EMG signal-based gait phase recognition using a GPES library and ISMF

This paper presents a gait phase recognition method using an electromyogram (EMG) with a gait phase EMG signal (GPES) library and an integrated spectral matching filter (ISMF). Existing pattern recognition using an EMG signal has several innate problems in trying to control a gait assistant robot. The GPES library is robust against amplitude bias owing to the use of derivative analysis and integral reconstruction, while an ISMF reflects better timing characteristics of EMG signals than the time domain feature extraction algorithm. Therefore, it can provide fast detection of gait phase recognition using EMG signals. The experimental results show that the average accuracy of the proposed method is 17% better than that of the existing method. The result also shows that the average latency time of the gait phase recognition of the proposed method is 26.2msec whereas that of the existing method is 195.2msec.