Prediction of Twist Angle for Assistive Exoskeleton Based on EMG Signals

In this study, we propose a method based on electromyography (EMG) for real time prediction of human joint angle, which can be used as upper layer control in assistive exoskeleton. Light assistive exoskeleton can provide help in specific ways for old people hold or get articles, however, due to4the restriction of upper control method and the intrinsic shortage of structure, many exoskeletons can merely be driven in a certain mode, sometimes may not be appropriate to particular situation. Purpose of this work is to figure out the corresponding relationship between EMG and kinematics of human limb, at the same time, to increase diversity of usage of EMG. Prediction of joint angle can save time for control and mechanical delay in service. Without the model of exoskeleton, evaluation of our model is based on the graphics of the error between practical signal curves and predicted signal curves.