An sEMG-Driven Neuromusculoskeletal Model of Upper Limb for Rehabilitation Robot Control

This study proposes a new method for modeling the complicated dynamics between motor neural signal (surface electromyography, sEMG) and corresponding joint torque of muscle contraction (sEMG-driven neuromusculoskeletal model), which has potential to be used for rehabilitation robot control and neuromuscular evaluation after stroke, etc. In this model, muscle activation dynamics and contraction dynamics are built based on Hill-type muscle model, which has many parameters to be determined using optimization methods, and training samples of sEMG, joint angles, and joint torques are acquired with the aid of an upper-limb rehabilitation robot. Subject-specific parameters are initialized with scaled Standford VA model data by subjects' weight, limb length, etc., and the model is optimized using the genetic algorithm (GA). Based on this study, characteristics of a certain single muscle during voluntary movements can be obtained by measuring their sEMG signals, and the motor commands of brain are decoded in some degree.