Foot adaptation to obstacle with vision-based landing angle estimation for biped robot

Abstract

The paper presents a novel landing foot angle detection for biped walking robot under uneven ground environments. The authors have past proposed a vision-based motion stabilization method with considering landing foot angle. However, since the landing angle is calculated from internal relative angle between both feet, the past method causes accumulated error of the estimation of landing foot angle at every step. To overcome this problem, the landing foot angle is directly estimated by using visual target position in the field of view. The validity of this method is evaluated by several experimental results.