EMG-driven human model for orthosis control

Abstract

In this paper we present a simplified body model of the human lower extremities used for the computation of the intended motion of a subject wearing an exoskeleton orthosis. The intended motion is calculated by analyzing EMG signals emitted by selected muscles. With the calculated intended motion a leg orthosis is controlled in real-time performing the desired motion. To allow motions with different velocities and accelerations, the body model contains physical properties of the body parts and is animated with data recorded from the pose sensors as a basis for the prediction. Computing the intended motion is achieved by converting calibrated EMG signals to joint torques and forces which are also part of the model. The extrapolation is performed for a short period of time, calculating the joint coordinates for the actuator control loop. The algorithm was examined with the experiment of flexing and extending the knee while raising and lowering the thigh. The discussion compares the motion performed by the leg orthosis and the desired motion. The algorithm of the model and the preliminary experimental results are both presented.