Multiple Model-Based Robot Control: Development and Initial Evaluation

AA-1 'APprove(I for public release; distribution unlimited-- l • I II i II il I Preface The purpose of this thesis was to develop and evaluate a new adaptive robot control technique. The approach included the use of a Multiple Model Adaptive Estimator (MMAE) to determine unknown parameters needed for robot tracking and a PD feedback loop to reject disturbances. There are presently many esti-niation techniques used for parameter identification in robot control. Before a preferred approach can be established, the range of possible identification schemes must be expanded and experimentally verified. The MMAE was combined with a model-based description of the robot. Model-based control is a mature control algorithm that has been shown to produce superior tracking performance when the payload in known. The MMAE was used to provide the niodel-based control algorithm with an estimate of the mass of the payload. Simulation and experimentation on the PUMA-560 clearly demonstrate