The mechanism of synchronization-based control using neural oscillators for walking assist

Abstract

We proposed synchronization-based control using neural oscillators for walking assist. In order to synchronize a robotic suit's motion with that of a user, the mutual joint torque was used as input signal to a neural oscillator, and the output signal of neural oscillator as desired trajectory of each joint of the suit. This paper conducted computer simulations and experiments to explain the mechanism of synchronization-based control in respects of assist effect, friendly motion assist such as making a stop and restart, as well as the merit and demerit of our proposal. Basing on the results, we conformed that synchronization-based control was a simple system and helped the robot act synchronously and provide friendly assist. In addition, we found that a kind of “attenuation mode” occurred if the input and output signals of neural oscillator were in reverse directions. The attenuation makes the desired angle to be generated more reasonably for motion assist, such as facilitating stop and restart, thus the assist action to be more compliant.