Biomimetic myoelectric control of a dexterous prosthetic hand

A new biomimetic two-input EMG based control algorithm for a dexterous robotic hand is presented. The biomimetic controller maps the EMG control signals to the prosthesis in a more physiologically expected manner. By exploiting the muscular structure of the forearm, the controller permits control of both power and lateral pinch grasps in addition to a typing function with the index finger. Proportional control of wrist flexion and extension is also possible while simultaneously controlling the applied grip force. Experimental results show that there exists a correlation between hand position and EMG signals, which is used as a basis for the biomimetic control algorithm. The control algorithm was evaluated by four human test subjects who performed five different tasks. Results indicate a general trend of improvement using the controller with increased training time.