Active adjustments help cockroaches traverse obstacles by lowering potential energy barrier

Robots predominantly navigate complex environments by using geometric models to avoid physical contact with obstacles. By contrast, animals easily traverse obstacles by using physical interaction controlled by active feedback. Our previous study revealed that cockroaches made active adjustments to traverse grass-like beam obstacles. As the animal pushed against the beams, its body pitched up. Then, the animal often quickly rolled into the gap between beams to traverse. In this process, the animal’s head often flexed repeatedly, and one hind leg pushed against the ground while the other retracted. Here, we used physics modeling and robot experiments to understand the function of such active adjustments. We calculated system potential energy (resulting from gravity and beam elastic bending) as a function of head flexion. When the animal was in contact with the beams, head flexion lowered its potential energy barrier to transition from pitch to roll mode (by nearly half for the observed head flexion of 20°). We hypothesize that the animal uses proprioception (internal position and force sensing) to detect changes in beam resistive forces, which reflect the reducing transition barrier. We are developing a robotic physical model instrumented with force sensors to test this hypothesis.