Estimation of Ankle Joint Continuous Motion Based on Electromyographic Signals

Abstract

Human electromyogram (EMG) contains abundant body movement information, which is often used as the information medium of human-robot interaction. Precisely analyzing the muscle activity information contained in the patient's EMG and using it for the motion control of the rehabilitation robot is of great significance in the rehabilitation training process. Aiming at the problem that ankle rehabilitation robot needs to use the patient's EMG to identify the ankle motion intention and estimate the angle of continuous motion in the human-robot interactive rehabilitation training, this paper proposes a new method based on the patient's EMG to directly identify the type of ankle motion and estimate the continuous motion angle of the joint. Considering the nonlinear relationship between EMG and joint motion angles, Hammerstein model is selected to describe the relationship between them. The linear and nonlinear structure of the Hammerstein model simplifies the difficulty of model identification. First, obtain the real motion angle of the ankle joint while collecting the EMG of the muscles related to the ankle joint motion, and use the threshold judgment method to complete the recognition of the ankle joint motion intention. Then, use the EMG as the input of the motion model and the real motion angle of the ankle joint as the motion model output, identify and obtain the Hammerstein motion model of plantarflexion motion and dorsiflexion motion respectively. Finally, use EMG as the input of the recognized model to estimate the continuous motion angle of the ankle joint. The experiment shows that the motion model established by the method in this paper can realize the recognition of the motion intention of the ankle joint and the accurate estimation of the continuous motion angle, this method is worth promoting.