MODELING AND SIMULATION OF VISUAL SERVO CONTROL OF ROBOT MANIPULATORS DURING TRACKING

Abstract

This thesis is concerned with mathematical modeling and simulation of visual servo control of robot manipulators in certain target tracking and adaptive path control problems. We mainly consider aspects pertain to design of Artificial Neural Network controller for tracking a moving object using stereo vision. In chapter 1, a general introduction and brief review of literature is given. Chapter 2 contains some necessary concepts, definitions and results from robotics, visual servoing and artificial neural network that will be used in the subsequent chapters. Chapter 3 deals with the problem of tracking a moving polyhedral/cylindrical object using image based visual feedback control. Image based tracking algorithm are based on the optical flow equations which is given by X = J(X) r where X , X and r denote the vector containing error in image features, velocity of image features and end-effector velocity relative to the target, respectively. In conventional tracking algorithm the knowledge of the inverse (pseudo inverse) of J(X) at every instant of time is necessary to compute a feedback control, namely the end-effector velocity r . In this chapter, we introduce two types of controller design. One is based on linear feedback and other based on Artificial Neural Network. One feature point is used from each of the camera image in the control design. The use of minimum number of feature points ensures faster convergence of the algorithm.