A practical EMG-driven musculoskeletal model for dynamic torque estimation of knee joint

Abstract

Multichannel electromyography (EMG) signals have been used as human-machine interface (HMI) to control robot systems and prostheses in recent years. EMG-based torque estimation is a widely research method to obtain motion intent. However, the existing torque models usually have the disadvantage of complexity for modeling or time consuming for model tuning. This paper presents a practical EMG-driven musculoskeletal model for the knee joint, which can estimate muscle force and active torque from EMG signals. The EMG-driven model consists of a muscle tendon model and a proposed musculoskeletal model. The muscle tendon model is used to calculate muscle force for each muscle group first. Then the forces are input to the musculoskeletal model to estimate the active joint torque. The dual population genetic algorithm (DPGA) is applied to optimize the model parameters. This tuning process takes only a few minutes and can reduce risk of fallen into local minimum. The ability to accurately predict the active torque of knee joint with relatively low root-mean-square error (RMSE) demonstrates the proposed EMG-driven model has potential applications towards the development of HMI.