Stable adaptive controller using MLP neural networks for rigid robot manipulators

摘要

In this article an adaptive controller based on neural networks is proposed to achieve output trajectory tracking of rigid robot manipulators. The neural network used is of the MLP (multi-layer perceptron) type with one hidden layer. Our method uses a decomposed connectionist structure. Each neural network approximates a separate element of the dynamical model. These approximations are used to perform an adaptive stable control law. The controller is based on direct adaptive techniques, and the Lyapunov approach is used to derive the adaptation laws of the nets' parameters. A compensation term is added to make up for the modelization error inherent to neural network approximation. The performances of the proposed adaptive neural network based controller are tested in simulation for a 2-DOF PUMA robot.