A Patient-Specific EMG-Driven Musculoskeletal Model for Improving the Effectiveness of Robotic Neurorehabilitation

摘要

An EMG-driven musculoskeletal model for controlling the human-inspired robotic neurorehabilitation is proposed in this paper. This model is built upon the state-of-the-art computer generated musculoskeletal framework which provides patient-specific muscular-tendon physiological, muscular-tendon kinematics parameters. Muscle forces and joint moment during locomotion are predicted through activation dynamics and contraction dynamics based on the hill-type muscle mechanics model. A hybrid Simulink-M simulated anneal algorithm is used for parameters optimization. The preliminary result showed that based on only a few EMG channels, the proposed model could efficiently predict joint moment and muscle forces. The proposed model has the potential to control the rehabilitation robot based only on a few of EMG channels from extensor and flexor muscle.