Map Merging with Suppositional Box for Multi-Robot Indoor Mapping

• DOI: 10.3390/electronics10070815
• Publication Date: 2021-03-30T13:57:04Z
• Authors: Baifan Chen; Siyu Li; Haowu Zhao; Limei Liu
• Keywords: Global Map; Merge (version control); Transformation matrix; Feature (linguistics); Line segment; Grid reference

For the map building of unknown indoor environment, compared with single robot, multi-robot collaborative mapping has higher efficiency. Map merging is one fundamental problems in mapping. However, process grid merging, image processing methods such as feature matching, a basic method, are challenged by low matching rate. Driven this challenge, novel method based on suppositional box that constructed right-angled points and vertical lines proposed. The paper firstly extracts selected from point which intersection line. Secondly, common edge characteristics between points, constructed. Then transformation matrix obtained according to pair boxes. Finally, for errors length pairs, Kalman filter used optimize matrix. Experimental results show can effectively merge different scenes successful rate greater than other features.