sEMG signal-based portable hand and wrist integrated rehabilitation robot for hand dysfunction

Abstract For patients with hand movement dysfunction caused by diseases such as stroke and hemiplegia, a portable hand and wrist integrated rehabilitation robot is presented based on the human factor engineering. The rehabilitation robot can achieve rehabilitation training of 14 DOFs of fingers and 2 DOFs of wrist. The four-channel sEMG signals of finger and wrist movements are collected by DELSYS equipment, and the pre-processing operations such as denoising and segmentation are carried out by MATLAB. Three kinds of time domain signal features are extracted and combined. The classification training model is obtained by LDA, CNN and LSTM classification method, respectively for different hand movements. Comparisons of the accuracy and learning speed of the three methods show that the CNN method has the highest accuracy and the fastest learning speed. The prototype model of the rehabilitation robot is generated by 3D printing, then the CPM rehabilitation and sEMG control experiments are carried out. Experiment results verify the effectiveness and efficiency of the proposed multi-function rehabilitation robot.