Passing under Obstacles with Humanoid Robots

Introduction

Since humanoid robots have human-like shapes, they are expected to work in various environments designed for humans. Such humanoids need to be able to recognize surroundings, move around, manipulate objects, and avoid obstacles. A humanoid robot is assumed to be able to change its body posture and pass underneath an obstacle. This is quite different from a typical mobile robot.

There have been several attempts to develop methods for humanoid passing under motions. Shiller et al. presented a practical motion planner, which considers the most common motion postures, represented by a 2D grid, for humanoids and animated human figures [1]. Fujimoto et al. proposed a method that selects typical pre-designed locomotion forms such as a forward-facing walk, a sideways walk, and a crawl based upon visual information of obstacles [2]. Kanehiro et al. [3] proposed a method that generates a 3D local map from visual information and plans the appropriate locomotion based on the map for bipedal walking with variable height and width, and for crawling. In all these methods however, the appropriate motion pattern for passing under an obstacle is selected from several prepared motion patterns according to the environment. Therefore, the created motion is limited to prepared motion patterns.