Meeting Abstract
Cursorial ground birds are paragons of bipedal running, even over terrain of highly variable geometry. We hypothesize that these behaviors emerge from underlying task-level control priorities. We formulate this hypothesis as a control optimization problem, seeking to explain and predict features of avian locomotion, across species, by optimizing control applied to reduced-order math models of locomotion. We investigated the relative priority for energy economy and attenuation of gait perturbations (a proxy for gait stability) by running birds over obstacles that put these priorities in direct conflict. We compared the observed maneuvers to optimal control for each priority applied to a reduced-order model (a simple spring-mass damper model with a single linear leg actuator). After validating the model against steady bird locomotion, our obstacle negotiation analysis suggested that birds, from quail to ostrich, prioritize energy economy and avoidance of musculoskeletal injury risk over attenuating deviations from steady gait. Our optimizations suggest that priority for diminishing gait deviations would demand higher mechanical work and leg forces, increasing energy costs and risk of injury. These findings suggest a novel approach to bipedal robot control, and our ongoing work seeks to apply task-level control priorities of economy and injury avoidance on a suitable robot model, ATRIAS, a human-scale biped. Since ATRIAS is designed to mechanically match to our reduced-order model, we hypothesize control optimization for economy and injury avoidance will produce running that is dynamically similar to birds.
