Multi-directional impedance control with electromyography for compliant human-robot interaction

Abstract

Compliant human-robot interaction is necessary for the seamless integration of robotic devices into everyday life. Surface electromyography (sEMG) provides a potential interface for compliant control due to the natural relationship between muscle activity and joint stiffness. Previous works have used sEMG to predict and control impedance along a single degree-of-freedom (DOF). However, humans interacting in unstable environments tune both the magnitude and direction of their impedance. This paper proposes a framework for multi-directional impedance control with sEMG to enhance compliant interactions. The framework allows subjects to simultaneously control both the direction of motion and primary stiffness axis of a robot, enabling stable behavior with controlled compliance to external forces. The efficacy of the approach is shown in a five day experiment with a 3-DOF virtual environment. Subjects displayed significant performance enhancements consistent with motor skill learning as they learned to follow paths and stay at desired targets while compensating for external forces. The framework is also demonstrated with compliant simultaneous and proportional control of a robot arm, suggesting the approach as a natural interface for enhancing capabilities of compliant human-robot interaction.