Adaptation in Bipedal Locomotion Using Phase Oscillator Networks

Stable and robust dynamic locomotion has been gaining increasing attention to enhance the applicability of biped humanoids to dynamically changing environments. In this work, we propose an efficient neural oscillator network that controls the phase of the gait by incorporating sensory signals detecting the changes in terrain slope. This enables biped humanoids to reliably control periodic features of legged locomotion adapting to changing environments. Specifically, locomotion trajectories of individual limbs are designed as predetermined functions of the reference phase of the gait sent from the phase generator. These trajectories are autonomously modified to keep the posture of the humanoids balanced by exploiting phase adaptation through sensory feedback. In order to verify the validity of the proposed scheme, we perform experiments with a real robot under various conditions of terrain.