A Method for Rapid Generation of Passive Dynamic Walking Gait Based on Gait Library

Passive dynamic walking offers an effective way achieving energy-efficient bipedal locomotion. In this paper, we present a method for rapidly generating gaits based on a gait library. Initially, a set of period-1 passive dynamic walking gaits is pre-computed. Then, these gaits are parameterized using Bezier curves. Finally, by employing polynomial regression to fit gait features and parameters, a gait library is constructed. Case studies are conducted on a compass-like biped model and a kneed biped model. The results demonstrate that the gait library is capable of rapidly generating a desired gait within 1 millisecond, with energy consumption almost equivalent to that of passive dynamic walking.