The heart of the SLAM problem is to determine the ``best'' map, the
physical locations of features such that the constraints have maximum
probability. We consider the case where the features are locations
visited by the robot; as shown by \cite{Montemerlo_2003}, positions of
other features can be efficiently computed once the robot trajectory
is known. In this paper, we present an algorithm for optimizing pose graphs that
is dramatically faster than the published state of the art.
This paper has been revised and expanded. Click here to go to the updated paper.
Best Paper
@inproceedings{olson2005csw, AUTHOR = {Edwin Olson}, TITLE = {Incremental Optimization of Large Robot-Acquired Maps}, BOOKTITLE = {CSAIL Student Workshop Proceedings}, YEAR = {2005}, }