Dynamic Walking on Stepping Stones with Gait Library and Control Barrier Functions

Dynamical bipedal walking subject to precise footstep placements is crucial for navigating real world terrain with discrete footholds such as stepping stones, especially as the spacing between the stone locations significantly vary with each step. Here, we present a novel methodology that combines a gait library approach along with control Barrier functions to enforce strict constraints on footstep placement. We numerically validate our proposed method on a planar dynamical walking model of MARLO, an underactuated bipedal robot. We show successful single-step transitions from a periodic walking gait with a step length of 10 (cm) to a stepping stone with a 100 (cm) separation (10x step length change), while simultaneously enforcing motor torque saturation and ground contact force constraints. The efficacy of our method is further demonstrated through dynamic walking over a randomly generated stepping stones requiring single-step step length changes in the range of [10:100] (cm) with a foot placement precision of 2 (cm).