Linear Parameter-Varying Identification of the EMG–Force Relationship of the Human Arm

Abstract

In this paper, we present a novel identification approach to model the EMG–Force relationship of the human arm, reduced to a single degree of freedom (1-DoF) for simplicity. Specifically, we exploit the Linear Parameter Varying (LPV) framework. The inputs of the model are the electromyographic (EMG) signals acquired on two muscles of the upper arm, biceps brachii and triceps brachii, and two muscles of the forearm, brachioradialis and flexor carpi radialis. The output of the model is the force produced at the hand actuating the elbow. Because of the position-dependency of the system, the elbow angle is used as scheduling signal for the LPV model. Accurate modeling of the human arm with this approach opens new possibilities in terms of robot control for physical Human-Robot Interaction and rehabilitation robotics.