Minimal Resource Allocating Networks for Discrete Time Sliding Mode Control of Robotic Manipulators

Abstract

This paper presents a discrete-time sliding mode control based on neural networks designed for robotic manipulators. Radial basis function neural networks are used to learn about uncertainties affecting the system. The online learning algorithm combines the growing criterion and the pruning strategy of the minimal resource allocating network technique with an adaptive extended Kalman filter to update all the parameters of the networks. A method to improve the run-time performance for the real-time implementation of the learning algorithm has been considered. The analysis of the control stability is given and the controller is evaluated on the ERICC robot arm. Experiments show that the proposed controller produces good trajectory tracking performance and it is robust in the presence of model inaccuracies, disturbances and payload perturbations.