Hand Pose Estimation Using EMG Signals

Abstract

In this paper, we report a hand pose estimation using electromyogram (EMG) signals consisting of two methods. One is a method of hand motion classification with a support vector machine (SVM). The other is a method of operator's joint angle estimation based on EMG-Joint angle models, which express the linear relationships between the EMG signals and joint angles. By incorporating the motion classification and joint angle estimation, it is not discrete but continuous hand pose estimation is achieved. To examine the effectiveness of our method, we performed experiments in which seven hand motions are estimated by our method with eight subjects. Experimental results show that motion classification was performed with high accuracy and that three joint angles were estimated well for subjects experienced in our methodology.