Enhancing upper-limb rehabilitation robot: autonomous self-adaptive resistance generation using EMG-based Fuzzy-PI control

Abstract

ABSTRACT Upper-limb rehabilitation is a technique for improving muscle development and nervous function in stroke patients through repetitive rhythmic hand movements. However, conventional robot-assisted therapy cannot generate the natural resistant responses seen in therapy administered by therapists, leading to critical limitations in its effectiveness compared to conventional therapy. To overcome this limitation, we propose a novel strategy enabling a rehabilitation robot to generate self-adaptive resistance using electromyography (EMG) autonomously signals from the patient. A position-based force control system based on Fuzzy-PI (proportional and integral) control was successfully implemented in robot-assisted upper-limb rehabilitation. Performance evaluation comprised two primary aspects: examining rehabilitation durations and assessing movement smoothness through jerk values. Experimental outcomes revealed that the self-adaptive control (SAC) strategy, which automatically adjusts resistance, outperforms the non-adaptive control (NAC) approach. SAC demonstrated notable superiority in movement smoothness and extended the duration of upper-limb rehabilitation tasks nearly twofold compared to NAC. Furthermore, applying the jerk concept, and assessing motion smoothness, indicated that SAC exhibited more controlled and seamless movements than NAC. The alignment between participant surveys and the outcomes of parallel tests confirms the validity of the study's conclusions. Thus, this research makes a significant contribution towards improving the effectiveness of robot-assisted therapy in stroke rehabilitation.