Robot Navigation Based On Vision

The robot navigation is an important issue in the filed of robotics .Because of the advantage of the visual sensor, which is non-contact and capturing a wealth of goal information, the robot navigation based on vision is becoming the important research direction.In this paper, we research the current theory about the robot navigation and the robot visual navigation. And then, considering the robot application and experimental environment, we design two robot visual navigation systems. The first system is a binocular vision ranging system based on the zoom lens. It is dependent on the angle of camera rotation rather than the focal length that is the essential element in the traditional binocular vision ranging methods. So we can use the zoom lens camera and capture clearer image of object. The second system is the robot corridor walk vision navigation system. We specially design the visual servoing system and the improving artificial potential field method for the circumstances of the corridor. The visual servoing system is planed to extract the target to the navigation system and the improving artificial potential filed method is planed to plan the path for the autonomous mobile robot. In this system, the improving artificial potential field method not only solved the force-balance problem but reduced the walking oscillation.We test the proposed techniques from the binocular vision ranging system based on the zoom lens indoor and the robot corridor walk navigation system and good results have been obtained.