Proportional sEMG Based Robotic Assistance in an Isolated Wrist Movement

Abstract

Numerous studies have demonstrated the efficacy of robots for motor therapy. Surface electromyography (sEMG) is appealing for user intent detection as the signal relates the individual’s desired muscle contractile force. A drawback to sEMG interfaces is subject- and session-dependent calibration. We sought to investigate the effect of a simple sEMG assistive controller on user performance in the MAHI Exo-II, therapeutic exoskeleton. Agonist-antagonist muscles were related after normalization based on sub-maximal isometric contraction. Six subjects performed a target tracking task with four levels of assistance in wrist flexion-extension. Performance metrics were mean absolute position error and estimated muscular activity. In low levels of assistance, subject performance was not significantly affected, while increasing the assistance resulted in higher position errors. In characterizing the performance assistance trade-off, we better understand the capabilities of this simple controller. This investigation validates the feasibility of using a proportional control scheme for a therapeutic wrist exoskeleton system and motivates further testing with impaired subjects to optimize the system for use in a clinical setting.