Neural network output feedback control for uncertain robot

Abstract

A neural network output feedback control scheme is proposed for the trajectory tracking of uncertain robot manipulators without measuring joint velocities. First, a nonlinear sliding observer based on a neural network is presented to estimate the joint velocities. Then, an observer-based neural network output feedback controller is proposed. The overall closed-loop system composed of a robot, an observer, and a controller is shown to be robust. Besides, it has been proven that both velocity estimation errors and trajectory tracking errors asymptotically tend to zero, and the estimation errors of neural network weights are uniformly bounded.