Vision-based compliant control with landing adaptation to obstacle for biped walking robot

Abstract

The paper presents a vision-based compliant control method with considering the adaptation to unknown floor object for biped walking robot. Generally, while waking motion, stepping on some obstacle causes critical instability in biped robot. The authors have past proposed a reaction force feedback method based on visual feedback manner, and it contributes to stable contact between foot sole and ground environments. However, this method was limited only for flat and level floor environments, because visual feedback includes the influence of landing pitch angle of foot sole. In order to improve this issue, visual feedback method is enhanced to reject the uncertainty of the landing foot angle in the paper. The effect of this method is evaluated by several experimental results.