Adaptive Control for Robotic Manipulators base on RBF Neural Network

Abstract

An adaptive neural network controller is brought forward by the paper to solve trajectory tracking problems of robotic manipulators with uncertainties.  The  first  scheme consists of  a PD feedback  and  a  dynamic  compensator  which is  composed by  neural  network controller and  variable  structure controller.  Neutral network controller is designed to adaptive learn and compensate the unknown uncertainties, variable   structure controller is designed to eliminate approach errors of neutral network. The adaptive weight learning algorithm of neural network is designed to ensure online real-time adjustment, offline learning phase is not need; Global asymptotic stability (GAS) of system base on Lyapunov theory is analysised to ensure the convergence of the algorithm. The simulation results show that the kind of the control scheme is effective and has good robustness.