A New Approach for Optimal Control of Multiple-Arm Robotic Systems

Optimal control frameworks are very important in optimization of multiple-arm robotic systems, although complexity, non-linearity and having large scales usually cause some computational problems in the capacity and time needed. In this paper, it is shown how by conducting the study of multiple-arm robotic systems, within a decomposition-coordination framework, the overall optimum can be achieved in only a few iterations. By using this methodology, the overall problem is considered as optimization of a two-level large-scale system. So with the aim of optimization, the problem is first decomposed into m sub-problems at the first level, where each sub-problem is solved using a typical gradient optimization method. Then by using the proposed methodology, which is based on the gradient of the interaction prediction errors, at the second level, the coordination of the overall system is done.