A humanoid robot that can go up and down stairs, crawl underneath obstacles or simply walk around requires reliable perceptual capabilities for obtaining accurate and useful information about its surroundings. In this work we present a system for generating three

dimensional (3D) environment maps from data taken by stereo vision.

At the core is a method for precise segmentation of range data into planar segments based on the algorithm of scan-line grouping extended to cope with the noise dynamics of stereo vision. In off-line experiments we demonstrate that our extensions achieve a more precise segmentation. When compared to a previously developed patchlet method, we obtain a richer segmentation with a higher accuracy while also requiring far less computations. From the obtained segmentation we then build a 3D environment map using occupancy grid and floor height maps. The resulting representation classifies areas into one of six different types while also providing object height in formation. We apply our perception method for the navigation of the humanoid robot QRIO and present experiments of the robot stepping through narrow space, walking up and down stairs and crawling underneathatable.

Note:

1. Main contribution is segmentation of the environment (plane) using stereo camera data

2. Maintain the floor map and a corse 3D occupancy grid map. Horizontal floor map is assumed due to the humanoid stable walking. (consideration point for the biped type robots)

3. Range data obtained by a stereo camera is segmented into planes and integrated into a floor height and 3D occupancy grid from which environment cells are classified

and associated with a height value.

4. [weak point] To get the good disparity map, the environment should contain enough texture.