Abstract: Bilateral rehabilitation systems with bilateral or unilateral assistive robots have been developed for hemiplegia patients to recover their one-side paralysis. However, the compliant robotic assistance to promote bilateral inter-limb coordination remains a challenge that should be addressed. In this paper, a biomimetic variable stiffness modulation strategy for the Variable Stiffness Actuator (VSA) integrated robotic is proposed to improve bilateral limb coordination and promote bilateral motor skills relearning. An Electromyography (EMG)-driven synergy reference stiffness estimation model of the upper limb elbow joint is developed to reproduce the muscle synergy effect on the affected side limb by independent real-time stiffness control. Additionally, the bilateral impedance control is incorporated for realizing compliant patient–robot interaction. Preliminary experiments were carried out to evaluate the tracking performance and investigate the multiple task intensities’ influence on bilateral motor skills relearning. Experimental results evidence the proposed method could enable bilateral motor task skills relearning with wide-range task intensities and further promote bilateral inter-limb coordination.

Key words: Biomimetic stiffness modulation , · Compliant physical human–robot interaction (pHRI) , · Electromyography (EMG) , · Variable stiffness actuator (VSA) , · Rehabilitation robotics , · Synergy-based control , · Skill relearning