Robust Adaptive Wavelet Network Control for Robot Manipulators

Abstract

A novel robust adaptive wavelet network controller (RAWNC) is proposed for trajectory tracking of robot manipulator with uncertainties and disturbances. The controller is composed of an approximate ideal controller based on the adaptive wavelet network and a robust part. The adaptive wavelet network controller (AWNC) with finite wavelet family is designed to approximate the ideal controller which assumes the uncertainties and disturbances are known exactly. The parameters of the wavelet network are updated online motivated from the sliding mode control. The robust controller is proposed to compensate the errors between the AWNC and the ideal controller. It attenuates the uncertainties and disturbances in the sense of the Lyapunov function with guaranteed Hinfin tracking performance. The simulations for a two-joint robot manipulator demonstrate its robustness with satisfactory performance and approximate quality.