An approach to adaptive decentralized control of flexible joint robots

This paper presents an approach to adaptive decentralized control for flexible joint robots based on an adaptive observer technique. The joint dynamics is decomposed into a linear term plus nonlinearities, where the parameters are unknown. Under the assumption, that the variation of nonlinearities is slower than the adaptation speed of the observer, the system states and parameters will be reconstructed by the adaptive observer. A two-level adaptive control scheme is used to guarantee the trajectory tracking and the stability of the control system. Simulation results of the control on a realistic robot model are given to show the performance of the proposed control system