This thesis addresses the use of radio-frequency identification (RFID)
for three fundamental tasks in mobile robotics: mapping, self-localization,
and trajectory estimation. These topics have widely been studied
in recent years, because they are regarded as key ingredients to
autonomous robots. In particular, we investigate the application
of long-range passive RFID in the work at hand. This RFID standard
is expected to play a major role in the automation of identification
processes in industry and commerce.
We present solutions to the automated inventory of RFID-tagged goods,
which we interpret as mapping the locations of RFID transponders.
In order to quickly provide the required model of the detection behavior
of the RFID reader on board the mobile robot, we devise a semi-autonomous
learning method for probabilistic sensor models. Moreover, we compare
different techniques to self-localize in an RFID-tagged environment.
In the last part, we show how the trajectory of the robot can be
estimated in previously unknown surroundings by tailoring approaches
from the field of simultaneous localization and mapping (SLAM).
@phdthesis{vorst2011thesis,
author = {Vorst, Philipp},
title = {Mapping, Localization, and Trajectory Estimation with Mobile Robots
Using Long-Range Passive {RFID}},
school = {University of Tuebingen, T\"ubingen, Germany},
year = {2011},
month = aug,
abstract = {This thesis addresses the use of radio-frequency identification (RFID)
for three fundamental tasks in mobile robotics: mapping, self-localization,
and trajectory estimation. These topics have widely been studied
in recent years, because they are regarded as key ingredients to
autonomous robots. In particular, we investigate the application
of long-range passive RFID in the work at hand. This RFID standard
is expected to play a major role in the automation of identification
processes in industry and commerce.
We present solutions to the automated inventory of RFID-tagged goods,
which we interpret as mapping the locations of RFID transponders.
In order to quickly provide the required model of the detection behavior
of the RFID reader on board the mobile robot, we devise a semi-autonomous
learning method for probabilistic sensor models. Moreover, we compare
different techniques to self-localize in an RFID-tagged environment.
In the last part, we show how the trajectory of the robot can be
estimated in previously unknown surroundings by tailoring approaches
from the field of simultaneous localization and mapping (SLAM).},
isbn = {978-3-8439-0060-7},
keywords = {radio-frequency identification (RFID), passive UHF RFID, mobile robot,
self-localization, mapping, trajectory estimation, simultaneous localization
and mapping, SLAM, sensor model, location fingerprinting},
pdf = {http://www.cogsys.cs.uni-tuebingen.de/mitarb/vorst/publications/Dissertation_Philipp_Vorst.pdf},
publisher = {Verlag Dr.~Hut},
series = {Robotik und Automation},
url = {http://www.dr.hut-verlag.de/978-3-8439-0060-7.html}
}