On-line adaptive robust neural network tracking control for robot manipulators

Abstract

The robust tracking control for a class of robot manipulators with disturbance and uncertainties is considered. The controller consists of an adaptive radial basis function(RBF)neural network controller and a PD controller.The initial structure and parameters of RBF neural network are determined on-line by the growing-and-pruning(GP-RBF)algorithm based on the sensitivity of neurons as well as the winner neuron concept.When the errors meet certain requirements, the adaptive law based on the Lyapunov stability further adjusts the weights of networks to ensure the asymptotic convergence of the tracking error to be zero.The controller guarantees the stability and robustness of the system.Simulation results demonstrate the efficacy of this method.