Motion planning for humanoid robots stepping over obstacles

Abstract

In this paper, we address the problem of how a humanoid robot can step over a given obstacle. Obstacle stepping-over has two aspects, namely, feasibility analysis and motion planning. The former determines whether the robot can step over the obstacle, and the latter discusses how to realize the stepping-over, if it is feasible, by trajectory planning. The paper focuses on the latter. Specifically, based on our previous analysis of feasibility, we present a novel algorithm to plan suitable trajectories for obstacle stepping-over, taking into account two basic requirements. The first requirement is to avoid any collision between the robot and the obstacle, and the second to maintain stability or balance of the robot. To meet them, we decompose the whole body motion of the robot into two parts, corresponding to the upper body motion and the lower body, respectively. We first plan collision-free trajectories of the feet and the waist for lower body motion, and then adjust upper body motion by resolved momentum control to guarantee the robot stability. This novel planning method is adaptive to obstacle sizes and hence oriented to autonomous stepping-over of humanoid robots guided by vision or other range finders. Its effectiveness is shown by simulation and experiment on our humanoid platform HRP-2.