Interaction Approach for Movement-Assist Control Using Neural Oscillators

Abstract

In this paper we propose a framework of realizing natural assist behavior with a movement-assist suit inspired by human interaction. Since human interaction can be thought as synchronization behavior common in movement assistance between human beings, to achieve human-like movement assistance, synchronization-based control is applied to a movement-assist suit. We use neural oscillators to entrain and synchronize suit movement with that of users. To determine validity and feasibility, we examine the proposal for whether (1) synchronization of action between human and movement-assist suit can be realized, (2) the assist effect can be obtained, and (3) the proposed method is comfortable for users. To determine these points, we simulated movement assistance and conducted experiments with a joint-torque-sensing assist suit. Results demonstrated that synchronized movement was realized and that a movement-assistance effect was implemented. Results of evaluation experiments showed the good usability the suit has as proposed, confirming the proposal's applicability and performance.