We present the evolution of a highly-efficient system for visually detecting number signs in real-time on a mobile robot with limited computational power. The system was designed for use on a robot participating in the 2010 "SICK robot day'' robotics challenge, in which the robot needed to autonomously find 10 number signs and drive to all of them in a given order as fast as possible. Our most efficient method has proven to be robust and successful, since our robot won the competition using the final vision systems described in this paper.
[pdf]
@inproceedings{scherer2011ecmr,
author = {Scherer, Sebastian A. and Daniel Dube and Philippe Komma and Andreas
Masselli and Andreas Zell},
title = {{Robust Real-Time Number Sign Detection on a Mobile Outdoor Robot}},
booktitle = {Proceedings of the 6th European Conference on Mobile Robots (ECMR
2011)},
year = {2011},
address = {\"Orebro, Sweden},
month = sep,
abstract = {We present the evolution of a highly-efficient system for visually
detecting number signs in real-time on a mobile robot with limited
computational power. The system was designed for use on a robot participating
in the 2010 ``SICK robot day'' robotics challenge, in which the robot
needed to autonomously find 10 number signs and drive to all of them
in a given order as fast as possible. Our most efficient method has
proven to be robust and successful, since our robot won the competition
using the final vision systems described in this paper.},
url = {http://www.cogsys.cs.uni-tuebingen.de/publikationen/2011/scherer11.pdf}
}