A five-fingered underactuated prosthetic hand controlled by surface EMG (electromyographic) signals is presented in this paper. The prosthetic hand was designed with simplicity, lightweight and dexterity on the requirement of anthropomorphic hands. Underactuated self-adaptive theory was adopted to decrease the number of motors and weight. The control part of the prosthetic hand was based on a surface EMG motion pattern classifier which combines LM-based (Levenberg-Marquardt) neural network with the parametric AR (autoregressive) model. This motion pattern classifier can successfully identify the flexions of the thumb, the index finger and the middle finger by measuring the surface EMG signals through two electrodes mounted on the flexor digitorum profundus and flexor pollicis longus. Furthermore, via continuously controlling a single finger's motion, the five-fingered underactuated prosthetic hand can achieve more prehensile postures such as power grasp, centralized grip, fingertip grasp, cylindrical grasp, etc. The experimental results show that the classifier has a great potential application to the control of bionic man-machine systems because of its fast learning speed, high recognition capability and strong robustness.
