A modified model reference adaptive control scheme for rigid robots

This paper seeks to improve the zero-state performance of a standard model reference adaptive control scheme for rigid robots by augmenting the usual model reference control law with an auxiliary component synthesized from the tracking error. A condition involving a certain design parameter is given which can be used to decide a priori whether the modified scheme can potentially outperform its unmodified counterpart. If the joint accelerations are available for measurement, then it is shown that this condition can always be satisfied and the zero-state performance of the modified scheme can in fact be arbitrarily improved.