A Prosthetic Arm Based On Emg Pattern Recognition

Although myoelectric prosthesis has been researched for almost 60 years, a high quality prosthetic arm with dexterous hand manipulation and stable control system is always hard to find and amputee acceptance remains low. One of the major challenges is the lack of a portable and powerful embedded system to implement the electromyography (EMG) pattern recognition (PR) algorithms, other challenges include design of dexterous prosthetic hands, the placement of multichannel electrodes, trade-off between low power consumption and small size and so on. This paper presents our latest progress on EMG PR controlled prosthetic hand. In the system, an armband with 8-channel electrodes is used to acquire the EMG signals. A powerful embedded system which deals with the EMG signals decoding algorithms is introduced. On board EMG training and real-time surface myoelectric signals decoding is implemented in the embedded system to control a 6 degree of freedoms (DOFs) prosthetic hand. The promising results of this work show its potential of speeding up the translation of PR prosthetic arm into daily application.