Study on Motion Recognition for a Hand Rehabilitation Robot Based on sEMG Signals

Abstract

Robot-assisted self-rehabilitation training can overcome the shortcomings of clinical rehabilitation, which is significant for stroke patients. This paper proposed a hand robot rehabilitation training system based on surface electromyography (sEMG) signals. This study combines pattern recognition technology and biomedical technology to collect the sEMG signals of the daily movements of the stroke patients, and then pre-processes signals, feature extraction and pattern recognition classification, and recognizes the movement intention of the patients. Through the hand rehabilitation robot to help the affected side of the hand for rehabilitation training. In the online self-rehabilitation training, a certain motion mode performed by the contralateral side of the patient is obtained according to the pattern recognition offline trained classifier to obtain the classification result, and the motor-controlled hand robot is driven to perform the corresponding motion to realize the self-rehabilitation training. Finally, simulation experiments and volunteer rehabilitation training experiments were carried out. The experimental results show that the recognition rate of pattern recognition reaches 93.70%±2.22%. Compared with BP neural network in other literatures, our proposed wavelet neural network is better than 8% in the accuracy of recognition classification. This system we designed can effectively help patients with self-rehabilitation training.