MATHEMATICAL MODELING, CONTROL AND RELIABILITY ANALYSIS OF ROBOT MANIPULATORS

Abstract

This thesis concerns mathematical modeling, control and reliability analysis of robot manipulators. The neural network (NN) based adaptive control schemes are de-'eloped for coordinated multiple manipulator system, hybrid force/ position and nonlinear trajectory tracking for redundant robot manipulators. The controllers are designed and analyzed for convergence and stability using Lyapunov theory. The performance of the controllers is evaluated via numerical simulations for different robot manipulators. A distributed parameter dynamic model of a single flexible link robot manipulator is developed and a new feedback trajectory control scheme is presented and demonstrated through simulation experiments. Finally, reliability analysis of a complex multi-robot system has been performed. The present thesis is divided into eight chapters. The chapterwise description is given below. Chapter 1 is introductory in nature. It gives a review of earlier work in the field. A summary of the thesis is also given. Chapter 2 deals with basics and preliminaries which are used in subsequent chapters. Chapter 3 contains a NN based nonlinear tracking control of kinematically re-dundant robot manipulators. The NN controller achieves end effector trajectory tracking as well 'Ls sub-task tracking effectively. A neural network is"employed to learn the existing unknown dynamics of redundant robot manipulator, which 11 requires no preliminary learning. The stability of the system is proved using Lya-punov function, generated by weighting matrices_ Finally simulation is carried out for a 3R planar manipulator to illustrate the control methodology. The simulation results show that the feedforward neural network with the on-line updating law can compensate the full robot dynamics effectively. Chapter 4 deals with a neural network based adaptive control scheme for hybrid force/ position control for rigid robot manipulators. Firstly the robot dynamics is decomposed into force, position and redundant joint subspaces. Based on this de-composition, a controller is proposed that achieves -desired interaction force between the end-effector and the environment as well as regulates robot tip position in carte-sian space. A neural network is employed to learn the existing unknown dynamics of redundant robot manipulator, which requires no preliminary learning. The stability of the system is proved using Lyapunov function, generated by weighting matrices. Finally, to illustrate the control methodology, simulation is carried out for a two link rigid robot manipulator. Chapter 5, contains the coordinated motion control of a multiple robot manipu-lator system carrying a common object. Firstly, an integrated dynamic model of the manipulators and the object is derived in terms of object position and orientation as states of the derived model. Based on this model a controller is proposed that achieves the asymptotic convergence of the trajectory tracking for the object as well as tracking of the desir€d internal forces arising in the system. A neural netwbrk is employed to learn the existing uncertainties in the manipulators and object dy-namics. Finally, numerical simulation studies are carried for two three-link planar manipulators moving a circular disc on specified trajectory. In Chapter 6, based on distributed parameter dynamic model of a single flexible link robot manipulator, a new feedback trajectory control law has been presented. The control scheriie is stable throughout the entire trajectory tracking: In this scheme the drawbacks of truncated models are avoided and satisfactory results are In achieved. Chapter 7, is based on a multi-robotic system, with coordinated operations, in which two robots are used: operating independently and in coordinated manner. Moreover, the two robots are cooperated and synchronized independently with a conveyer unit. To represent the asynchronous and concurrent processing system, a Petri net (PN) model is applied. To enhance the relevance of the reliability study, fuzzy numbers are developed from available data on components and using fuzzy possibility theory to define membership functions. The use of fuzzy arithmetic in the PN model increases the flexibility for application to various systems and condi-tions. Various reliability parameters (such as MTBF, ENOF, reliability, availability etc.), are computed using Fuzzy Lambda-Tad methodology. As the available data is imprecise, incomplete, vague and conflicting, the fuzzy methodology can deal easily with approximations. Finally, the conclusions and future extensions of the work are presented in Chap-ter 8.