AGV Of Indoor Positioning Method Research Based On Monocular Vision

The widespread use of AGV in various industries and various places greatly improves the efficiency of production and changes people's life.If you want AGV to autonomous navigate,you need to first determine the position of AGV.So AGV needs positioning technology as the first skill.In this paper,I design AGV of indoor positioning system based on monocular vision,using artificial landmarks information.I use image processing technology to extract artificial landmarks from the background and use image matching technology to recognize artificial landmarks which have been extracted.Using known artificial landmarks coordinates,I can calculate the position and orientation of AGV through the establishment of distance measuring model and positioning model.Firstly,I extract artificial landmarks from the background by using image processing technology,such as median filter,color space transition,fixed threshold segmentation and mathematical morphological open operation.I use median filter to remove the noises on image.The image which has been removed the noises is transformed RGB color space to HSV color space.To realize coarse segmentation of the image by fixed threshold segmentation,I use mathematical morphological open operation to precise the result of coarse segmentation.Secondly,I use SIFT algorithm to recognize artificial land marks which have been extracted and use RANSAC algorithm to improve recognition accuracy.And then,using known artificial landmarks coordinates,I can calculate the position and orientation of AGV though the establishment of distance measuring model and positioning model.Finally,I build AGV of indoor positioning system through hardware equipment and build a human-computer interaction interface by software platform.I test this AGV of indoor positioning system and analysis the test result.