Iterative Learning Control for Trajectory Tracking of Robot Manipulators

Abstract

Iterative learning control (ILC) has been shown to be effective in improving tracking performance of repetitive tasks, and is widely used in the motion control systems of CNC machines, semiconductor manufacturing equipment, hard disk drives, etc. However, applying ILC to robot manipulators requires careful consideration of nonlinear dynamics. We propose using a computed torque controller and the disturbance observer (DOB) to robustly linearize the dynamics of robot manipulators. The PD feedback controller is then applied for each joint to achieve the desired bandwidth and damping ratio. Both control-based ILC and command-based ILC are implemented separately in the linearized system as a feedforward compensator to enhance trajectory tracking accuracy. The proposed control system is realized in a six-axis industrial robot. Experimental results show that DOB is indispensable for robust feedback linearization so that ILC can work on the linearized system to improve the tracking performance for repetitive motion. Satisfactory and similar performance is accomplished by both control-based ILC and command-based ILC.