Decentralized Adaptive and Nonadaptive Position/Force Controllers for Redundant Manipulators in Cooperations

Abstract

This paper addresses the problem of controlling positions and forces of multiple redundant manipulators cooperatively handling an ob ject in a decentralized manner while optimizing a performance in dex. The authors propose one adaptive and two nonadaptive de centralized controllers. The adaptive controller is able to cope with unknown or uncertain physical parameters of the system. The Lya punov approach proves that this controller not only yields asymp totic convergence of the position and force errors but also opti mizes the specified performance index. To improve the transition performance, besides the nonadaptive version of the controller, an implicit decentralized force controller with exponential stability is developed. The controllers realize the force and position controls simultaneously by using coupled force and position signals. The coupled signals distinguish these methods from traditional hybrid controllers in which decoupling of the signals is a must. The de centralized architectures of all the controllers must be emphasized. In detail, the joint input for each robot is designed in its own state space; thus, no coordinator exists, and the computation cost is less expensive than that of centralized controllers designed in the state space of the whole system.