MOTION ANALYSIS, TRACKING CONTROLLER DESIGN AND SIMULATION FOR STATIONARY AND MOBILE ROBOTS

Abstract

This thesis is concerned with, kinematics/dynamics modeling, simulation and controller design for wheeled mobile robots in target tracking and path following problems, new method for computation of coriolis and centripetal forces of n-link rigid robot manipulator, with illustration of a five link robot manipulator dynamics. The thesis is also concerned with vision based control of mobile robots and use of kalman filter in visual servoing. In chapter *1, a general introduction and review of literature is given.In chapter 2,some necessary concepts, definitions and results from robotics, design of controllers, kalman filters and visual servoing which will be used in subsequent chapters is given. In chapter 3, the kinematic, dynamic analysis and a tracking control for four wheeled mobile robot is given. Non-holonomic constraints are derived using the condition of pure rolling and no side slipping. There are seven nonholonomic constraints. The kinematic model is converted into chained form using suitable change of coordinates. The trajectory planning is done for the platform centre of gravity (x0, yo) . Using the Euler-Lagrange's dynamic equation of motion the dynamics of the four wheeled mobile robot is presented and a controller is designed using the method of backstepping and Lyapunov stability theory so that the error between actual and desired states converges to zero. Simulation results are presented. The chapter 4, deals with motion planning for nonholonomic mobile manipulators. Trajectory tracking control and stabilization for a four wheeled nonholonomic mobile robot with a manipulator mounted on it is given. The role of the mobile platform is to bring the configuration of the manipulator into a preferred operating region in the workspace. When the manipulator is currently at configuration outside of the preferred operating region, the mobile platform should be planned and controlled so that the configuration of the manipulator returns to preferred operating region. There are two control inputs to the moble robot. In the first step assuming that the stabilization has been established for the first control input control law is designed to use second control input for stabilization. The manipulator motion are described and simulation results for mobile platform are illustrated using a circular trajectory. ii In chapter 5, A sensor based controller to drive a mobile robot visually towards a target is presented. This formulation can be applied to manipulators as well as to nonholonomic mobile robots provided that, the camera is able to move independently from the base. The visual servoing method used here requires that image features remain always in the field of view of the camera and they are never occluded during the whole execution of the task. The simulation result and the control strategy are illustrated. In chapter 6, we give the dynamic analysis of a n-link rigid robot manipulator. The general form of dynamics equation for any robot manipulator can be written as M(q):4 +C(q,4)4 + G(q) = r Where q E R" is joint angle vector, r e R" is torque vector, M(q) E Kix" is the inertia matrix C(q,4) E R" is coriolis and centripetal torque vector, G (q) e R" is gravitational torque. For typical manipulators the generation of manipulator dynamic equation is lengthy and tedious process. In this chapter a new formula for computing the corioles and centripetal terms of a n-link rigid robot manipulator is developed from the above Euler-Lagrange equation. This formula can be used for any open chain n-link rigid robot manipulator when the dynamic analysis is performed with Euler— Lagrange equation. The procedure is illustrated with the dynamic equation of the five link rigid robot manipulator. In chapter 7, a kalman filter approach for visual tracking is discussed. The kalman filter provides a recursive procedure for calculating the estimates of a state vector. The relation between the ego coordinates and the camera coordinates are given. A tilt camera fixed on the mobile robot with respect to horizon is considered. Using Kalman filter, position of the moving point on the ground, Kalman gain and covariance matrices are computed at each iteration and numerical table containing the measured coordinates, predicted feature points, updated feature points and their images are given. Simulation result are presented.