Locomotion mode Classification using Frequency Domain Filter based on Feature Array of sEMG Signals

초록

This paper proposes a locomotion mode recognition method using frequency domain filter based on feature array of lower limb surface electromyogram (EMG) signals in gait phase. Force sensors are used to detect before and after starting points of stance and swing phases. For detecting locomotion mode, EMG signals of all channels are extracted and assembled to feature array in time domain and then converted to those in frequency domain by using Fast Fourier Transform. Locomotion mode is classified by applying frequency domain filters to the feature array. Experimental result shows that the average accuracy of classifying locomotion mode at stance and swing phases is 76.8%.