Neural Network Control for Tele-rehabilitation Robot based on Variable Gain

Abstract

The patient's muscle spasm can cause the tele- rehabilitation robot system with force feedback to become instable and result in the slave unsmoothness of movement, which make the physical rehabilitative exercise inefficient during the tele-rehabilitative training. In order to guarantee the stability and reduce its fluctuation of the speed, a new method based on variable gain with back propagation neural network control was brought forward. With neural network adapting the control gains, not only the stability was guaranteed, but also the slave speed unsmoothness was lessened. The system can be used by sorts of rehabilitants and exhibits strong robustness. The simulation results demonstrate that this method is much more stable and smooth than that of the traditional control.