This work is about metric localisation across extreme lighting and weather conditions. The typical approach in robot vision is to use a point-feature-based system for localisation tasks. However, these system typically fail when appearance changes are too drastic. This research takes a contrary view and asks what is possible if instead we learn a bespoke detector for every place. Our localisation task then turns into curating a large bank of spatially indexed detectors and we show that this yields vastly superior performance in terms of robustness in exchange for a reduced but tolerable metric precision. We present an unsupervised system that produces broad-region detectors for distinctive visual elements, called scene signatures, which can be associated across almost all appearance changes.


  • [PDF] C. McManus, B. Upcroft, and P. Newman, “Scene Signatures: Localised and Point-less Features for Localisation,” in Proceedings of Robotics Science and Systems (RSS), Berkeley, CA, USA, 2014.
    [Bibtex]

    @inproceedings{McManusRSS2014,
    Address = {Berkeley, CA, USA},
    Author = {Colin McManus and Ben Upcroft and Paul Newman},
    Booktitle = {Proceedings of Robotics Science and Systems (RSS)},
    Date-Added = {2014-04-29 01:07:27 +0000},
    Date-Modified = {2014-06-11 08:02:17 +0000},
    Month = {July},
    Pdf = {http://www.robots.ox.ac.uk/~mobile/Papers/2014RSS_McManus.pdf},
    Title = {Scene Signatures: Localised and Point-less Features for Localisation},
    Year = {2014}}