Simplified control of upright walking by exploring asymmetric gaits induced by leg damping

Abstract

A simple trunk stabilization strategy, called the virtual pendulum (VP) concept, was recently proposed based on human walking data. The implementation of this concept in a simulation model extending the bipedal spring-loaded inverted pendulum (BSLIP) model with a rigid trunk yielded stable upright walking and running patterns. In this study, a first step towards the transfer of the VP concept to real robotic platforms is made by investigating how energy losses due to damping along the leg axis influence the system behavior during walking. We found that the introduction of damping improves the predicted stability of the gait patterns and the robustness of the system with respect to the trunk control parameter. However, further increase of damping reduces the robustness with respect to the leg control parameters. Hence, the control of upright walking based on compliant leg function may be facilitated if an appropriate amount of leg damping is provided.