An EMG-controlled Mobile Robot Based on a Multi-layered Non-contact Impedance Model

This paper proposes an obstacle avoidance method for EMG-controlled mobile robots based on a noncontact impedance model. The proposed system can voluntarily control a mobile robot by classifying EMG signals using a recurrent probabilistic neural network and can avoid obstacles without user handling based on virtual repulsive force through a multi-layered non-contact impedance model. In the experiments, two obstacles were arranged in the path of the mobile robot, and the participant was asked to control the robot toward a target. The robot passed through the obstacles smoothly without any avoidance operations, indicating that the proposed system could be used for obstacle avoidance in mobile robots.