A study on obstacle avoidance for stable walking of biped robots

The autonomous mobility for the biped robot KU-ROBO in the home environment is realized base on the development of a small stereo vision system, the recognition of floor and obstacles using plane extraction. The terrain is represented in a robot centric coordinate system without making any structural assumptions about the surrounding world. And the representation of a terrain map based on these observations, robot motion, and the generation of a walking path on the terrain map. We therefore believe, our approach is well suited for many different a home environment where no a priori information about the environment is given. The limitation of our system is that the terrain has to contain enough texture in order to obtain reliable stereo data.