Decentralized Control Mechanism Underlying Morphology-Dependent Quadruped Turning

Abstract

Quadruped mammals can control the movement of their center of gravity when turning by skillfully utilizing their bodies to achieve adaptive turning movements. Interestingly, the low-speed turning behavior also changes depending on the animal’s morphology. Therefore, this study aims to understand the control algorithm of low-speed turning, which can reproduce the turning behavior according to the location of the center of gravity. Specifically, we constructed a control algorithm based on the knowledge that animals steer with the leg closest to the center of gravity and verified it with a quadruped robot whose center of gravity could be adjusted. Consequently, the behavior observed in animals was successfully reproduced, with a stable and large turning angle per time when the proposed control algorithm was used.