Robust vision-based indoor localization
PubDate: April 2015
Teams: University of Oxford
Writers: Ronald Clark；Niki Trigoni；Andrew Markham
Vision-based positioning has proven to be highly successful and popular in mobile robotics and computer vision applications. These methods have, however, not enjoyed the same popularity in the field of indoor localization.
In this work we highlight some of the issues that arise when using vision-based methods for indoor localization. We then propose means of addressing these issues and implement a proof-of-concept visual inertial odometry system for a mobile device.
Preliminary experiments have been carried out in a small library where sub-meter positioning accuracy was attained. Based on our proof-of-concept, we believe that visual inertial odometry techniques can provide the levels of positioning accuracy needed for widespread adoption.