Research On Moving Target Tracking In Mobile Robot System Based On Omni-Vision And Laser Rangefinder

Target tracking is a hot issue in the field of robotics. As an interdisciplinary cutting-edge technology, it combines the knowledge of many areas and has some theoretical significance. The reliable target tracking technology is widely applied in the areas of military, civilian and scientific research. Real-time and accuracy have been a serious problem in target tracking. In this dissertation, omni-vision and laser ranging are applied to the mobile robot reasonably for improving the real-time and accuracy of tracking target.Firstly, the target was detected by use of moving target detection algorithm. The omni-vision image was effectively extended to ensure the real-time of tracking target. The frame difference method was applied to the expanding image sequence and the sequence data from laser range finder. The moving target area was detected after removing the noise. The target information detected from single sensor was effectively integrated, so that moving target was found accurately and reliably.Then, the color sampling points were determined and the color feature was extracted by the color feature extraction algorithm. The single sampling point of moving target was fixed through the integration of information and multi-sensor sampling algorithm after target detection. The average color feature of point sets which were similar to a sampling point was extracted at the neighborhood of its center.Finally, the target location and tracking were achieved by target tracking algorithm. The tracking window was determined and filled with the feature color. The target color within window was projected in the horizontal direction, and its projection strength was enhanced according to the real-time data from the laser. According to the projection, the specific location and distance were determined. The track strategy was adjusted to track the moving target effectively and in real-time for robot according to the location of target relative to the robot. The still obstacles were successfully avoided. If the moving target were lost over a sustained period of time, the tracking algorithm was stopped.The experimental results show that the proposed target detection algorithm not only meets the requirement for real-time but also detect the moving target accurately and reliably. The algorithm of extracting color feature can extract and enhance the color characteristics of moving target quickly and effectively. The moving target tracking algorithm can improve real-time and robustness for tracking of mobile robot.