Learning Visibility of Landmarks for Vision-Based Localization

Show simple item record

dc.contributor.author Alcantarilla, Pablo F.
dc.contributor.author Oh, Sang Min
dc.contributor.author Mariottini, Gian Luca
dc.contributor.author Bergasa, Luis M.
dc.contributor.author Dellaert, Frank
dc.date.accessioned 2011-03-29T17:19:33Z
dc.date.available 2011-03-29T17:19:33Z
dc.date.issued 2010
dc.identifier.citation Alcantarilla, P.F., Oh, S. M., Mariottini, G.L., Bergasa, L.M., Dellaert, F.(2010). “Learning Visibility of Landmarks for Vision-Based Localization”. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA 2010), 3-7 May 2010, 4881-4888. en_US
dc.identifier.issn 1050-4729
dc.identifier.uri http://hdl.handle.net/1853/38323
dc.description ©2010 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works. en_US
dc.description Presented at the 2010 IEEE International Conference on Robotics and Automation (ICRA), 3-7 May 2010, Anchorage, AK.
dc.description DOI: 10.1109/ROBOT.2010.5509383
dc.description.abstract We aim to perform robust and fast vision-based localization using a pre-existing large map of the scene. A key step in localization is associating the features extracted from the image with the map elements at the current location. Although the problem of data association has greatly benefited from recent advances in appearance-based matching methods, less attention has been paid to the effective use of the geometric relations between the 3D map and the camera in the matching process. In this paper we propose to exploit the geometric relationship between the 3D map and the camera pose to determine the visibility of the features. In our approach, we model the visibility of every map feature w.r.t. the camera pose using a non-parametric distribution model. We learn these non-parametric distributions during the 3D reconstruction process, and develop efficient algorithms to predict the visibility of features during localization. With this approach, the matching process only uses those map features with the highest visibility score, yielding a much faster algorithm and superior localization results. We demonstrate an integrated system based on the proposed idea and highlight its potential benefits for the localization in large and cluttered environments. en_US
dc.language.iso en_US en_US
dc.publisher Georgia Institute of Technology en_US
dc.subject Camera pose en_US
dc.subject Data association en_US
dc.subject Localization en_US
dc.subject 3D landmark en_US
dc.subject 3D maps en_US
dc.subject Visibility en_US
dc.title Learning Visibility of Landmarks for Vision-Based Localization en_US
dc.type Post-print en_US
dc.type Proceedings
dc.contributor.corporatename Georgia Institute of Technology. Center for Robotics and Intelligent Machines
dc.contributor.corporatename Georgia Institute of Technology. College of Computing
dc.contributor.corporatename Georgia Institute of Technology. School of Interactive Computing
dc.contributor.corporatename Universidad de Alcalá. Departamento de Electrónica
dc.contributor.corporatename University of Minnesota. Dept. of Computer Science and Engineering
dc.publisher.original Institute of Electrical and Electronics Engineers

Files in this item

Files Size Format View
Alcantarilla10icra2.pdf 639.7Kb PDF View/ Open

This item appears in the following Collection(s)

Show simple item record