ACTIVE LATERAL FOOT PLACEMENT FOR 3D STABILIZATION OF A LIMIT CYCLE WALKER PROTOTYPE

Abstract

This study focuses on the application of active lateral foot placement for 3D stabilization of bipedal walkers. Within the paradigm of "limit cycle walking" foot placement is an important strategy as it can provide cyclic stability for walkers that are locally unstable. Moreover, human gait analysis studies suggest that the stability of human walking depends highly on lateral foot placement. Various simulation studies have already successfully implemented lateral foot placement in walking models, but this study demonstrates that an active lateral foot placement strategy can actually (cyclically) stabilize a physical walking robot that is locally unstable. In order to come to this result, first a study is performed on a simple 3D point mass walking model. This study establishes that, for a model with fixed step length, cyclic stability can already be obtained with a simple linear lateral foot placement strategy that only uses lateral state information (lateral position and velocity) of the center of mass. Moreover, it is found that increasing the walking speed and increasing the ankle roll stiffness enlarges the range of stable feedback gains. With this knowledge of stable feedback gains and parameter sensitivities, the same foot placement strategy is applied to the physical 3D walking prototype called Flame. Similar to the model, this prototype is shown to be unstable without foot placement and stable with the application of the simple, linear lateral foot placement strategy.