the 16 output ports of the PLC. T erminals are connected to I/O ports of PLC using 10 pin network connectors. T able 4 1 shows the I/O addresses with their assigned actuations and Figure 4 11 presents operating flow between actuators and PLC units.

PAGE 30

30 Preparations for programming All actuators in the prototype model operate based on open loop actuating time since the prototype does no t have transducers to detect tilting angles of BCU and moving distance of LMU. T herefore, data related to actuating time is required to program ladder logic Relationship between Actuating Time and Moving Distance T he moving distance of an actuator respon ds linearly with actuating time but actuating speed varies depending on applied load. In order to obtain correct operating time and moving distance, the actuator should be calibrate d T able 4 2 shows the actuating time versus distance data which is average value from three independently repeated experiments. Based on this data, the relationship between actuating time and moving distance of LMU shown in Figure 4 12 can be described with the first order linear equation like below: (4 1) W here x is actuating time and f(x) is moving distance of linear actuators. Relationship between Actuating Time and Tilting A ngle The tilting system consists of three revolute joints and one prismatic joint shown in Figure 4 13. Relationship between actuating distance and tilting angle is described with an inverse trigonometrical function. (4 2) w here is tilting radian angle of BCU from the base frame and is an o uter tilting angle which is equivalent to the sum of two revolute joint angles from the actuator. (4 3)

PAGE 31

31 (4 4) and are calculated using the arctangent function and are constant due to the fixed value of which are distances among five points in Figure 4 13. (4 5) (4 6) (4 7) (4 8) Inserting constant into Equation 4 8, relationship between actuating distance and tilting angle follows as : (4 9) Applying Equation 4 1, distance is transformed to time variable like below. (4 10) Finally, the relationship between actuating time and tilting angle of BCU from the equation (4 10) is described in Figure 4 14. D riving Speed of Tree Row Driving speed of tree row should be decided according to the maximum actuating distance in the catching system. W hen LMU moves from catching position of right misaligned tree to position of left misaligned tree, the actuators should move 140 mm

PAGE 32

32 for 1200 0 ms which is the maximum actuating length in this study. T herefore, driving speed of tree row is as low as 12 mm/sec. Tree row is pulled manually to simulate the operation of the canopy shakers. Case Programming PLC program was divided into four catching operation cases according to tree positions and maturity: center located left misaligned, right misaligned and immature tree T aking into consideration the delayed response times between lateral moving actuators, 150 0 ms was added to each actuat ing time of LMU. The modified equation 4 1 is following: Initial Preparation case I n t he initial preparation step, all of BCUs tilt down to zero degree and all of LMUs move fully backward Lateral moving actuators retract for 11 00 0 ms and b rush tilting actuators tilt down for 600 0 ms using timer on delay operators T4:3 and T4:4 controlled by a toggle switch I:0/3 shown in Figure 4 15. Center Catching case In case of t he center catching, LMUs c ome to center and then extends actuator to 70 mm for 65 0 0 ms and BCUs tilt 20 degree s up by extending for 330 0 ms Timer operators T4:6 and T4:7 control tilting angle of BCUs and T4:8 and T4:9 control moving distance of LMUs. Rear catching units start operation 50 0 ms after finishing the catching operation of front units usin g T4:5 shown in Figure 4 16 Left and Right M isaligned Catching Case

PAGE 33

33 T hese catching cases are for 7 0 mm m isaligned tree s which range depends on the maximum actuating distance of the linear actuator W hen the front proximity sensor detects the left misaligne d tree, T4:11 and T4:12 let BFG move to left side for 700 0 m s without tilting BCU shown in Figure 4 17 T here is a 100 0 ms interval between front and rear LMU moving by T4:10 ( preset with 800 0 ms ) A s the front proximity sensor detects the right misaligned tree, the front LMU set moves right side for 1200 0 ms using timer on delay operators T4:14, T4:15 shown in Figure 4 18. BFG move s 140 mm to right. This moving requires the longest actuating time of 1200 0 ms whi ch dominate s the driving speed of tree row as mentioned above. Open Catching Case Since the identifying sensing system is not applied in this prototype catching system, the method to avoid catching immature fruits is only explained. The catching system wi ll recognize the immature tree from a counting up operator connected to the front proximity sensor and preset with order of an immature tree in the tree row. W hen the counting up operator accumulates four, front BCU set starts to tilt 20 degree s up with ex tending actuators for 280 0 ms and front LMU set moves back to 150 mm while retracting actuators for 1100 0 ms. Beyond 800 0 ms from starting operation of front catching units, rear BCUs and LMUs start to actuate with same actuating time of front BCUs and LMU s shown in Figure 4 19. Static Catching Mode In static catching mode the tree trunk stops between front and rear catching units to catch and collect the falling citrus. A fter finishing catch the first tree moves out of catching units and the next tree trunk moves in. Manually operated catching is used to

PAGE 34

34 evaluate the efficiency of catching citrus using the brush catching system and it s result s are compared with results from the two sensor based catching modes. Proximity Sensing based Catching Mode Th is catching system operates using signals from two proximity sensors attached to in the catching system shown in Figure 4 20. Front proximity sensor is mount ed 850 mm at the front of catching system using an adjustable quadratic aluminum shafts. T he quadr atic shaft can fix the front proximity sensor out of tilting. An additional sensor is set up between front and rear catching units. R eflectors are located in corresponding side of each sensor. When the front proximity sensor detects tree, the catching sys tem starts to align BFG to center and tilts BCUs up by extending actuators Using the counting up operators, the catching system is moved according to the following order of trees: center located, left misaligned, right misaligned and immature. A nother counting up operator is applied to track front tree during catching. P roximity sensors are connected to 2 input signal ports I :0 /1 and I:0/2. Catching operation for the tree sensed by the front proximity sensor should start after finishing the prev ious catching operation. Since the length among two trees is 175 mm shorter than the distance from the front proximity sensor to the end of the first brush catching unit 1,315 mm. As shown in Figure 4 15, Figure 4 16, Figure 4 17, and Figure 4 18, timer on delaying operators are controlled by count up operators C5:2, C5:4, and C5:6 which are connected to the rear proximity sensor. Using a toggle switch wired to an input port I:0/13 in PLC, all of count up operators can be reset manually. Range Sensing B ased Catching Mode The catching system using a range sensor operates along with the tree position. Range sensor detects a tree and measures distance to the tree trunk. Based on th is

PAGE 35

35 distance data, LMUs move forward and backward and BCUs tilt up and down automa tically. IR range sensor GP2D12F manufactured by Sharp Electronics Co. has 39 ms response time and 5.0V as the maximum output voltage under operating voltage with 5.2V. It can sense range efficiently up to 800 mm. T he range sensor equipped to an input por t I:2/0 of PLC has different responses with various materials and shapes of objects. Figure 4 21 shows different output voltages depending on various types of materials. In addition, the output response for the PVC has a sinusoidal signal which makes it d ifficult to classify output ranges for tree positions. Besides, as the output response has more noise in sensing of dynamic objects than the static one, it is necessary to modify output signals in order to decide tree position using range data. Filtering and Modulat ing Due to heavy noise and sinusoidal response in output signals from the range sensor, the original data should be filtered and modulated. Tens and units of output data can be removed using a simple removal filter shown in 4 22. T he output data is divided by 100 and then using an integer operator, the decimal places are removed. In addition, multiplying 100 again to the data in the integer operator creates simple filtered values. The range sensor is mounted at 15 mm front of the front proximity sensor in order to correlate the sinusoidal signals to step signals as shown in Figure 4 23. W henever the front proximity sensor detects a tree the range sensor operates for 50 0 ms using time on delaying operator and detects range from tre e trunk face to the sensor. Even if the range sensor turns off after 50 0 ms, the output signal maintains the last sensed value until detecting next tree.

PAGE 36

36 Comparing and Determining Using th e filtered step signal shown in Figure 4 23, the output voltage ranges for tree position are classified as following: Right misaligned tree: D 4800 mV Center located tree: 3800mV D <4100mV Left misaligned tree: 2500mV D <3200mV Deviations in output voltage from sensing equally located trees are up to 12.3% due to the self vibrations of tree row during its moving This explains why the first and the last step signal are different as shown in Figure 4 23, for output values for center located tr ees. Since the range sensor is more sensitive to closer objectives, the output voltage becomes larg er as the tree approaches the range sensor. Detecting tree on catching In order to keep BFG centered on the tree position, the range sensor detects each tre e and tracks its location while in the catching system. Start of the catching operation in the rear unit for next tree is initiat ed by a proximity sensor mounte d between the front and rear catching units as shown in Figure 4 24. While three count up opera tors are used for detecting tree in the proximity sensing based catching mode, only one c ount up operator is used in range sensor catching mode. Figure 4 25 shows that a count up operator C5:1 connected to the rear proximity sensor is turned on for 30 0 0 ms using timer on delaying operator T4:1. C5:1 is reset after 300 0 ms. However, this algorithm does not apply to the first tree, since there is not a tree in front of it W hen the front proximity sensor detects the first tree in row, C5:1 get 1 count from the binary operator

PAGE 37

37 B3:3 from another count up operator C5:0 which is connected to the front proximity sensor as shown in Figure 4 26. Skip catching for an immature tree In order to create a larger space in BFG the tilting angle of BCUs is raised to 55 de gree s using 400 0 ms of tilting time as shown in Figure 4 27. T he retracting times of LMUs are same as used in other modes. Tree maturity cannot be determined at this time. Therefore, the same method is used as in the proximity sensing based catching mode

PAGE 38

38 Figure 4 1 Output Data of Range S ensor for Tree Positions Figure 4 2 Algorithm of Proximity Sens ing Based Catching Operation

PAGE 39

39 Figure 4 3 Algorithm of Range Sens ing Based Catching Operation Figure 4 4 Revolving Brush Type Catching System

PAGE 40

40 Figure 4 5 Self Rotating Brush Type Catching System Figure 4 6 Strip Brush Type Catching System

PAGE 41

41 Figure 4 7 Components of A Catching System Figure 4 8 Bracket P arts in BCU Figure 4 9 Strip Brush Brooms

PAGE 42

42 Figure 4 10 Subparts in LMU Figure 4 11 Actuating System C ontrolled by PLC

PAGE 43

43 Figure 4 12 Time P lot for A ctuating D istance Figure 4 13 Tilt Actuating System with 3 Revolute Joints and 1 Prismatic Joint 0 1 2 3 4 5 6 7 8 9 10 11 12 Actuating Time (sec) Actuating Distance (mm) Mean Time for expanding Meam Time for Retracting

PAGE 44

44 Figure 4 14 Time Plot for Tilting Angle Figure 4 15 Initially Positioning Mode

PAGE 45

45 Figure 4 1 6 Normal Catching Mode Figure 4 1 7 Left Misaligned Catching Mode Figure 4 1 8 Right Misaligned Catching Mode

PAGE 46

46 Figure 4 1 9 Skip Catching Mode with Opening A ngle of 45 D egree s Figure 4 20 Location of Proximity S ensors Photos Courtesy of Kyusuk You Figure 4 21 Range Sensor Output Signals for Different Materials

PAGE 47

47 Figure 4 22 Algorithm of Data F iltering Process

PAGE 48

48 Figure 4 23 Range after Filtering and Stepping Figure 4 24 Location and Sensing of Rear Proximity Sensor Photo Courtesy of Kyusuk You

PAGE 49

49 Figure 4 25 Tracking Tree during Catching and Detecting Immature Tree Figure 4 26 Tracking Tree during Catching for The First Tree Figure 4 27 Modified Skip Catching Mode with O pening A ngle of 55 D egree s

PAGE 50

50 Table 4 1 I/O a ddresses with a ssigned a ctuation Address Assigned Operation Description Location I:0/0 Turn on System Toggle Switch I :0/1 Front Proximity Sens or Sensing Tree before Catching 835 mm front of System I :0/2 Rear Proximity Sens or Sensing Tree on Catching Between Front and Rear LMU I :0/3 Pause System Temporary Stop System I n PLC I:0/13 Reset All Counter Toggle Switch In PLC I:2/0 Range Sensor Sensing Range to Tree 850 mm front of System O:0/0 Extending Actuator #1 BCU #1 Tilting Up Front Right O:0/1 Retracting Actuator #1 BCU #1 Tilting Down O:0/2 Extending Actuator #3 BCU #3 Tilting Up Front Left O:0/3 Retracting Actuator #3 BCU #3 Tilting Down O:0/4 Extending Actuator #4 BCU #4 Tilting Up Rear Left O:0/5 Retracting Actuator #4 BCU #4 Tilting Down O:0/6 Extending Actuator #7 LMU #3 Tilting Up Front Left O:0/7 Retracting Actuator #7 LMU #3 Tilting Down O:0/8 Extending Actuator #8 LMU #4 Tilting Up Rear Left O:0/9 Retracting Actuator #8 LMU #4 Tilting Down O:1/0 Extending Actuator #2 BCU #2 Tilting Up Rear Right O:1/1 Retracting Actuator #2 BCU #2 Tilting Down O:1/2 Extending Actuator #5 LMU #1 Tilting Up Front Right O:1/3 Retracting Actuator #5 LMU #1 Tilting Down O:1/4 Extending Actuator #6 LMU #2 Tilting Up Rear Right O:1/5 Retracting Actuator #6 LMU #2 Tilting Down

PAGE 51

51 Table 4 2 C alibrating d istance t ime d ata of a ctuators Distance (mm) Extending Time (sec) Retracting Time (sec) 0 0.00 0.00 10 0.88 0.92 20 1.42 1.62 30 2.07 2.32 40 2.84 3.08 50 3.49 3.80 60 4.22 4.45 70 4.99 5.14 80 5.60 5.86 90 6.32 6.56 100 7.04 7.31 110 7.71 8.04 120 8.42 8.74 130 9.11 9.44 140 9.83 10.12 150 10.53 10.73 1. A ll values are averaged from three times of measurement 2. There is ignorable difference in responses between extension and retraction

PAGE 52

52 CHAPTER 5 EXPERIMEN TS T he u ltimate objectives of this study are to seal around tree trunks using strip brushes, to align BFG for tree positions and to catch citrus selectively according to maturities. The experiment focuses on these objectives and shows the feasibility of the brush catching system for its operation on real field. The recovery rate and the skipping rate are evaluated for a tree row including 4 complementary trees with 1140 mm intervals between trees. T he diameter of the tree trunk was 90 mm. E ach tree has 6 canopies and 25 citrus which are simulated by tennis balls. A t otal of 100 balls we re observed in the experiment. Of the 100 balls, 25 balls are marked with X to identify immature citrus shown in Figure 5 1. T rees are arranged in center o f the machine 70 mm left misaligned and 70 mm right misaligned. Experiments are independently repeated three times for each catching mode. The first experiment evaluates the recovery rate and skipping rate for the catching system in static catching mode The last two experiments are performed with proximity sensing and range sensing. T he Micrologix 1200 model is used for the main control system with an 1762 IF2OF2 and 1762 OX6I expanded I/O. Eight electrical actuators controlled by two motor controllers actuate the LMUs and tilt BCUs in the catching system. Static Catching Test The recovery rate and skipping rate are evaluated by manually operating the catching system without PLC control. In these experiments strip brushes provided a high catching efficiency. M ature citrus were caught by the catching system as shown in Table 5 2. F or immature tree s 19 out of 25 immature citrus fell on the ground resulting in a skip rate of 76%. These results are average values from 3 repeated experim ents.

PAGE 53

53 Dynamic Test Based on Proximity Sensors Using two proximity sensors shown in Figure 5 2, the recovery rate and skipping rate are measured. When the front proximity sensor detects a tree, the catching system starts to operate. A s mentioned in previo us chapter, the catching system operates according to preset catching order; center left offset right offset open catching. Using count up operators connected to proximity sensors, PLC recognize s tree posi tion sensed and determine s catching sequence W hen accumulated order of the count up operator connected to the front proximity sensor is one LMUs move to center and BCUs tilt 25 degree s up. A s the accumulated order is two and the rear proximity sensor detects previous tree, the left misaligned catching st arts to operate. T he right misaligned catching starts to operate similar to the left offset catching mode. Seventy out of 73 dropped citrus were caught by the catching system resulting in a recovery rate of 95.89%. When the accumulated trunk count becomes four, the open catching case begins to operate I n this mode, 7 of 25 immature citrus fell on the ground resulting in a skip rate of 28%. T he a bove results are average data from 3 repeated experiments under same conditions. T his experiment s skip rate was the lowest rates among the three catching modes as shown in Table 5 2. T his was due to the limited BCU opening of 45 degree s while the range sensing based test was 55 degree s T his resulted in a difference of 88.12 mm in width of the opened BFG b etween the two tests. Dynamic Test Based on Range sensors T he aligning of the BFG can be observed from the I/O signal generated from each actuator and sensor. Range signals are taken as soon as the front proximity sensor detects a tree present Because there is no rear presence signal to start the catching

PAGE 54

54 operation at the first tree this test is programmed for two proximity sensors to be turned on at the same time for the first and second tree Using filtered and modulated range signal, PLC m easures range to the trunk and decides operation case appropriate for tree position. In Figure 5 3, the first movement of LMU can be observed between 100 0 ms and 170 0 ms with the rear LMU starts to move to center 20 0 ms after finishing the front catching operations. Figure 5 4 shows tilting operation of BCUs. T he #1 and #3 LMU extends ahead and #2 and #4 LMU extends later following the front LMU. S imilarly, BCU tilts up similarly to operating steps of LMU. The catching cases for the left misaligned tree an d for right misaligned tree are described in Figure 5 5 and 5 6. In open catching case, the catching system creates enough space to avoid any tree contact by tilting entire BCU up to 5 5 degree s and fully retracting entire LMU. Figure 5 7 shows the retract ion of each actuator to skip catching. In Figure 5 3, it is observed that front BCU set tilts up between 690 0 ms and 730 0 ms and rear BCU set tilts up between 790 0 ms and 830 0 ms. Sixty eight mature citrus among 69 harvested citrus were caught on the catching system resulting in a recovery rate of 98.55%. It avoid ed catching 17 immature citrus out of 25 dropped re sulting in a skip rate of 68%. T he effectiveness of the open catching mode is demonstrat ed not only by the skipping rate but also by moving response of LMUs and BCUs.

PAGE 55

55 Figure 5 1 Complementary Tree and Citrus Set Photo Courtesy of Kyusuk Figure 5 2 Step Signal from T he F ront and R ear Proximity Sensors

PAGE 56

56 Figure 5 3 Monitoring of LMU s Operation for T he Center T ree Figure 5 4 Monitoring of BCU s Operation for T he Center T ree

PAGE 57

57 Figure 5 5 Monitoring of LMU s Operation for T he Left O ffset T ree Figure 5 6 Monitoring of LMU s Operation for T he Right O ffset T ree

PAGE 58

58 Figure 5 7 Monitoring of LMU s Operation for Skipping Catching Citrus

PAGE 59

59 Table 5 1 R ecovery r ate from 3 c atching m odes Static Catching Test Dynamic Test under Proximity sensing Dynamic Test under Range sensing Total Amount (Ea) 75 75 75 Harvested (Ea) 75 73 69 Recovery (Ea) 75 70 68 Recovery Rate 100% 95.89% 98.55% 1. All values were averaged from 3 repeated experiments 2. Recovery rate is percentage value of amount of harvested citrus/amount of fallen citrus on the catching systems. Table 5 2 S kipping r ate from 3 c atching m odes Static Catching Test Dynamic Test under Proximity sensing Dynamic Test under Range sensing Total Amount (Ea) 25 25 25 Harvested (Ea) 25 25 25 Recovery (Ea) 19 7 17 Skipping Rate 76% 28% 68% 1. All values were averaged from 3 repeated experiments 2. Skipping rate is percentage value of amount of harvested citrus/amount of fallen citrus on the ground. 3. The lowe r skipping rate in Dynamic test under proximity sensing is due to the small space in the BFG T herefore more citrus fell on the catching system.

PAGE 60

60 CHAPTER 6 CONCLUSION This study demonstrated the efficiency of a new type of catching system developed to reduce drop losses for continuous harvesting process. Tree position were sensed in PLC using two photoelectric sensors detecting tree presen ce and an IR range sensor measuring range to tree trunks. The catching system aligned its Between Frame Gap ( BFG ) according to tree positions us ing four linear actuators to LMUs. Four actuators were used to tilt BCUs for catching d ropped citrus and for sel ective catching of limited citrus depending on its maturit y T he power source for this system was 110 AC volts from wall outlet T he tree row was simulated manually by pulling the base board of tree row T he average driving speed of a tree row was 43 mm/sec. The driving speed discussed in Chapter 4 was lower than 12 mm/sec. However, during the test, the driving speed was as low as 12 mm/sec only when LMU set moved from left misaligned position to right misaligned position. E xcept in this case the driv ing speed was over 50 mm/sec. Simulated citrus was harvested by beating canopies with long polyethylene foam, strips instead of shaking canopies. The following conclusion s were found from this study The strip brush broom had enough strength to support and to hold falling citrus. In addition, brush brooms absorbed vertical impacts from fallen citrus which could prevent citrus out of bounce and damage, and help collect fruits I n addition the brush materials of strip brooms were flexible enough to seal arou nd moving and irregularly aligned trees. Shown in Table 5 1, during static catching test, it could catch all of falling citrus and the dynamic test reported average recovery rate wa s 98.55%. Comparing with results from previous study reported by Peterson et al. (1998), they presented recovery rate of 9 0%

PAGE 61

61 using plate type catching frames. However, one might expect that citrus could fall to ground through the V space between brush bundles as shown in Figure 6 1. Active aligned catching units made from strip brushes keep on sealing continuously for misaligned trees. Figure 5 2 and Figure 5 4 to 5 6 shows the operations of actuators moving along with BFG for differently located trees at the right time. However, in o rder to increase catching speed, it needs to upgrade the stroking speed of actuators D uring test, the speed of actuator decided operating time of the catching system. The open catching case was expected to avoid contacting tree trunks and catching fruits by creating a large clear space in BFG T he mean skipping rates from three catching test modes were evaluated based on assumption that immature citrus fell on the ground. E ven though the harvesting machine stops shaking for immature tree in real fields, s ome of citrus dropped to the ground. T his open catching case is to skip catch ing immature citrus during harvesting process. However, the dynamic test under proximity sensing reported the lowest skipping rate shown in Table 5 1 due to different tilting degr ee. T here was difference of 88.12 mm in width of opened BFG between two tests. Future Work From results of test it is expected that the catching operation s using strip brushes are feasible a nd active aligning system s seal BFG efficiently. However, further studies are required to certify its practical performance for real citrus harvesting. Applied static actuators should be faster than 50 mm/sec in order to stabilize operating speed of the catching system. T he stroking speed of actuators d ecides the driving speed of catching frames.

PAGE 62

62 Additional catching units will be required in field test T wo catching sets were use d in th is experiment. T he effective harvesting length of a double drum canopy shaker is 6.7m as reported by Peterson s paper. I n order to catch all fruits from a tree the catching system should be longer than 7m. Closed loop control can be implemented by applying t ransducers to measure tilting angle of BCUs and moving distance of LMU in the catching system. In this study, rotati ng angle and moving distance was calculated by actuator operation time. H owever, stroking length of actuators was different depending on applied load and starting position of cylinders. R eal time feedback data from each transducer can make systems operate more precisely T he strip brushes used in this study is manufactured for street sweeping vehicles that would inevitab ly damage the tree trunk for continuous contact from brush filaments. T herefore, a unique brush material size and stiffness should be selected specifically for citrus.

PAGE 63

63 Figure 6 1 V S pace among T ree B undles

PAGE 64

64 APPENDIX 3D PART MODELING AND DRAWING 1. Brush Catching Unit (BCU) 2. Brush Bracket

PAGE 65

65 3. Drawing of Bracket 4. Drawing of Collecting Bed

PAGE 66

66 LIST OF REFERENCES Adrian P. A. and R. B. Fridley. 1959. New Concept of Fruit Catching Apparatus Tested. American Society of Agricultural Engineers 2 (1): 30 31. Adrian P. A. R. B. Fridley, and L. L. Claypool. 1968. Adapting Shake Catch Method of Harvesting to Cling Peaches. American Society of Agricultural Engineers 11 (2): 159 163. Antuniassi U.R., Z.N. Figueiredo, W.P.A. de Carvalho and D.R. Pinto 2002. Performance Evalu ation on Electronic Control Systems for Field Sprayers Using Programmable Logic Controllers Proceedings of the World Congress of Computers in Agriculture and Natural Resources : 272 278. Burkner P. F. and J. H. Chesson. 1976. Preliminary Investigations for Jojoba Harvesting 19 (4): 614 616. Burks, T. F. 2010. Machine Enhancements: Catch frame/ Recovery rate Improvements. Annual Research and Extension Report of IFAS Citrus Initiative. Chen, S., Y. C. Chiu, and Y. C. Chang. 2010. Developmen t of a Tubing Grafting Robotic System for Fruit Bearing Vegetable Seedlings. American Society of Agricultural and Biological Engineers 26 (4): 707 714. Chinchuluun R. W. S. Lee, and R. Ehsani. 2009. Machine Vision System for Determining Citrus Count and Size on a Canopy Shake and Catch Harvester. American Society of Agricultural and Biological Engineering 25 (4): 451 458. Churchill, D. B. 1980. A Direct Loading, Offset Pickup Machine for Citrus. American Society of Agricultural Engineers Paper No. 80 152 9. Coppock, G. E. 1976. Catching Frame Development for a Citrus Harvest System. American Society of Agricultural Engineers 19 (4): 627 630. Fridley, R. B., L. L. Claypool. 1975. A New Approaching to Tree Fruit Collection. American Society of Agricultural E ngineers 18 (5): 859 863. Futch, S. H., and F. M. Roka. 2005. Continuous Canopy Shake Mechanical Harvesting Systems. Horticulture Science Depts. in Institute of Food and Agricultural Science. Documentation No. HS1006. Grosz, F., Y. Sarig, J. Shamruk, R. Ke ndel, and H. Egoai. 1989. A Roll Out Catching Harvester for Citrus Fruit Destined for Processing. American Society of Agricultural and Biological Engineers 5 (3): 307 310. Millrier, W. F., G. E. Ruhkugler, R. A. Pellerin, J. A. Throop, and R. B. Fridley. 1 973. Tre Fruit Harvester with Insertable Multi Level Catching System. American Society of Agricultural Engineers 16 (5): 844 850.

PAGE 67

67 O Brien, M., R. B. Fridley, and L. L. Claypool. 1977 Food Losses in Harvest and Handling Systems for Fruits and Vegetables. American Society of Agricultural Engineering Paper No. 76 6509. Peterson, D. L. 1998. Mechanical Harvester for Process Orange. American Society of Agricultural Engineers 14 (5): 455 458. Robert, C E., K. B. Jacqueline, T. M. Kelly, and F. M. Roka. 2010 Ab scission Agent Application and Canopy Shaker Frequency Effects on Mechanical Harvest Efficiency of Sweet Orange. Horticulture Science 45 (7): 1079 1083. Schmilovitch, Z., A. Zaltzman, A. Hoffman, and Y. Edan. 1995. Firmness Sensing and System for Date Sor ting. American Society of Agricultural Engineering 11(4): 555 560. So, J. D., and J. W. Cho. 2010. Design Considerations and Feasibility of a Prototype Mechanical Bokbunja Harvester. Meeting Presentation of American Society of Agricultural and Biological E ngineering Paper No. 1008911. Specification of Oxbo Blueberry Harvester Model 7420, http://www.oxbocorp.com/Portals/0/Oxbo/Berries/7420/742011.pdf Accessed on Oct 2010 Thomas R. S. and D. R. Buckmaster. 2003 A Programmable Multiple Supplement Cattle Feeder for Pasture Use. American Society of Agricultural Engineers 19 (4): 511 520. Video of BEI Jatropha Wave Harvester in Honduras, http://www.youtube.com/watch?v=D5oaeTRZpOo&feature=grec_index Accessed on Dec. 2009. Zocca, A., P. Rodati and A. Mazza. 1995. Performance of a New Tree Fruit Harvester. American Society of Agricultural Engineers. Paper No. 85 1565. ASAE Winter Meeting. Chicago, IL.

PAGE 68

68 BIOGRAPHICAL SKETCH Kyusuk You was born in Seoul South Korea He received his degree in Bachelor of Engineering in m echanical e ngineering from Hongik University Seoul South Korea in 200 6 After graduation, He worked at KUMHO Tire Co. in Seoul, South Korea until 2007 He joined the concurrent m aster s program in mechanical engineering and agriculture and biological engineering from University of Florida in Aug 200 7 He graduated Master of Eng ineering in mechanical engineering in August 20 1 1 and Master of Engineering in agricultural and biological engineering in December 201 1 He hopes to study in area of developing automation system applying in agriculture.