A deadbeat trajectory controller for nonlinear systems via neural network: a robot case study

Abstract

A method for trajectory control of nonlinear systems is presented. Nonlinear feedback is used in order to obtain a linear model, and the nonlinearities and uncertainties remain are modelled by a multilayer neural network. The tracking error between the nonlinear system output and a desired trajectory is attempted to be removed by a deadbeat controller. The algorithm is applied to the trajectory control of a robot arm. >