An adaptive learning control scheme for manipulators

The author presents a simple adaptive learning control scheme for multi-joint robotic manipulators based on the independent joint adaptive control and iterative learning control concepts. The control scheme consists of an adaptive law and a learning law. The adaptive law is designed by the Lyapunov method; less knowledge about the manipulator dynamics is required, and only the position and velocity signals are used. The control objective is to achieve accurate tracking to the desired motion trajectory by trails. The convergence of the scheme is theoretically proved for a linear time-varying mechanical system, which is an appropriate representation of robot dynamics in the vicinity of the desired trajectory. The proposed control scheme is implemented on a super PC/XT computer for the motion control of a revolute-type rigid link robot MOVEMASTER-II (R501) with five degrees of freedom. The desired motion is given not only in terms of joint-angle coordinates but also in terms of task-oriented coordinates. The learning convergence speed and the robustness of the system are discussed.<>