The Development of a Wheelchair Control Method Based on sEMG Signals

Abstract

This paper proposed a control method of the electric wheelchair based on surface electromyography (sEMG) signals. In this method, a mapping between hand motions and control commands was established. When a certain kind of hand motion was recognized from sEMG signals, corresponding control would be applied in the wheelchair. The sEMG signals was as raw material for the pattern recognition type of classifier, which promoted the accuracy rate and robustness. The fusion features of Autoregressive (AR) model coefficient and root mean square ratio (RMSR) were used as features of data of hand motions. Support vector machine (SVM) as one of state-of-the-art supervised learning models, was used as classifier. Furthermore, comprehensive real-time simulation and control experiment were implemented. The accuracy rate of hand motions recognition in real-time reached 95% and the success rate of control experiment was up to 88%, which showed the proposed method was feasible and practical.