We investigate the question whether loop closure detection using depth images is feasible using currently available depth features. For this reason, we collected a benchmark dataset consisting of a total number of 15 logfiles with several loops in various environments, implemented a modular and easily extensible loop closure detector and used this to evaluate the adequacy of state-of-the art depth features on our benchmark dataset. To allow for a fair comparison, we determined the best values for the sometimes large number of user-chosen parameters using a large-scale grid search. Since our benchmark dataset contains both depth and RGB images, we can compare the performance relying on depth features with the performance achieved when using intensity image features.
[pdf]
@inproceedings{scherer2013ecmr,
author = {Scherer, Sebastian A. and Alina Kloss and Andreas Zell},
title = {{Loop Closure Detection using Depth Images}},
booktitle = {European Conference on Mobile Robots (ECMR 2013)},
year = {2013},
address = {Barcelona, Catalonia, Spain},
month = {September},
abstract = {We investigate the question whether loop closure detection using depth
images is feasible using currently available depth features. For
this reason, we collected a benchmark dataset consisting of a total
number of 15 logfiles with several loops in various environments,
implemented a modular and easily extensible loop closure detector
and used this to evaluate the adequacy of state-of-the art depth
features on our benchmark dataset. To allow for a fair comparison,
we determined the best values for the sometimes large number of user-chosen
parameters using a large-scale grid search. Since our benchmark dataset
contains both depth and RGB images, we can compare the performance
relying on depth features with the performance achieved when using
intensity image features.},
pdf = {http://www.cogsys.cs.uni-tuebingen.de/publikationen/2013/scherer2013ecmr.pdf},
}