Research And Design For Mobile Robot Localization System Based On Computer Vision

Localization is an important issue in the study of the mobile robot and its utilization. It is the prerequisite to realize robotic intelligence. Aiming at the shortages of existing localization methods of mobile robot, a new localization method based on the vision system of the robot is implemented here. The method makes use of the camera which is fixed on the top of the robot to capture the images of floor and recognize landmarks on the floor, then determine the orientation and distance of the camera relative to these landmarks. So the location of robot can be gotten.Firstly, considering the requirement for time crisis, a gray-level transformation method based on shifting operation is put forward. Because the Robert's operator is very sensitive to noise, some extension is done to it. A 2×2 smoothing cell is introduced here to alleviate noise. Secondly the Robert's operator is extended to 3×3 template. The result proved to be more satisfying. By improving the Hough line-fitting process, the influence of local maximum on the accuracy of line fitting can be reduced. In order to fit the edge-lines of a image, a Hough transformation should be done to fit each one of the edge lines. The fitted line will be erased from the image before fitting the next one. The accuracy of fitting line is increased.Secondly, both Hough transform and Kalman filter are used to straight lines tracking. This method avoids calculating and matching the model of the straight lines.Thirdly, a simple method is used to camera calibration. The relationships between the image coordination and the world coordination are derived, so the point in the image can be transformed to the world coordination. With these relationships, the robot is localized by the change of the projection of the camera's optical axis on the floor.Experiments have shown that all these methods are correct and feasible. As a result, a mobile robot location system is established by C++ under VS.NET environment. It basically realizes the location of the mobile robot in laboratory. At last location errors are analyzed.