In this paper, we propose a sEMG signal based gait phase recognition method using adaptive feature and channel selection. This method can be used to control powered artificial prosthetic for lower limb amputees and can reduce overhead in real-time pat...
In this paper, we propose a sEMG signal based gait phase recognition method using adaptive feature and channel selection. This method can be used to control powered artificial prosthetic for lower limb amputees and can reduce overhead in real-time pattern recognition by using only adaptive channels and features in embedded environment. The proposed method can enhance the classification accuracy by selecting channels and features according to sensitivity and specificity of each subject because EMG signal patterns may vary according to subject’s locomotion convention. Experimental results show that the accuracies of channel selection, feature selection and channel/feature selection are better than that of existing method using all channels and/or features.