Robust learning control for robot manipulators based on disturbance observer

Abstract

A robust learning control algorithm based on a disturbance observer is proposed. The robust controller which is based on a disturbance observer compensates disturbances due to parameter variations, mechanical nonlinearities, unmodeled dynamics and external disturbances. In doing so, it provides robustness and makes the whole system stable, and makes easy to design learning controllers which acquire good performance. A novel iterative learning control scheme comprising a unique feedforward learning controller and a robust controller is presented. In this paper, two kinds of learning controller are designed: firstly, the learning controller solves the zero initial problem; secondly, it solves the nonzero initial position problem. The convergence and robustness of the proposed controller is proved by a method based on Lyapunov stability theorem. The results of numerical simulation are shown to verify the effectiveness of the proposed control scheme.