A Robust Neural Predictive Control Approach for Robotic Manipulators with Online Learning Ability

In this paper, we develop a new predictive controller for tracking control problems of robotic manipulators. Internal dynamics of the robotic model are first modeled using proper neural networks under support of an output feedback control signal. A new model predictive control signal is next derived to realize the control objective in a robust manner. Novel adaptation laws are then proposed to activate the network learning in an effective way. Effectiveness of the proposed controller has been validated throughout intensive simulation results on two degree of freedom robot.