A hierarchical hand motions recognition method based on IMU and sEMG sensors

Human hand is crucial to the activity of daily living. To improve amputees' quality of life (QOL), it is imperative to propose a new method for the prosthesis hand to recognize hand motions. According to previous researches, multi-sensor can obtain more information with limited sensors for improving the classification accuracy rate. Therefore, this paper proposed a hierarchical hand motions recognition method based on one inertial measurement unit (IMU) sensor and two surface electromyography (sEMG) sensors. We dealt with two types of signals separately to classified six predefined hand motions in a hierarchical structure. The method utilized SVM classifier for EMG signals and a decision-tree classifier for IMU signals. To verify the feasibility of the hierarchical hand motion recognition method, ten subjects were asked to join the experiment. And experimental results show that proposed method is stable and the overall average accuracy rate is 95.6%. The combination of sEMG and IMU signals can be used for prosthetic user interface applications.