Research On Mobile Robot Positioning Based On Panoramic Vision Sensor

Positioning is the premise and foundation for finishing the task of navigation in mobile robot, mobile robot has to realize where it situations, so that it can plan the next task. Mobile robot needs sensor to perceive outside information, 75 percent of outside information is obtained by visual in human, which shows that visual censor has irreplaceable advantage than sensors in other types. Panoramic vision sensor, a kind of wide view field sensor, is efficient and convenient in perceiving feature information in surrounding environment, and suitable used in mobile robot positioning.This paper aims at studying mobile robot positioning based on panoramic vision, the panoramic vision sensor installed on the top of mobile robot needs to collect and extract artificial landmarks in environment to finish positioning. Aiming at the problem of image difficult to segment because of abundant pixel points in panoramic image, a method of the combination of HSV threshold segmentation method and OTSU method was proposed, to finish the segmentation of panoramic vision and extraction of artificial landmarks, and then finish the task of positioning. The research in this thesis main as follows:Firstly, this article studied panoramic vision system of mobile robot, introduced the mobile robot Voyager II and panoramic vision sensor UP-OmniVision, and studied the imaging principle and image collection of panoramic vision system, then analyzed some features of panoramic vision.Secondly, the segmentation of panoramic vision was studied. First was the explanation of several color spaces, and the study of the conversion of color spaces. Then by contrast of several common methods in image segmentation, this paper found that the combination of HSV threshold segmentation method and OTSU method was applicable in segmentation of panoramic image. Then this paper studied image morphological operations, and used expansion and corrosion to fill the void.The final section was the process of mobile robot positioning. By the use of colorful artificial landmarks which were extracted, intersecting circle method of triangulation positioning method was used in this paper for positioning. By the acquisition of angles between each artificial landmarks and global coordinate information of artificial landmarks, the global coordinate of panoramic vision sensor optical center could be acquired, and then mobile robot positioning was accomplished. The positioning result acquired by experiment shows that the combination of HSV threshold segmentation method and OTSU method which proposed by this paper could segment panoramic image well, as well mobile robots based on panoramic vision could perceive artificial landmarks in environment stably and efficiency, and adapted to position in wide areas such as workshops in factory, so it shows the significance in reality research.