An approach to balance and posture estimation using image feature points for biped walking robot

Abstract

This paper presents an approach to balance and posture estimation based on image feature points from camera image for biped walking robot. In this paper, the image feature tracker is employed in order to conduct the balanced motion mainly from visual information. In the proposed approach, the point cloud of tracking features is evaluated through the image moment calculations to obtain the inclined angle and the reaction force from ground, and they are used for real-time motion control of biped robot. The validity of the proposed method is evaluated by several experimental results.