Active Vision Based Relocalization Of Wheel Robot

Cultural relics has always been the focus of the national protection,and high value of the cultural relics of the scene changes are often subtle,some changes can be observed even after hundreds years.It also increase the difficulty for the monitoring and protection of cultural relics of high value scenario tiny changes.This paper solves the problem of small change detection of outdoor high-value cultural relics,and main contributions are divided into three aspects.Firstly,it put forward to the method of robot relocalization based on active vision.This method uses by extracting the feature points between reference image and camera current image,matching them by similarity,and eliminating the error pairs of matching points to get a series of matching pairs of feature points.According to these pairs of feature points,we can calculate the single transformation matrix of the two images,then we will obtain the navigation box of two images by using the real-time update of the single transformation matrix.After that,we can get the moving information by the navigation box of two images.Finally,the driving wheel robot independently carries out the relocalization process by the algorithm of semi-fixed path;secondly,we designed and developed a large scale,high load,and all-round robot platform,which provides the foundation of sports platform based on the visual relocalization method;finally,we realized the active vision based relocalization of wheel robot.Compared with the traditional method of manual relocalization,the method proposed in this paper don't need to staff intervention,while guarantee the relocalization precision,shortens the time and improves the working efficiency.