Towards Obstacle-aided Legged Locomotion in Cluttered Environments

Modern robots are often required to perform tasks in environments filled with obstacles and disturbances. However, most legged platforms are not yet capable of coping gracefully with unanticipated repeated disturbances, and often rely on active sensing to avoid all engagements with their physical surroundings. In this study, we propose a novel framework wherein we abstract the obstacle-cluttered environment into a horizontal-plane disturbance force field, and we consider robot legs as disturbance selectors. With different gait patterns, the robot can generate different disturbance forces on its center-of-mass from the same physical environment. Our simplified model significantly reduces the complexity of representing interactions between robot and obstacle-cluttered environments, and begins to suggest an approach to using gait space affordances for purposes of generating desirable obstacle responses in multi-legged robot locomotion amidst complex environments.