Control and Experiment for Exoskeleton Robot Based on Kalman Prediction ofHuman Motion Intent

To accurately acquire human motion intent for exoskeleton control, torque sensors are used to measure human-robot interaction information. Based on the single pendulum model of human swing leg, the motion trajectory of lower limb joints is obtained, which is predicted with Kalman filter to compensate the delay of motion intent. The PD (proportional-derivative) control law is used to control exoskeleton to track the joint trajectory of human swing leg, and real-time position of the exoskeleton joint is fed back by the encoder, which forms a closed-loop position control. Experiments on the exoskeleton swing leg are conducted and the results show that motion intent of human swing leg can be predicted with the measured human-robot interaction information using Kalman filter, and the exoskeleton robot can achieve the joint trajectory tracking of human swing leg. Therefore, the proposed method is effective.