Research Humanoid Robot Walking Based on Vision-Guided

Walking is a basic function of humanoid robot, this paper presents key ideas of stereo vision based humanoid walking. Image processing techniques and pattern recognition techniques are employed for the obstacle detection and object recognition, data fitting technique is also used to plan the path of the humanoid robot. High precision visual feedback is provided by the combination of real time high precision feature detection and high actuary object detection method. The proposed stereo vision based approach and robot guidance system were evaluated partly by experiments and partly by the simulation with the humanoid robot.