<%BANNER%>

The Control System Design of an Omni-Directional Vehicle Platform

Permanent Link: http://ufdc.ufl.edu/UFE0024642/00001

Material Information

Title: The Control System Design of an Omni-Directional Vehicle Platform
Physical Description: 1 online resource (65 p.)
Language: english
Creator: You, Kyu
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

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

Notes

Abstract: Advancements in robotics technology have enabled robots to become more popular and it is no longer unusual to see autonomous vehicles moving in industrial and other environments. When an autonomous vehicle is operating in a crowded environment such as a manufacturing area or in a building, there are more chances of interacting with people or social infrastructures and thus requiring high maneuverability of the vehicle platform. In this paper, I introduce an Omni-directional vehicle platform. The omni-directional aspect of the platform is expected to give high maneuverability of the autonomous vehicle by allowing the vehicle to move virtually in all direction on a plane. This feature will generate more options for path planning and obstacle avoidance when applied to an autonomous vehicle. This thesis describes part of the design and development of an omni-directional vehicle platform. It focuses on the hardware control system. A prototype vehicle controller using an embedded micro controller and a Windows program is developed and tested. This omni-directional vehicle has three independently steerable wheels and steering actuators. The control methods of different types of actuators are summarized followed by a series of tests.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kyu You.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Crane, Carl D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024642:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024642/00001

Material Information

Title: The Control System Design of an Omni-Directional Vehicle Platform
Physical Description: 1 online resource (65 p.)
Language: english
Creator: You, Kyu
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

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

Notes

Abstract: Advancements in robotics technology have enabled robots to become more popular and it is no longer unusual to see autonomous vehicles moving in industrial and other environments. When an autonomous vehicle is operating in a crowded environment such as a manufacturing area or in a building, there are more chances of interacting with people or social infrastructures and thus requiring high maneuverability of the vehicle platform. In this paper, I introduce an Omni-directional vehicle platform. The omni-directional aspect of the platform is expected to give high maneuverability of the autonomous vehicle by allowing the vehicle to move virtually in all direction on a plane. This feature will generate more options for path planning and obstacle avoidance when applied to an autonomous vehicle. This thesis describes part of the design and development of an omni-directional vehicle platform. It focuses on the hardware control system. A prototype vehicle controller using an embedded micro controller and a Windows program is developed and tested. This omni-directional vehicle has three independently steerable wheels and steering actuators. The control methods of different types of actuators are summarized followed by a series of tests.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kyu You.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Crane, Carl D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024642:00001


This item has the following downloads:


Full Text

PAGE 1

1 THE CONTROL SYSTEM DESIGN OF AN OMNI DIRECTIONAL VEHICLE PLATFORM By KYUHYONG 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 SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Kyuhyong You

PAGE 3

3 To my mom and father

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my deepest thanks and gratitude to my advisor, Prof. Carl Crane, for his invaluable support supervision and advice for this research project. His guidance and provision lead this thesis to become successful. I would like to thank Prof. Antonio Arroyo for helping me to pursue my interest in robotics and become my committee. Without his a dvice, I would not have a chance to start this research. Thanks also go to the rest of my committee Prof. Scott Banks for hi s interest in my research project. I am also grateful to Shannon Ridgway for his technical support and practical advice throughout the development and testing of this project. I t hanks to all my fellow students at the Center for Intelligent Machines and Robo tics for their help and friendships I would never forget the time I shared with them. Lastly, I would like to thank my parents, Miyong Oh and Keunoh You, for their endless support and love throughout my life. I thank my brother, Sunhyong, for his care and support. I would also like to thank my friend s Se okke un Kim Sangwook Park and M. Park for their friendships and encouragements.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 7 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 10 ABSTRACT ................................ ................................ ................................ ................................ ... 11 CHAPTER 1. INTR O D U CUTION ................................ ................................ ................................ .................. 12 Introduction ................................ ................................ ................................ ............................. 12 The Concept of Omni Directional Mobility ................................ ................................ ........... 13 Locomotion ................................ ................................ ................................ ...................... 13 Wheeled mobile robots ................................ ................................ ............................. 14 Degree of mobility ................................ ................................ ................................ ... 18 Degree of steerability ................................ ................................ ............................... 18 Omni Directional Vehicle Platform ................................ ................................ ....................... 18 Special Wheel Designs ................................ ................................ ................................ .... 19 Conventional Wheel Design ................................ ................................ ............................ 20 The OmniGator ................................ ................................ ................................ ................ 21 Thesis Focus ................................ ................................ ................................ ........................... 22 2. VEHICLE SYSTEM DESIGN ................................ ................................ ................................ .. 24 Vehicle System Architecture ................................ ................................ ................................ .. 24 Power System ................................ ................................ ................................ .................. 24 Hardware Control by Embedded Micro Controller ................................ ......................... 25 Motor Control System ................................ ................................ ................................ ............ 26 3. A CTUATION SYSTEMS AND CONTROL ................................ ................................ ............ 28 Throttle Syste m ................................ ................................ ................................ ....................... 28 Throttle System H ardware C omponents ................................ ................................ ......... 28 Brushledd DC w heel m otor ................................ ................................ ...................... 29 B rushless DC motor amplifier ................................ ................................ .................. 29 Digital to a nalog c onverter ................................ ................................ ....................... 29 Encoder and m icro c ontroller ................................ ................................ ................... 31 Detecting the RPM ................................ ................................ ................................ .......... 33 Noise issue with using encoder in high switching current environment ......................... 34 PID C ontroller I mple mentation ................................ ................................ ....................... 38

PAGE 6

6 Servo Actuator System ................................ ................................ ................................ ........... 41 SmartMotor Servo Actuator ................................ ................................ ............................ 41 Steeri ng A lignment ................................ ................................ ................................ .......... 42 4. S OFTWARE CONTROL ARCHITECTURE ................................ ................................ ........... 45 Embedded Wheel M otor and Hardware Controller ................................ ................................ 45 Computer S oftware and C ommunication ................................ ................................ ............... 45 Control by Joystick ................................ ................................ ................................ ................. 48 5. TESTING AND RESULTS ................................ ................................ ................................ ....... 51 Methodology ................................ ................................ ................................ ........................... 51 Fabrication of the OmniGator C ontrol Board ................................ ................................ 51 Data Acquisition ................................ ................................ ................................ .............. 51 Test Using D ynamometer ................................ ................................ ................................ 52 Results ................................ ................................ ................................ ................................ ..... 53 Motor R esponse C omparison ................................ ................................ .......................... 53 Tuning the D igital PID C ontroller ................................ ................................ ................... 54 6. CONCLUSIONS AND FUTURE WORK ................................ ................................ ................ 59 Conclusions ................................ ................................ ................................ ............................. 59 Future Work ................................ ................................ ................................ ............................ 59 APPENDIX A O MNIGATOR CONTROLLER BOARD DESIGN ................................ .............................. 61 LIST OF REFERENCES ................................ ................................ ................................ ............... 63 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 65

PAGE 7

7 LIST OF TABLES Table page 3 1 Summary of backlash considerations for the gear train ................................ .......................... 33 4 1 Communication protocol ................................ ................................ ................................ ......... 47 4 2 Command message definitions from PC to the wheel controller ................................ ............ 47 4 3 Command message definitions from the wheel controller to PC ................................ ............ 48

PAGE 8

8 LIST OF FIGURES Figure page 1 1 TailGator ................................ ................................ ................................ ................................ .. 12 1 2 Finished design of an omni directional wheel and the vehicle platform ................................ 13 1 3 Physical dimensions of the platform. Dimensions are measured from the prototype platform and drawn on a CAD system. ................................ ................................ .............. 15 1 4 Reference frames ................................ ................................ ................................ ..................... 16 1 5 Wheel types ................................ ................................ ................................ ............................. 17 1 6 Kinematic expression of the omni directional wheel mechanism ................................ ........... 17 1 7 One, two, and three degree of mobility conf igurations ................................ ........................... 18 1 8 Example of Omni directional wheels ................................ ................................ ...................... 19 1 9 An industrial lift truck using Mecanum wheel ................................ ................................ ........ 20 1 10 The Seekur, a commercially available omni directional wheeled robot ............................... 21 1 11 The OmniGator redesigned on a CAD system ................................ ................................ ...... 2 2 2 1 Schematics of power system of the omni directional vehicle ................................ ................. 25 2 2 The component interface architecture ................................ ................................ ..................... 27 3 1 Schematic diagram of an R 2R ladder type digital to analog converter ................................ .. 30 3 2 Connections between the controller, DAC, motor amplifier and the motor ............................ 31 3 3 A HEDM 6140 Code wheel used for detecting angular velocity of the wheel motor ............ 32 3 4 Schematic diagram of the gear train inside the wheel motor ................................ .................. 32 3 5 Safe operating area of ATmega128 micro controller ................................ .............................. 35 3 6 A w iring diagram shows connections between motor and the motor controller side ............. 35 3 7 The effect of electric isolation for encoder input ................................ ................................ .... 36 3 8 Switching charac teristic and Voltage Waveforms ................................ ................................ .. 37 3 9 Noise removed from the output signal of the encoder sensor ................................ ................. 38

PAGE 9

9 3 10 Simplified block diagram of discrete digital PID controller ................................ ................. 39 3 11 Proportional control and steady state error ................................ ................................ ............ 40 3 12 Actual implementation of a PID controller for the wheel motor ................................ ........... 41 3 13 Configuration of the optical switch for homing ................................ ................................ .... 43 3 14 A configuration of the laser level for wheel alignment ................................ ......................... 44 4 1 A c omputer software developed to control the vehicle ................................ ........................... 46 5 1 Pilot board developed for the vehicle platform ................................ ................................ ....... 51 5 2 Software developed to t une the PID controller ................................ ................................ ....... 52 5 3 Configuration of the dynamometer used for the test ................................ ............................... 53 5 4 Motor comparison test result ................................ ................................ ................................ ... 54 5 5 Step responses of the PID motor controller for different P gain. ................................ ............ 55 5 6 Step response of the motor with PI control ................................ ................................ ............. 56 5 7 K D gain tuning for motor C ................................ ................................ ................................ ..... 57 5 8 Step response of all three motors ................................ ................................ ............................. 58 A 1 Schematic diagram of the embedded motor control board. ................................ .................... 62

PAGE 10

10 LIST OF ABBREVIATIONS CAD Computer Aided Design CIMAR Center for Intelligent Machines and Robotics DOF Degree of Freedom HMV High Mobility Vehicle IGVC Intelligent Ground Vehicle Compe tition JAUS Joint Architecture for Unmanned System

PAGE 11

11 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 THE CONTROL SYSTEM DESIGN OF AN OMNI DIRECTIONAL VEHICLE PLATFORM By Kyuhyong You August 2009 Chair: Carl Crane Major: Mechanical Engineering Advancement s in robotics technology ha ve enabled robots to become more popular and it is no longer unus ual to see autonomous vehicles movin g in industrial and other environments When an autonomous vehicle is operating in a crowded environment such as a manufacturing area or in a building, there are more chances o f interacting with peo ple or social infrastructures and thus requiring high mane uverability of the vehicle platform. In this paper I introduce an Omni directional vehicle platform. The omni directional aspect of the platform is expected to give high maneuverability of the autonomous vehicle by allowing the vehicle to move virtually i n all direction on a plane. This feature will generate more options for path planning and obstacle avoidance when applied to an autonomous vehicle. This thesis describes part of the design and development of an omni directional vehicle platform It focuse s on the hardware control system. A prototype vehicle controller using an embedded micro controller and a Windows program is developed and tested. Th is omni directional vehicle has three independently steerable wheels and steering actuators. The control me thods of different types of actuators are summarized f o llowed by a series of tests.

PAGE 12

12 CH APTER1 INTROD U C TION Introduction The Center for Intelligent Machines and Robotics (CIMAR) in the Mechanical and Aerospace Engineering Department at the University of Florida has long participated in research and development of autonomous vehicles. Among achievements made by CIMAR in robotics, participating in the Intelligent Ground Vehicle Competition ( I GVC) resulted in great success in past years The TailGator shown in F igure 1 1 was developed in 200 3 and proved competitive in autonomous ground vehicle design by winning 1st place in the competition Figure 1 1 TailGator (CIMAR, 2003) Like the vehicle shown above, most of the mechanisms u sed in developing autonomous vehicle system s at the University of Florida were based on Ackerman steering or tracked vehicles [1] Some res earchers at CIMAR found limitation s of vehicles mobility due to the non holonomic constraints of the wheels. T his stud y aimed to develop a high mobility vehicle (HMV) that would not be subject to the same motion constraints Fulmer (2003) successfully demonstrated the performance of an omni directional wheel for the HMV system[2] The omni

PAGE 13

13 directional wheel developed at t he CIMAR is basically a type of brushless motor embedded in the hub of a wheel with all the gear trains and cooling pipes around it. The figure 1 2 shows the work done by Fulmer and the platform of the vehicle. Figure 1 2 Fi nished design of an omni directional wheel and the vehicle platform (Fulmer et al., 2003) The vehicle platform is composed of a main body frame, three independent steering motors and omni directional wheel motors. The wheel motors are assembled to the ste ering actuator with an L shape wheel bracket. Though the platform was developed no further test or study has been done due to a lack of available controllers. The design and development of a controller for this vehicle platform will be described in the re st of this thesis. The Concept of Omni Directional Mobility Locomotion Typically, an autonomous vehicle system has a designated mission or target which may be located o n the ground se a or in the sky. One method to describe effected motion on a vehicle is

PAGE 14

14 by the forces required for operation. L ocomotion describes the process of causing an autonomous vehicle to move [3] Implementation of locomotion deals with the design of a mobile platform with a control mechanism that delivers an autonomous system to the target most efficiently and precisely. Wheeled mobile robots Wheeled mobile platforms are widely used in autonomous ground vehicle system s Several methods exist to describe the behavior of a vehicle based on the number of wheels and their configuration. The d esign of a wheeled robot involves kinematics and dynamics knowledge of the specific vehicle platform. Kinematics deals with the geometric combination of components that construct the vehicle platform such as the location of wheel s and types of bracke ts used. The s tudy o f kinematics helps identify the motions of the vehicle without consideration of the forces that are applied to the vehicle. Kinematics describes the constraints of control mechanisms and behavior of a system. Figure 1 3 presents the phy sical dimensions of the vehicle system. Note that the arbitrary heading position is defined here for reference. Three wheel motors are named as motor A, motor B and motor C as they represent front side, rear left and rear right respectively.

PAGE 15

15 Figure 1 3 Physical dimensions of the platform. Dimensions are measured from the prototype platform and drawn on a CAD system. If forces are applied to a vehicle system, they will cause the vehicle to start moving and the magnitude and dir ection of the forces will affect the resulting motion. Study of dynamics involves understanding of dynamic motion and predicts the future orientation and acceleration of the vehicle. The fo llowing assumptions are made to model the wheeled vehicle. 1. The rob ot is built from rigid mechanisms. 2. No slip occurs in the perpendicular to the direction of rolling (no slipping) 3. No translational slip occurs between the wheel and the floor (pure rolling) 4. The robot contains at most one steering link per wheel. 5. All steerin g axes are perpendicular to the floor. To describe the wheeled robot the f o llowing parameters are applied

PAGE 16

16 Figure 1 4 Reference frames {X b Y b } represents fixed base frame {x m y m } represents moving reference frame of the vehicle Then the posture of t he vehicle in the base frame can be expressed as q = x y Vehicles posture in base frame Now the vehicle kinematics can be expressed as the types of wheels attached to the platform. The various types of wheels can be categorized by steerability, orienta tion of the steering, and wheel configurations. Figure 1 5 shows the most commonly used wheels on vehic ular platforms.

PAGE 17

17 Figure 1 5 Wheel types ( Xiao, J., Mobile Robot Locomotion ) The steerable wheel can be expressed with the orientation of the rotation axis. If the wheel rotates about the geometric center of the wheel as described in the figure 1 2 Figure 1 6 Kinematic expression of the omni directional wheel mechanism r = wheel radi us v = whee l linear velocity w = wheel angular velocity t = steering velocity

PAGE 18

18 Degree of mobility Degree of mobility describes the degree of freedom of the vehicle motion. The term instantaneous center of rotation (ICR) or instantaneous center of curvature (ICC) is u sed to define the number of degrees of freedom, as shown in Figure 1 7 The omni directional vehicle described in this thesis has a mobility of three, consisting of two translational and one rotational Figure 1 7 One, two, a nd three degree of mobility configurations ( Xiao, J., Mobile Robot Locomotion ) Degree of steerability The degree of freedom of a vehicle is often denoted as degree of steerability and is defined by the number of centered orientable wheels that can be ste ered. If no centered orientable wheels exist, the vehicle is said to have zero degree s of steerability. Omni Directional Vehicle Platform As described, omni directional mobility of wheeled vehicle s can be achieved by expanding the steerability of the wheel s. There have been several approaches proposed to implement an omni directional vehicle. Among those, s teerable w heels and Mecanum 1 wheels are most widely discussed and researched previously. In the pa st, four wheel ed steering vehicle s [4] ha ve been 1 Mecanum wheels refer to the name of the original Swedish manufacturer.

PAGE 19

19 develo ped. The vehicle has steering wheels with conventional tires and can move omni directionally by orienting all wheels in the same direction and moreover it can rotate. Omni wheel or Mecanum wheel approach introduced rolling mechanisms to the conventional t ype wheel so that it can move laterally. Special Wheel Designs Special wheel designs often include modification of conventional wheels or rollers. Mecanum wheels, referring to the name of the original Swedish manufacturer, have rollers as treads. The rolle rs have axles skewed with respect to the wheel axle [5] These wheels provide a practical way of providing simultaneous vehicle motion in all three direction s, longitudinal, lateral, and yaw, without singularities A B C D Figure 1 8 Example of Omni directional wheels A) Mecanum Wheel (Stephen, L. D., Brett, D. L) B) Omni Wheel (Society of Robots, http://www.societyofrobots.com/robot_omni_wheel.shtml Las t accessed 04/ 20/2009) C) WESN Wheel D) Omni Ball (Kenjiro, T. et al., 2007)

PAGE 20

20 Yasumasa K et al, have developed an omni directional wheelchair using Mecanum wheel s [6]. An Omni directional lift truck from Airtrax shown in the figure 1 9 is one of the e xamples of using Mecanum wheels. This vehicle is commercially available now. Figure 1 9 An industrial lift truck using Mecanu m wheel (Airtrax ) Note: http://www.airt rax.com/vehicles/sidewinder.html Last accessed 04/20/2009 Some miniature sized Omni wheels are widely used in smaller applications such as robot soccer player s [7] developed by Cho, S.H. Mecanum wheels also are available in miniature sized robot s [10]. Kenjiro and Riichiro developed even more rigorous and complex omni directional wheel using spherical wheel s [8]. Each of t hese special wheel designs greatly extended the maneuverability of vehicles but certain limits also apply The relatively smaller dia meter of the rollers often limits the range of operation to even surface. Also, when omni wheels or Mecanum wheels are operating on an unpaved road similar to where IGVC is held, it is hard to predict the future orientation for a particular command Moreov er rolling mechanisms often increase s power loss due to conflicts among other actuators [ 9]. Conventional Wheel Design Conventional wheel type design s often employ indecently driven steering actuators for omni directional mobility. These types of design u sually have larger load capacities and are

PAGE 21

21 robust to irregular ground surface s Also the simplicity of implementing steerable wheel mechanisms made it widely used in industry, robotics and military [11] The SEEKUR robot shown in the Figure 1 10 was devel oped by Mobile R o bots and demonstrated its ability of operating in tough environment with four omni directional steerable wheels. However conventional wheels with steering mechanisms for omni directional vehicle s are not considered as they have non holono mic mobility [12] because when the vehicle encountered a non continuous curve, there is a finite amount of time before the steering mechanism can reorient the wheel to the next projected curve [2]. Figure 1 10 The Seekur, a commercially available omni directional wheeled robot (Mobile Robot) The Omni Gator Based on the preliminary research on the omni directional vehicle platform, research ers at CIMAR developed the Omni Gator. For the purpose of having robust mobility on unpave d road, t he OmniGator has three independently steerable hub drive w heel motor. One of the distinct features of the OmniGator is that its steered driving wheel is designed to have zero offset. This design is helpful to avoid extra forces that may be applied when steering the wheel instantaneously. Figure 1 11 shows computer aided design of the omni directional vehicle platform developed at CIMAR.

PAGE 22

22 Figure 1 11 The OmniGator redesigned on a CAD system Thesis Focus The OmniGator vehicle platform including wheel motors, steering actuator, carbon fiber body and wheel brack ets was built but no further attempt s were made to build a control system for it. In order to make the OmniGator to become operational as an autonomous ground vehi cle, getting the control of the wheels and actuators should be the first step into the system design. Since most of the mechanical parts were already made, certain restrictions such as the performances of the motor and gear trains exist in designing the sy stem. Furthermore, many difficulties were expected mostly due from using fully customized brushless DC motor for the omni directional wheel. A functional motor controller based on classic PID control method will allow for autonomous computer control of the platform. Control of the steering actuators was part of designing the control architecture of the OmniGator.

PAGE 23

23 This thesis will cover the design and implementation of the vehicle actuation systems with omni directional mobility. An electric hardware system architecture and c ontrol interface is introduced Details of actuation system are presented with operational theory of control system. With designed PID motor throttle controller, PID gain tuning procedures and results will also be discussed.

PAGE 24

24 C HAPTER 2 V EHICLE SYSTEM DESIGN This chapter discusses conceptual design of the entire power system of the vehicle platform and control architecture of the actuation system. Since the vehicle is expected to be operated autonomously, sensor architectures and higher le vel control system should be discussed as well as actuation system of the vehicle In the later part of this chapter, a vehicle control architecture is presented. Vehicle System Architecture Power System T he autonomous vehicle system is entirely electric p owered and sourced from l ead a cid batteries. The stack of batteries can supply 48 Volt DC as a main power source and it relies on a n onboard gas powered generator for charging. The AC input from the power generator is converted to DC and is used for charg i ng the batteries. Main power is directly supplied to the three motor amplifiers and SmartMotor s which consume s most of the power The remaining power is ste p p ed downed to respective device s power rating using DC to DC converter. Detailed schematic of elec tric power system is drawn in figure 2 1 For the purpose of emergency stop (E Stop) features, both soft and hard kill method have been considered. As described in the IGVC rules, every autonomous ground vehicle is required to have a mechanical and wirel ess E Stop method [13]. The main kill switch is located between a 12VDC converter and relays. This would not be connected to the onboard computer which contains higher level controller and only affect s low level device controller causing immediate reset.

PAGE 25

25 Figure 2 1 Schematics of power system of the omni directional vehicle Hardware Control by Embedded Micro Controller In order to prevent electric al leakage or possible damage from unexpected operation, the high powered elect ric actuator systems must be turned on or off by user s For safety, it is recommended that a ny autonomous vehicle have functions that allow operators to shutdown the vehicle at anytime. All hardware component s ha ve relays betwe en the power s ource and ground that are controlled by an embedded micro controller. ATmega128 micro controller from ATMEL is used to control the relays. The controller I/O pins are connected to the relays allowing it to turn the connected device on or off The microcont roller will determine when to activate the devices based on control inputs received from the onboard computer. To prevent a n accidental switch on

PAGE 26

26 during microcontroller initialization an inverter is inserted between the relay arrays and microcontrol ler s I/O pins. Motor Control System Before design of the control system can begin it is essential to understand the characteristics of the system including types of inputs and output s communication method, and control method. The vehicle control methodologi es in designing an omni directional vehicle are broadly separated into two types: centralized and decentralized [1 4 ]. In the centralized type, wheel modules control their own motions based on information commanded by a central controller. By contrast the decentralized type has no central controller, and individual wheel modules synchronize their motions using local information. The centralized type can easily reduce the conflict among motions of wheel modules under non holonomic constraints. Therefore a c entralized type controller has been selected for this project. The entire vehicle platform is composed of two different types of co mponents : steering and throttle system s For the steering system, a pre existing SmartMotor was chosen as it contains an embe dded controller and can natively talk through a RS 232 serial connection Figure 2 2 shows a simplified view of the control architecture for the omni directional vehicle platform. The onboard computer takes inputs from either an autonomous controller or ma nual control input such as a joystick. Depending on the mode of operation, the omni directional motion generator determines the control efforts of each component. A steering angle command is sent to the SmartMotor controller via an RS 232 serial port and t he latest position of the steering angle is updated. A different approach is required for throttle control since the motor is incapable of directly processing a command from the system. Instead, a real time embedded micro controller is used to interpret th rottle input. This controller has a RS 232 serial port to communicate with the main

PAGE 27

27 computer while controlling the motor. A real time system is required for completing the digital PID controller. Figure 2 2 The component int erface architecture

PAGE 28

28 CHAPTER 3 A CTUATION SYSTEMS AND CONTROL This chapter d escribes the control implementation for both throttle and servo actuator. Hardware configurations and control methods are discussed in this chapter. For the throttle control system a discrete digital PID controller is introduced. Steering and alignments for servo actuator is discussed in the later part of this chapter. Throttle System The vehicle s throttle system is responsible to control the velocity of the vehicle determined by the higher level control architecture. The OmniGator has three brushless DC motor (BLDC) for throttling the vehicle. For driving the motors, three analog input BLDC motor amplifier s are used The motor amplifier converts voltage input to switching currents for the motor. The throttle control system mainly produce analog voltage as an input to the motor driver and sense the velocity of the wheel using encoder pulse output. A PID controller has been implemented using the micro controller. This section discuss thrott le control system. Throttle System H ardware C omponents The throttle system of the OmniGator is mainly co mposed of three BLDC wheel motors, motor amplifiers and a motor controller. In the implementation of the control system, it is required to understand the kind of inputs and outputs of certain components and to find ways to connect different components together. The Omni Directional wheel motor generates the main throttle power for the OmniGator and is driven by a br ush less motor amplifier. The fully customized motor s were designed and fabricated by Fulmer [2]. A d etailed explanation of its design, construction and testing is fully described in his thesis.

PAGE 29

29 Brushledd DC w heel m otor Designing the BLDC motor controller requires basic understanding of BLDC motor itself. There are several parameter s affect the velocity control of a BLDC motor such as Back EMF, torque constant. Following torque and Back EMF equations describes the property of the motor. The Back EMF magnit ude can be written as = 2 (3 1 ) A nd the torque term as = 1 2 2 1 2 2 + 4 (3 2 ) W here N is the number of winding turns per phase, l is the l ength of the rotor, r is the internal radius of the rotor, B is the rotor magnet flux density, is the motor s angular velocity, i is the phase current, L is the phase inductance, is the rotor position, R is the phase resistance. Equation 3 1 shows the Back EMF is proportional to the motor velocity. Equation 3 2 explains relationship between torque and the phase current. The first two terms in the torque expression are parasitic reluctance torque components. The third term produces mutual torque, which i s the torque production mechanism used in the case of BLDC motors. Brushless DC motor amplifier BE40A8 motor amplifier from Advance Motion is used to power the drive wheels. This particular driver takes analog voltage range from +/ 15 Volts as an input an d translates it into switching power source to the motor. Digital to a nalog c onverter As explained earlier, the motor amplifier requires analog signal as a reference. A DAC7715 from TI was selected for this application The DAC interprets the receiving se rial peripheral interface (SPI) command from the embedded micro controller to convert reference

PAGE 30

30 voltage to an analog voltage output. Figure 3 1 shows the basic operation of the DAC The DAC 7716 is a quad, serial input, 12 bit, voltage output DAC. It has a rchitecture of classic R 2R ladder configuration in conjunction with following operational amplifier which served as a buffer. As shown in the figure Figure 3 1 Schematic diagram of an R 2R ladder type d igital to a nalog conve rter The minimum and maximum voltage output is set by external voltage references (V RFE H for higher reference voltage and V REF L for lower reference voltage ). In this case they are set to +/ 10 Volt for both directions and are supplied by the motor amplif ier. 12 bit DAC code which determines the voltage output is transmitted with respective output channel in front completing a 16 bit serial input to the SDI pin. Detailed connection between the DAC and micro controller is shown in the figure 3 2

PAGE 31

31 Figure 3 2 Connections between the controller, DAC, motor amplifier and the motor Analog output voltage from the DAC is given by following equation = + 4096 (3 3 ) Where N is the digital input code in decimal which is transmitted as a binary format to the DAC. Each new instruction requires at most 10 micro second settling time. Encoder and m icro c ontroller When implementing a motor controller, knowing the current velocity of the motor is very important. Rotary type encoders are most widely used for this purpose. The wheel motor has a code wheel with 1000 Count per Revoluti on (CPR) attached to the core shaft of the motor and an optical encoder is fixed about the code wheel as shown in the F igure 3 3 The optical encoder generates pulses whenever it detects the slits surrounding the code wheel. A microcontroller is

PAGE 32

32 used to co unt the pulses from the encoder. By detecting both rising and falling edges of the pulses, it can double the resolution of the encoder code wheel as 2000 CPR. Figure 3 3 A HEDM 6140 C ode wheel used for detecting angular veloc ity of the wheel motor (Datasheet, Two and Three Channel Codewheels for Use with Agilent Optical Encoder Modules Agilent Technologies, Figure 12, pp 8) As the Figure 3 4 shows, the gear train includes several stages of gear and the final gear ratio of t he gear train is 30.3:1 allowing 5mph operation of the vehicle with a motor speed of 4250 rev/min. Figure 3 4 Sche matic diagram of the gear train inside the wheel motor (Fulmer, 2003)

PAGE 33

33 With the given gear ratio of the motor, t he actual encoder resolution becomes 6 0 6 00 CPR or approximately 0.0 06 /Count. However the accuracy of the encoder measurement can be affected by other factors such as gear backlash. Gear backlash as a definition is the play or clearance between mating teet h and is described as the amount of lost motion due to clearance. Although gear backlash is undesirable as it may introduce errors, it is an unavoidable mechanical property and should be considered in the design process. Table 3 1 summarizes the inaccuraci es accounted for the design of the gear trains. Note that all dimensions in inches. Table 3 1 Summary of backlash considerations for the gear train (Fulmer, 2003) Gear Stage 1 st Stage epicyclic gear train 2 nd Stage epicyclic gea r train Backlash S ources Mesh1 Mesh2 Mesh 1 Mesh2 Sun Planets Ring Sun Planets Ring Design Backlash 0 .00342 0.00383 0.00159 0.00331 0.0015 0.0025 0.0037 0.00409 0.00225 0.00374 0.00285 0.00444 S ums Minimum Radial Clearance 0.00501 0.00481 0.00 595 0.00659 Maximum Radial Clearance 0.00714 0.00409 0.00783 0.00669 Detecting the RPM The micro controller has 6 I/O pins dedicated for external trigger interrupt that can be used for counting the pulses. Whenever a rising pulse from an encoder channel is detected, an interrupt is triggered, incrementing the number of counts. Periodic internal timer clock triggers an interrupt and calculate s the RPM f o llowed by resetting the counter. The i nternal timer is set to 8 ms for this application. The RPM can be obtain ed by following equation. RPM = 1 Rev 2000 Counts N counts 1 Sampling Time 1 Sampling Time 8 ms 1000 ms 1 sec 60 sec 1 min (3 4 ) Where N denotes number of pulses counted during single sampling time.

PAGE 34

34 If considering the gear ratio between the encoder shaft and the wheel output, the equation becomes = ( ) 1 30 3 (3 5 ) Also note that sampling time is set to 8 ms. This sampling time is determined by internal tim er interrupt of the micro controller. The Timer/Counter in the ATmega128 is clocked by an external clock source. The external clock source used for the controller is rated at 16MHz and this clock is divided by a prescaling factor set in the Timer/Counter C ontrol Register. Depending on the value selected for the output compare register, sampling time is then calculated with the following equation. = 1024 ( ) 1 (3 6 ) Where f clk is the clock signal frequency and OCRn is the value set for the output compare register. By applying 16MHz clock signal and OCRn value of 128 we get = 16 10 6 1024 1 125 = 125 (3 7 ) T his is equal to 8ms timer period. Noise issue with using encoder in high switching current environment While testing the controller of the wheel motor, the controller reset was observed because of unexpected voltage dropping across the micro control and the encoder. This usually happens at higher motor speeds As specified in the datasheet of the ATmega128, any voltage drop below 4.5V makes the controller run at a slower rate. Figure 3 5 shows the safe operating area of the controller.

PAGE 35

35 Figure 3 5 Safe operating area of ATmega128 micro controller As described earlier, the omni directional wheel has integrated brushless DC motor A Hall Effect sensor and encoder sensor are also mounted inside the motor. Thus every cable runs along the same path as the amplifier to motor cables. The w iring diagram show n in Figure 3 6 illustrates how cables are connected between the controller s and the wheel motor Note that each cable is shielded and terminated to the ground. Figure 3 6 A w iring diagram shows connections between motor and the motor controller side Although it is not a good practice, for the ease of connection assembly an d to prevent wires from wrapping around the rotating axis, all the wires have to share the same path across the

PAGE 36

36 motor and controller. When high frequency switching current is flowing near a cable, noise becomes an issue especially when the feedback signal is located. The noise generated on the feedback signal simply can be described as f o llowing equation. = (3 8 ) W here I = Currents induced from the noise C = Capacitance between the cables dv/dt = Rise time of the servo amplifier signals The capacitance of the cables is largely related to the length of the cab le. It can be explained how antenna works and longer cable suddenly become an antenna that collects signals from the near cables. Generally, the best practice to minimize the effect of noise is to design the shortest path between the controller and the enc oder signal. However the kinematic constraints and mechanical configuration do not allow any cables shorter than 8 feet. Figure 3 7 The effect of electric isolation for encoder input 200 150 100 50 0 50 100 150 200 250 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Speed (RPM) Time (sec) w/ shield w/o shield

PAGE 37

37 Another method to solve this problem is i solating the noisy circuit f ro m the controller. This can be done by employing an electric isolator between the controller and encoder. The isolator or coupler is an integrated circuit widely used to protect controller s form high current source. When select ing an electric isolator for this kind of application, the rise time or response time of the isolator is much more important than other factors. A digital isolator ISO 7240 from TI semiconductor is used to isolate the noisy circuit from the controller. Fi gure 3 8 shows schematic and timing diagrams of the isolator. ISO 7240 has total of four channels and runs as fast as 48Mbps. This is equivalent of 2 ns of rise time. Figure 3 8 Switching characteristic and Voltage Waveforms (ISO 7240 Quad digital isolators) Figure 3 9 shows how the digital isolator effectively removed noise from the original sources. Also note that the delay of the output is almost negligible compare to the original input signal.

PAGE 38

38 Figure 3 9 Noise removed from the output signal of the encoder sensor PID C ontroller I mplementation When controlling a motor, there is no guarantee such that linear relationship exists between input and motor speed. System disturbance or loadings such as fri ction or noise induced from the control circuit can affect the output speed of the motor causing it to turn more slowly than desired. To minimize the error between the desired input and the system output, a feedback control system is used to compensate and provide corrections A PID controller is the most commonly used closed loop system in controlling motors. Figure 3 10 shows a typical application of a digital PID controller to a motor control system with analog motor amplifier. The symbol D/A represents digital to analog converter and in this case it is a 12 bit +/ 15 V olt DC converter.

PAGE 39

39 Figure 3 10 Simplified block diagram of discrete digital PID controller Every time the Counter/Timer updates the latest encoder count, PID controller tries to compare it with the control input. The error found in here can be expressed as (3 9 ) Where GoalRPM is desired speed of the motor, CurrentRPM is the latest encoder count measured during a sampling time. The sampling time in this case is set to 8 ms a sampling rate of 125 Hz. The first term in the PID controller tries to compensate the error by multiplying the current error by proportional gain constant K P as, Output P = K P E (3 10 ) In many cases, proportional control is the easiest and quick est way of solving a feedback control problem. Obviously increasing the K P gain does affect the response time of the system. However as figure 3 11 implies, when high internal f riction is present in the system, certain error may exist and that prevent s the system from reaching the desired speed. More commonly known as steady state error it can be addressed by introducing integral control.

PAGE 40

40 Figure 3 11 Proportional control and steady state error Integral control in a control system can reduce the steady state error to zero by adding previous errors. Output I = K I (E t) (3 11 ) W here K I is integral gain con s tant and (E t) i s sum of past errors. Since the integral control multiplies sum of all past errors, sometimes this term becomes so large that the error windup happens. By applying the limit of the error, this integral windup could be prevented. Integral control does pr ovide way to eliminate steady state error but at the same time it may increase overshoot or fluctuation of the output. Derivative control helps make system response faster while reducing overshoots. As the name implies, derivative control compares the last amount of error with current error as f o llowing equation, Output D = (3 12 ) Where K D is derivative control gain constant and is the rate of change of error.

PAGE 41

41 By summing all terms describ ed above, the PID controller takes the form (3 13 ) Figure 3 1 2 shows the simplified block diagram expression of just implemented PID controller. The detailed result and response of the controller is discussed in Chapter 5 of this thesis. Figure 3 12 Actual implementation of a PID controller for the wheel motor Servo Actuator System The Omni directional wheel motor is independently steered by a servo actuator system. When controll ing an omni directional vehicle, each wheel has to be aligned to the desired position before motion can be commanded, requiring fast and accurate response from the servo actuator system A SmartMotor wa s used as a main actuator for this system and a gear h ead with 49:1 ratio was attached to the end of the motor shaft to provide high resistance against the external forces such as shocks from tough terrain environment. Smart Motor Servo Actuator A S mart M otor SM3440D from ANIMATICS is a complete servo system e quipped with individual embedded motor controller amplifier and an encoder with resolution of 4000 counts

PAGE 42

42 per revolution ; this enables the system to control the motor without any external devices SmartMotor Interface ( SMI ) provides a programming environm ent which is very similar to basic programming language to program the action of the motor Furthermore when using the I/O channel of the motor, external switches or a potentiometer can be added to allow the motor to detect edges or home position. Steeri ng A lignment Physical alignment of the drive wheel is very important in controlling the steering angle of the omni directional wheel. If any of the wheels are not correctly aligned towards the omni directional motion, abnormal forces will be applied to the mechanism and that can potentially damage or destroy it. To prevent such a catastrophic event from happening, e very time the device is turned on the SmartMotor must be initialized prior to operation This initialization process includes homing and alignme nt. Homing can be easily done by polling limit switch output at the home position. However since the mechanical structure has already been designed and made without considering the location of such limit switch es non physical contact switch es such as op to electric switch is considered Optical switch is basically an optical sensor which responds to light with certain wavelength. Typical diffuse type optical switch takes infrared light as an input and light over threshold intensity will set the switch eithe r on or off. Figure 3 1 3 shows the configuration of the optical switch. A reflective tape is att ached to the top of the bracket to reflect infrared light from the emitter of the sensor.

PAGE 43

43 Figure 3 13 Configuration of the opti cal switch for homing Once the switch detects the edge of the reflective tape on the bracket, it triggers an input to the SmartMotor making it complete stop. The motor sets this position as a new home position. Then the steering alignment can be measured r elative to this home position. For the purpose of precision an d alignment, a laser level is used. Figure 3 1 4 shows how the omni directional wheel is aligned using laser level.

PAGE 44

44 Figure 3 14 A configuration of the laser level for wheel alignment

PAGE 45

45 CHAPTER 4 S OFTWARE CONTROL ARCHITECTURE Embedded Wheel M otor and Hardware Controller As described earlier, the digital PID controller is highly dependent on sampling time thus requires real time controller to count the encoder signal and update the current velocity of the motor Not like regular operating system in personal computer, Real Time Operating System (RTOS) is an event driven or interrupt driven operating system [ 21 ]. A micro controller is used to control the Omni directiona l wheel motor and the rest of devices that complete the autonomous vehicle system. The micro controller used for this project is ATmega128 from ATMEL. The ATmega128 controller is one of the 8 bit AVR micro controller series with 128K bytes of In System pro grammable flash memory [15] Computer S oftware and C ommunication A simple GUI program was developped using Visual C# to evaluate the omni drectional control system. This program opens serial ports to communicate with embedded wheel motor controller and th ree SmartMotor s. Figure 4 1 shows how the program looks. Two displays on top left side of the panel graphically show the input and current pose of the vehicle. Control panel below the displays is responsible for controlling the motor. With this control pan el, PID gains can be updated on line and can record the current velocity report of the wheel from the embedded controller.

PAGE 46

46 Figure 4 1 A c omputer software developed to control the vehicle A communication protocol is defined to make communication between the controller and PC. Following message sets are transmitted through a serial port at the rate of 19200 bytes per seconds. Every message has to be start with $ sign fallowed by a string type header message indicating what instruction is being transmitted. Each instruction may contains certain number of parameters and these parameters are separated by ) sign.

PAGE 47

47 Table 4 1 Communication protocol Category Size Type Description Start 1 byt e s tring Header 1~6 bytes s tring Data 0~20 bytes i nt Optional Include parameters passing to Parameters are separated by comma (,) Terminal 2 bytes h ex Indicates the end of message Every message must be finished wi \ r \ 0D 0A With the communication protocol defined PC to micro controller message is defined as following Table 4 2 Command message definitions from PC to the wheel controller Message Header Parameters Description $H ELLO n/a Establish serial communication $ MOTOR 1 or 0 1: Turn on Wheel motors fallowed by resetting the DAC 0: Turn off all motors $OFFM n/a Turn off all wheel motors $SMARTEN 1 or 0 1: SmartMotor enable 0: T urn off SmartMotor $SETS i nt, int, int Set m otor speed in RPM. $SETKP i nt, int, int Set Kp PID gain for each motor controller $SETKI i nt, int, int Set Ki PID gain for each motor controller $SETKD i nt, int, int Set Kd PID gain for each motor controller $REPORT 1 or 0 1: Request periodic reportin g of the speed of each motor 0: Stop reporting $STATUS int 1: Request PID gain inside the controller Controller to PC communication message is defined as following

PAGE 48

48 Table 4 3 Command message definitions from the wheel controll er to PC Message Parameters Description $HELLO n/a Report that the serial communication is successfully established $GAINUP int int, int, int, int, int, int, int, int, Report current PID gain setup to PC The first three parameters indicate they are Pr oportional gain. Second three parameters indicate they are Integrate gain. Last three parameters indicate they are Differential gain. $RPM int, int, int, Report the latest available RPM Control by Joystick Even though the vehicle will eventually be cont rolled by autonomous vehicle controller, manual control is important to demonstrate the controllability of the vehicle. Autonomous controller always has chances of failure Ability of manual control also relates to safety concern s of the autonomous vehicle Since the vehicle has omni directional platform with 3 DOF of mobility, manual control is more complex than conventional tricycle or buggy car type vehicle. Radio controlled cars usually take one steering and one throttle input to control the vehicle an d that is enough for most cases. However controlling omni directional vehicle requires more complex control inputs. An X box 360 control pad from Microsoft is selected for the manual controller. With two control sticks and triggers the Xbox control pad pro vides lots of features to the users. For the purpose of demonstrating the omni directional control, three different mode s of operation are proposed. Figure 4 2 shows how the Xbox control input represents the operation mode s

PAGE 49

49 Figure 4 2 Proposed modes of manual operation of the OmniGator F irst mode is translational motion. While this mode is selected, all of the three omni directional wheels are parallel to each other so that the vehicle moves with constant orientation. Left control stick represents the vector of the motion as the direction of the stick represents angle of wheels and magnitude represents throttle of the vehicle. The second mode is tricycle motion. While this mode is selected, two of three wheels are parallel to each other and the front wheel steers towards the path. Horizontal displacement of the right control stick determines the steering angle of the front wheel and vertical displacement controls throttle input of the vehicle. Also note that if the stick is pulled back, the vehicle moves backwards. The third and last mode is rotational motion. While this mode is selected, all three wheels are looking at the center of the vehicle making a complete circle and the vehicle is spinning

PAGE 50

50 about the center axis of the body Trigger input from the controller determines the rate of change of rotation.

PAGE 51

51 CHAPTER 5 TESTING AND RESULTS Methodology Fabrica tion of the OmniGator C ontrol B oard A pilot test board of the OmniGator is fabricated and tested. Figure 5 1 shows how the controller is laid out on the board. Note that the main 48V power supply and computer is not shown here. All components relate d to control ling the vehicle are laid on a single board. The entire package will eventually be inside of the body platform and th e housing is under construc tion at the writing of this thesis. Figure 5 1 Pilot board developed for the vehicle platform Data Acquisition Finding the best sets of PID gain requires repetitive trials and errors because differ ent gains affec the results. For the purpose of tuning the wheel motor, a windows program is

PAGE 52

52 developed using C#. The embedded motor controller updates the latest encoder counts per sampling time and report this to the computer at 33Hz through serial port. The program either can directly send a speed command or gradually increasing or decreasing speed command to simulate step input and ramp input respectively. It also can set PID gains of each motors on line basis. Figure 5 2 Software developed to tune the PID controller Test Using D ynamometer In order to test the wheel motor controller, a dynamometer is used to simulate the vehicle is on the ground. The roller used for the test is a solid steel mass with 4.9 inches in diameter and 10 inches in length. This roller weighs about 53.48 pounds or 24.26 kg. The large inertia of this roller well represents how the wheel motor controller will behave when the vehicle is on the ground. Figu re 5 3 shows configuration of the dynamometer wi th wheels attached to the vehicle. Although actual bench mark dynamometer test includes measuring torques while forces applied to the roller, such measuring devices were not available at the time of writing The micro controller with PID controller measure s the ve locity of the wheel and reports the RPM data to the

PAGE 53

53 host computer. While the motor is running, various types of external forces applied to the dynamometer and velocity response is monitored. Figure 5 3 Configurati on of the dynamometer used for the test Results Motor R esponse C omparison Before conducting the actual test for the tuning, it is important to know about the characteristics of each motors such as how they behave for certain input, how stable and etc. Diff erent windings of the armature or mechanical defect in gear train may affect the resulting performance of individual motor s For the purpose of this test, no external load is applied to the motors and the same set of gains for the controller were used Fig ure 5 4 shows the result from the first test. Each motor name refer s to the vehicle configuration mentioned in chapter 1

PAGE 54

54 Figure 5 4 Motor comparison test result The result clearly shows that one of three motor is responding differently from others T hus the controller should have different PID gain s set. Tuning the D igital PID C ontroller Another s eries of tests were conducted to find the best gain set for the PID controller. As described earlier, each gain has distinctive e ffect on the controller performance. In many applications of the PID control, where its dynamics are not well known or defined, the controller gains ( K P K I and K D ) must be determined experimentally [ 23 ]. Generally, f inding optimized PID gain s start s with increasing the proportional gain ( K P ) until the target system becomes unstable [2 4 ] Figure 5 5 shows how a proportional gain K P affects the response of the system. 50 0 50 100 150 200 3.0 5.0 7.0 9.0 11.0 13.0 15.0 Speed (RPM) Time (Sec) MOTOR_A MOTOR_B MOTOR_C

PAGE 55

55 During the actual test, it was observed that any K P larger than 40 makes the response unst able. Therefore K P =2 9 was selected for the wheel motor controller. A B C Figure 5 5 Step responses of the PID motor controller for different P gain A) K P = 45. B) K P =10. C) K P = 29.

PAGE 56

56 Although the found K P gain made the syst em to respond faster, certain steady state error was also observed. This may be resulted from internal motor friction or delaying update. It could be resolved by increasing K I gain of the controller. Since the K I gain multiplies the summation of all the pr evious errors, care must be taken when increasing it. Figure 5 6 shows resulting effect of increasing the K I gain. Some of the steady state errors were removed from the response. Figure 5 6 Step response of the motor with PI control In many PID motor control implementations, adjusting K P and K I gives acceptable results thus becoming PI controller. However some oscillations may occur during the rising and falling edge of the step input. Similar phenomenon is observed when tu ning the controller for motor C. Steady state error

PAGE 57

57 If such oscillations occur it can be reduced by increasing small amount of K D Figure 5 7 shows how K D gain affected the response of the motor C. Care must be taken when increasing K D gain as it may make the controller com pletely unstable. Figure 5 7 K D gain tuning for motor C Rest of the test s are relate to the synchronization of the motor. How all three motors well correspond to the input is very critical point for this kind of applicati on. Figure 5 7 shows how they behavior for the same input. The test conducted with gain set of Oscillation

PAGE 58

58 K PA = 2 5 .0, K PB = 5.0, K PC = 4 .0 K IA = 1, K IB = 4 K IC = 1 K DA = 0 K DB = 1 K DC = 5 While testing the motors, the motor B which wa s assigned as rear left motor, has been identified as having better mechanical assembly. This fact made the PID controller for the motor B can behavior similar to other motors with less K P gain. Figure 5 8 Step response of all three motors

PAGE 59

59 CHAPTER 6 CO NCLUSIONS AND FUTURE WORK Conclusions After months of hard working and testing, a n omni directional vehicle controller wa s designed and tested. During the development, making the embedded micro controller incorporated with motor amplifier was the most time consuming process. Even after the individual motor is successfully controlled with the control board, synchronizing it with other motors becomes a totally different problem. Especially in this case, the fact that the given motor is fully customized thus h ave no guaranteed specification escalates the difficulties. Noise induced from large amount of switching currents affected the entire control system and must be isolated. Finally a PID controller for the omni directional wheel is developed and tested. Alt hough t he PID controller still need s to be tuned at the time of writing this thesis, it successfully demonstrated the ability to control the omni directional wheel mechanism While the motor controller takes control of the throttling of the platform, thre e steering motors determine the direction of their respective wheel motor. Procedure for homing and alignment of the steering actuators also considered and tested. Non contact switch is used for homing and worked well. Future Work As the entire HMV resear ch project is still an ongoing project to build an intelligent ground vehicle, the omni directional vehicle controller has to be incorporated with higher level autonomous systems. Like other autonomous vehicles developed at CIMAR, JAUS architecture will be on top of the autonomous controller and it requires Primitive Drive (PD) for the vehicle control. In order to make the omni directional vehicle platform to be integrated in to the PD functionality, more research about control method has to be conducted. T he omni directional

PAGE 60

60 controller presented here shows its strength and potential in HMV system however it also brought number of complexity and uncertainty in controlling the vehicle. Kinematic analysis and precise dynamic modeling of the vehicle will help s olving such sophisticated control problems.

PAGE 61

61 APPENDIX A O MNIGATOR CONTROLLER BOARD DESIGN

PAGE 62

62 Figure A 1 Schematic diagram of the embedded motor control board

PAGE 63

63 LIST OF REFERENCES [1] C. Tobler, "Develop ment of an Autonomous Navigation Technology Test Vehicle," M.S. Thesis University of Florida, 2004 [2] C. Fulmer "The Design and Fabrication of an Omni Directional Vehicle Platform," M.S. Thesis, University of Florida, 2003. [3] J. Xiao, Mobile Robot Locomotio n Capstone Design Robotics, City Colle ge of New York (Last accessed 04/23/2009). [4] E. Nakano and N.Koyachi, An Advanced Mechanism of the Omni Directional Vehicle (ODV) and Its Application to the Working Wheelchair for the Disabled, in Proc of Int Con f on Advanced Robotics pp 277 284, 1983 [5] L. D. Stephen, D.L. Brett, Control of An Omni Directional Robotic Vehicle with Mecanum Wheels IEEE Telesystems Conf v ol.1., pp 323 328 1991 [6] K. Yasumasa, M. Takanori, T. Kazuhiko, Navigation Guidance Contr ol Using Haptic Feedback for Obstacle Avoidance of Omni directional Wheelchair IEEE Symposium on Haptic Interfaces for Virtual Environment s and Teleoperator Systems, pp. 437 444 2008 [7] N. S. Kuppuswamy, S. H. Cho, D. Stonier, S. L. Choi, J. H. Kim, D esi gn of an O mnidirectional R obot for FIRA ROBOSOT FIRA Robot World Congress 2006 [8] T. Kenjiro, T. Riichiro, Mechanical Design of Omni Ball :Spherical Wheel for Holonomic Omnidirectional Motion in Pro c. of the 3 rd Annual IEEE Conf on Automation Science and Engineering 2007 [9] W. Masayoshi, Omnidirectional and Holonomic Mobile Platform with Four Wheel Drive Mechanism for Wheelchairs Journal of Robotics and Mechatronics v ol. 19 n o.3, 2007 [10] I. Giovanni, Swedish Wheeled Omnidirectional Mobile Robots: Ki nematics Analysis and Control IEEE Transactions on Robotics v ol 25, n o. 1, February 2009 [11] T. Takahashi, E. Nakano, Y. Mori, and K. Takayama, A study on the mechanism and control of omni directional vehicle in Proc IROS 96, pp. 52 59, 1996 [12] W. Masayoshi, and M. Shunji, Holonomic and Omnidirectional Vehicle with Conventional Tires IEEE Int Conf on Robotics and Automation April 1996 [13] The 17TH Annual Intelligent Ground Vehicle Competition (IGVC), Official Competition Details, Rules and Forma t 2009.

PAGE 64

64 [14] M. Hashimotor, N. Suizu, I. Fujiwara, and F. Oba, Path Tracking Control of a Non Holonomic Modular Omnidirectional Vehicle IEEE Int Conf on Systems, Man, and Cybernetics v ol 6, pp. 637 642, 1999 [15] Atmel Corporation ATmega128L Summary Rev 2467Q 03/08 2008. [16] C. T. Kilian Modern Control Technology Thompson Delmar Learning, 3rd edition pp 433 488, 2006 [17] R. Rojas A. G. Forster Holonomic Control of a R obot with an O mnidirectional D rive. Kunstliche Intelligenz, BottcherIT Verlag, 2006. [18] K. Watanabe botic Systems, vol. 22, no. 3 4 Kluwer, 1998. [19] Advanced Motion Controls, AZ Analog Drives for Servo Systems Hardware Installa tion Manual, http://www.a m c.com/download/manual/mnalazin.pdf (Last accessed 04/24/2009). [20] D. E. Barnett, Encoder Electrical Interface http://www.optoresolver.com/help/tutorials/encoder_electrical_interface.htm (Last accessed 04/01/2009 ). [21] S. F. Barrett, D. J. Pack, Embedded Systems Design and Applications with the Pearson Prentice Hall,pp 205 251 2005 [22] Animatics, The SmartMotor Series 4 User s Guide Rev 5, 2008 [23] K. Ogata, Discrete Time Control Systems ,1 st ed, Prentice Hall ch 3, 4 1987. [24] T. Wescott, PID Without a PhD Embedded.com, http://www.embedded.com/2000/0010/0010feat3.htm (Last accessed 04/13/2009).

PAGE 65

65 BIOGRAPHICAL SKETCH Kyuhyong You was born on May 1981, in Se oul, the capital city of Korea. He received his Bachelor of Science degree in m echanical e ngineering from Hanyang University, Seoul, Korea in 2007 While an undergraduate, he organized a club HART with school mates to design and buil d a radio controlled air plane. H e began interested in robotics after taking m echatronics and m icro processor classes from the same school. Upon graduation h e enrolled as a graduate student at the University of Florida in the d epartment of m echanical and a erospace e ngineering. In 2007, he joined the Center for Intelligent Machines and Robotics (CIMAR) and started his research in robotics with intelligent ground vehicle. He received his Master of Science degree in m echanical e ngineering with minor in e lectrical e ngineering from the University of Florida, in August 2009. His research interests include control system and embedded controller for autonomous veh icle