A Compliance Control Strategy for an Upper-Limb Rehabilitation Robot with a Quick and Real-Time Human-Robot Interactive Force Estimation

Abstract

In order to improve the safety of rehabilitation training and estimate physical human-robot interaction(pHRI) force without force sensor, this paper designs a real-time interactive force learning controller in an End-Effector Type Robot. The adaptive sliding mode term in controller is used to solve the uncertainties in the robot modeling and inertial parameter measurements. The learning law term can estimate pHRI force. To investigate the performance of this controller, experiments were designed to repeat the tracking task on a circular trajectory. The result shows that this controller can make the tracking error and the estimate error of pHRI force gradually acquire a desired stability margin. The error of resultant force of pHRI between the estimating and the measuring by the force sensor is 2.87±2.50N. This controller enables the robot to exhibit high compliance and its estimation of the pHRI force takes less time and is more accurate.