Control of Hand Prostheses using EMG based Pattern Recognition: A Review of Existing Classification Techniques and Future Implementation

Abstract

Electromyography (EMG) signal is a type of biological signal that comes from the activity of nerves and muscles. Most of the time, an EMG device is used to record the EMG data. These signs are used to check for medical problems, find out how active an animal is, and figure out how it moves. Hand prosthesis is a device which used to replace the missing part of human parts. It enables amputees to perform their daily activity without anyone supports. This research provides a quick overview on how EMG classification methods work for hand prosthesis. Finally, this study found that 80% to 99% of the times, the present technologies are good enough to pick up and classify the EMG signals. Once the signal is recorded, the accuracy of recognition is based on the classification algorithms used. Signals are then decoded for their intended use, such as moving a robotic arm, recognizing speech, analyzing gait, etc. The accuracy of decoding the EMG signal has already gone above 99%, and it will get even better with the new deep learning technology. This review study helps to understand how well new methods are suitable for classifying EMG signals.