Research on Pattern Recognition of Lower Limb Motion Based on Convolutional Neural Network

Accurate motion recognition is essential for assist devices such as exoskeletons to achieve human-robot communion. However, at present, the technology of lower limb pattern still has problems small amount data and low accuracy. In this paper, was taken object, surface electromyography (sEMG) signals five gaits going upstairs without weight, downstairs with walking on a level weight were collected. Based feature extraction sEMG signal, convolutional neural network (CNN) set input constructed, new method proposed. The accuracy work proposed are compared several other classification methods. experimental results show that, traditional methods, using CNN can better represent features prediction model, higher. all greater than 96.96%, error rate less 7%, indicating that higher This provides theoretical support achieving compliant power assistance promoting motor function rehabilitation robots, power-assisted equipment.