Research On Mobile Robot Localization Method Based On Landmarks Information

Autonomous mobile robot localization is one of the most important research branches of intelligent robot. It is the prerequisite to realize robotic navigation. A new localization method which is based on the vision system of the robot and landmarks information in indoor environment is implemented here. The method makes use of the camera which is fixed on the top of the robot to continuously capture the images of floor, and using the edge information of ceramic tiles on the floor as the nature landmarks to confirm the orientation and distance of the camera relative to these landmarks, which is based on the relativity of these pictures and landmarks' change when the height and angle of camera are unknown. Then make sure the move distance and direction of robot, so the location of robot can be gotten. The method includes four parts: low level image processing, camera calibration and image mapping, straight lines tracking, robot localization.Low level image processing is the base of localization. The paper discusses time domain filtering and frequency domain filtering according to the noise disturbance in the surrounding images. Then, considering the shortages of Laplacian operator, the classical algorithm has been improved. We offered a new algorithm, which is disassembled the Laplacian operator in four direction:0°,45°,90°and 135°, then we get two separable Laplacian operator . We obtain a better effect on edge detection. In order to improve the accuracy of line-fitting process, we carry on several times Hough transformations to the picture, and the distance from the data to the straight line must be smaller than the error threshold at each Hough transformation.Robot vision system must constitute the relation between the spatial point and the plane point. A simple method has been given grounded on camera interior and exterior parameter model. Then, we derive a series of formulae from the world coordination to the image coordination. So the relationships from the image coordination to the world coordination are also gotten and the point in the image can be transformed to the world coordination directly under the circumstances that the height and angle of camera are unknown. With these relationships and formulae, we can get the height and angle of the camera real time. Then, the robot is localized by the change of the camera relative to these nature landmarks on the floor at the complex environment.We use a tracking line method based on Hough transform, the presented method transform line-tracking in image space into point-tracking in Hough space. The transformed point in Hough space is defined as feature point, so the feature point can be tracked. We get a good estimation of the lines tracking.Experiments have shown that it basically realizes the location of the mobile robot in laboratory. At last location errors are analyzed.