An adaptive control scheme for robots with unknown dynamics

In this paper, a stable adaptive control scheme for robot manipulators with unknown dynamics is proposed. It consists of an off-line least-mean-square (LMS) type identifier to identify structured system dynamics and an online dynamic compensator to compensating for dynamic uncertainties. Taking advantage of the unique structure of the robot regressor dynamics, the former uses an LMS type algorithm to identify, using a set of trial data, the structured dynamic parameters of the robot while the latter uses an online stable parameter updating mechanism determined using Lyapunov theory to compensate for both unknown and uncertain dynamics. The off-line identified parameters we used as initial values for the online dynamic parameter estimation. Since both identifier and compensator are implemented using the regressor dynamics, the recursive formula, for the computation of robot regressor dynamics previously proposed can be used to achieve high computational efficiency in real-time implementations. An illustrative simulation example is included to show the proposed adaptive control algorithm