Experimental evaluation of a PUMA manipulator model-referenced adaptive controller

Abstract

The robotic manipulator control performance database has been expanded by experimental evaluation of a joint space model-reference adaptive control technique. The trajectory tracking accuracy of the evaluated Model-Reference Adaptive Control technique was insufficient for gross motion control of a six degree of freedom PUMA manipulator. The adaption mechanism did not force the manipulator joints to track the reference model output. The Model-Referenced Adaptive Controller demonstrated efficacy inferior to dynamics based control techniques. Velocity reference input and friction compensation have been identified as essential components of any adaptive PUMA controller. Evaluation results provide valuable insight into further development of adaptive manipulator control methods.