DEVELOPING A NEUROCOMPENSATOR FOR THE ADAPTIVE-CONTROL OF ROBOTS

A neural-network compensator is developed for the adaptive control of robot manipulators. The proposed compensator is implemented using the adaptive-linear-combiner algorithm with a special learning rule derive d based on the Lyapunov method. Both system stability and error conver gence can be guaranteed. The resulting controller has an implementatio n advantage in that the adaptation part of the control structure is in dependent of the feedforward part of the same control algorithm and mu ltirate sampling for the whole control system can therefore be applied . Simulation studies on a single-link manipulator show that the adapti ve control system incorporated with the neurocompensator maintains a v ery good trajectory tracking performance even in the presence of large parameter uncertainties and external disturbance. The satisfactory co ntrol performance of this approach is also demonstrated by experimenta l results.