Impedance Control of Series Elastic Actuators in Exoskeleton Using Recurrent Neural Network

Abstract

Exoskeleton robot is a system with human-machine interaction and could help human to carry heavy load. Series elastic actuators (SEAs) in force-controlled robots can achieve compliant interactions with humans. In order to get flexible position and force control, this paper introduced impedance control into exoskeleton robot and use a recurrent neural network to improve the adaptive ability of the system. Finally, simulation results of force tracking and trajectory tracking are presented. By setting virtual impedance parameters, this controller can achieve stable track tracking and force tracking. Recurrent neural network can improve the performance of adaptability and robustness.