Running into a trap: Numerical design of task-optimal preflex behaviors for delayed disturbance responses

摘要: Abstract: Legged robots enjoy kilohertz control rates but are still making incremental gains towards becoming as nimble animals. In contrast, bipedal animals amazingly robust runners despite lagged state feedback from protracted neuromechanical delays. Based on evidence biological experiments, we posit that much of disturbance rejection can be offloaded and encoded into feed-forward pre-reflexive behaviors called preflexes. We present a framework for the offline numerical generation preflex to optimally stabilize legged locomotion tasks in presence response By coupling directly collocated trajectory optimizations, optimize preflexive motion simple running model recover uncertain terrain geometry using minimal actuator work. simulation, optimized maneuver showed 30-77% economy improvements over level-ground strategy when responding deviating just 2-4cm nominal condition. claim this “preflex-and-replan” designing efficient gaits is amenable variety extensible arbitrary tasks.