A New Procedure for Tuning an Allocator and Designing a Robust High-Level Control Law for Over-Actuated Systems.

Abstract

This paper presents a new integrated procedure to tune a control law for over-actuated mechanical systems that may encounter singularities. First, the allocator that divides the commands among the actuators is tuned thanks to a genetic optimization algorithm, that computes the optimal values of its parameters. Then, the open-loop system including the allocator is identified and a robust closed-loop controller is computed with the structured H∞ method. Indeed, near singularities, the system and the allocator may create errors to deviate from these points or create delays to reconfigure the actuators, hence there is a need to create a closed-loop controller robust to these characteristics and to parameter variations. This procedure is carried out on a planar redundant robotic manipulator example. Simulations of the behaviours of the open-loop system with the allocator and in a closed loop are presented.