A robust biped gait controller using step timing optimization with fixed footprint constraints

Abstract

Humans are able to wrestle each other on plum blossom pole (a kind of Chinese Kong Fu on tiny footholds), while our robots have yet to recover from a push while walking with fixed footprint constraints. We find it remarkable that combining step-timing optimization with classical modification of Center of Pressure (CoP) reduces the margin of instability. In this paper, we propose a simplified multi-objective optimization based algorithm to keep the Divergent Component of Motion (DCM) of the Linear Inverted Pendulum (LIP) model we use within stable margin, and make the robot more agile when recovering from tilted state. We have verified our algorithm robustness via dynamics simulation and hardware experimental results on our THU-Strider Platform.