Principal Base Parameter Analysis: Implementation and Analysis in an AdaptiveModel-Based Robotic Controller

摘要

Principal Base Parameter Analysis (PBPA) is a general and systematic procedurefor determining the dynamic parameters that directly contribute to the joint torques of a manipulator, ranked in order of sensitivity. The feasibility of employing PBPA as an aid in the design and tuning of adaptive model-based controllers for industrial manipulators is rigorously investigated. This is accomplished by employing PBPA to determine the minimal size of the adaptive parameter vector and more importantly, to develop a less heuristic procedure for controller tuning. A simple, step-by-step procedure is developed wherein the manipulator torque equations are used in conjunction with PBPA to develop a functional adaptive model-based control (AMBC) algorithm, then tune the algorithm for optimal performance. Experimental analysis contrasts this adaptive model-based controller, designed and tuned using PBPA, to the completely heuristic procedure employed in previous Air Force Institute of Technology research. The incorporation of PBPA into the AMBC design methodology reduces the time and expertise necessary to tune the controller for satisfactory tracking performance.