Artificial neural networks for identification in real time of the robot manipulator model parameters

Abstract

In this paper, the manipulator identification process was presented. To identify single-layer neural network with sigmoidal functions that describe individual neurons was used. The main goal was the approximation nonlinearities of manipulator model in real time. It was assumed that the nonlinearity of the manipulator are unknown. The stability of the identification system adopted by the law of the learning network weights generated based on Lyapunov stability theory.