PERFORMANCE IN ADAPTIVE MANIPULATOR CONTROL

Effective adaptive controller designs potentially combine high speed and high precision in robot manipulation and furthermore can considerably simplify high-level programming by providing consistent performance in the face of large variations in loads or tasks. A simple, globally tracking-convergent, direct adaptive manipulator controller has previously been developed and experimentally demonstrated. In this article, we further explore the performance issues linked to a computationally effective implementation. Specifically, we develop a recursive implementation applicable to both open and closed kinematic chains, as well as rules for obtaining minimal parametrizations. We also discuss implementations of the algorithm directly in Cartesian space, the exploitation of kinematic redundancies, and applications to adaptive compliant motion control. These developments are illustrated experimentally on a four-degrees-of-freedom articulated robot arm and suggest a wide range of application well beyond adaptation to grasped loads.