EMG Prosthetic Hand Control based on the Torque Estimation

Abstract

EMG signal includes the information of muscle activities for the arm movements. It has been utilized to manipulate the prosthetic hand under the similar control feeling to the natural limb. The present paper proposes a direct torque control method for the prosthetic hand. The artificial neural network is used for the torque estimation. Also the learning system based on the feedback error learning schema is used for the neural network renewing. The desired angle and EMG signals are measured from the normal and amputed arm. 2-DOF motions, i.e. hand grasp and open and arm flexion and extension, are picked up. Then it is verified that the neural network can learn the relation between the EMG signal and joint torque.