A robophysical model to study physical sensing of obstacles in legged traversal of complex 3-D terrain

Terrestrial animals transition between locomotor modes to traverse complex 3-D terrain, an ability that robots still lack. For example, the discoid cockroach traverses grass-like beam obstacles by transitioning from the pitch mode (pitches body up while pushing across the beams) to the roll mode (rolls body to maneuver through the gap). Our recent study discovered that body oscillation facilitates this transition by using kinetic energy fluctuation to probabilistically cross a barrier on a potential energy landscape (Othayoth, Thoms, Li, 2020, PNAS). Besides body oscillation, the animal also flexed its head and abdomen and adjusted its hind leg movement during transition. We hypothesized that the animal sensed physical interaction by flexing its head against the obstacles, then adjusted the abdomen and leg motions accordingly to facilitate pitch-to-roll transition. To study this, we developed a robophysical model with actuated flexible head and abdomen and underactuated body pitch and roll control which mimicked the effect of leg motions. To measure terrain contact forces, we added a force sensor in its head. We are doing experiments to discover how the sensed resistance relate to the potential energy landscape and guides abdomen and leg adjustments to facilitate transition.