AmbiSense

In the scope of the joint project AmbiSense, our department investigates passive UHF radio-frequency identification (RFID) as a sensor for localization and continuous inventory using mobile service robots. AmbiSense is a research cooperation of the Universities of Tübingen and Stuttgart and funded by the Baden-Württemberg Stiftung (former Landesstiftung Baden-Württemberg) in the scope of the program BW-FIT. Since 2006, the project partners have investigated the utilization of ambient sensors (RFID, WLAN, Bluetooth) for different purposes such as localization, inventory, stock control, etc. Among other work packages, we have developed solutions to the topics depicted subsequently.

We employ passive UHF RFID, particularly off-the-shelf hardware of the standard EPC Class 1 Generation 2 (ISO 18000-6C). This choice is motivated by the expectation that it will continuously replace barcodes for labelling goods in warehousing and logistics to a great extent. An onboard RFID reader with a range of several meters attempts to read inexpensive RFID tags by means of radio waves. This way, transponders serve as uniquely identifiable landmarks for robot navigation.

SCITOS G5 mobile robot with UHF RFID reader

Non-parametric Sensor Model Learning

Localizing RFID-tagged objects from known positions requires an RFID sensor model -- a model which describes at which rate RFID tags can be read, depending on their relative position to an RFID antenna. Such a model can efficiently be learned using mobile robots with reference positioning systems. Recorded reference positions and RFID data of known transponders serve as training data for model learning. Then, our approach applies nonparametric regression techniques (such as kernel or k-nearest neighbor smoothing) to derive a representation of the detection characteristics. Nonparametric Learning of an RFID Sensor Model

Mapping

Given a sensor model and RFID sensor data, our robot can inventory tagged objects and determine their positions. We examined a fusion method to improve transponder position estimates by incorporating knowledge about the geometrical properties of the environment. Using 2D laser occupancy grids, we were able to reduce the position estimation error by 10-33 % on average. Improved Mapping of RFID Tags by Fusion with Spatial Structure

Localization via Fingerprinting

Location fingerprinting signifies the position estimation from reference measurements taken a-priori at known locations. This technique requires no explicit sensor model and is supposed to be more accurate, as location-specific observations form the basis of inferring positions. Location fingerprinting is efficient using mobile robots, because they can move autonomously and use additional sensors to annotate measurements during the mapping stage. We have proposed two fingerprinting strategies, using particle filters in combination with vector similarity measures or detection rate estimators. Depending on the setup, we achieved a localization accuracy of 0.2-0.3 m.

Trajectory Estimation (SLAM)

In order to eliminate potentially expensive sensors for reference positioning (e.g., laser range finders), we have examined trajectory estimation techniques. The goal is to reconstruct the path of the robot in an unknown (not yet mapped) environment, using RFID and odometry only. Then, the estimated positions can be used as reference locations for fingerprinting. We have investigated approaches based on particle filters and on pose graph optimization (maximum likelihood mapping). These are effective techniques fro solving the related simultaneous localization and mapping (SLAM) problem. Example of trajectory estimation (SLAM) using long-range passive (UHF) RFID

Publications of this project

[1] Philipp Vorst, Artur Koch, and Andreas Zell. Efficient self-adjusting, similarity-based location fingerprinting with passive UHF RFID. In IEEE International Conference on RFID-Technology and Applications (RFID-TA2011), pages 160--167, Sitges, Barcelona, Spain, September 15-16 2011. IEEE. [ DOI | details | pdf ]
[2] Ran Liu, Philipp Vorst, Artur Koch, and Andreas Zell. Path following for indoor robots with RFID received signal strength. In The 19th International Conference on Software, Telecommunications and Computer Networks (SoftCOM 2011), Split, Hvar, and Dubrovnik, Croatia, September 2011. (Best paper award at the Symposium on RFID Technologies and Internet of Things). [ details | pdf ]
[3] Timo Schairer, Benjamin Huhle, Philipp Vorst, Andreas Schilling, and Wolfgang Strasser. Visual mapping with uncertainty for correspondence-free localization using Gaussian process regression. In Proceedings of the 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2011), San Francisco, California, USA, September 2011. Accepted for publication.
[4] Philipp Vorst. Mapping, Localization, and Trajectory Estimation with Mobile Robots Using Long-Range Passive RFID. PhD thesis, University of Tuebingen, Tübingen, Germany, August 2011. [ details | link | pdf ]
[5] Philipp Vorst and Andreas Zell. A comparison of similarity measures for localization with passive RFID fingerprints. In ISR/ROBOTIK 2010 (Proceedings of the joint conference of ISR 2010 (41st International Symposium on Robotics) and ROBOTIK 2010 (6th German Conference on Robotics)), pages 354--361. VDE Verlag, June 2010. [ details | pdf ]
[6] Philipp Vorst and Andreas Zell. Fully autonomous trajectory estimation with long-range passive RFID. In 2010 IEEE International Conference on Robotics and Automation (ICRA), pages 1867--1872, Anchorage, Alaska, USA, May 2010. IEEE. [ DOI | details | pdf ]
[7] Philipp Vorst and Andreas Zell. Particle filter-based trajectory estimation with passive UHF RFID fingerprints in unknown environments. In Proceedings of the 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2009), pages 395--401, St. Louis, Missouri, USA, October 2009. [ DOI | details | pdf ]
[8] Karsten Rohweder, Philipp Vorst, and Andreas Zell. Improved mapping of RFID tags by fusion with spatial structure. In Ivan Petrović and Achim J. Lilienthal, editors, Proceedings of the 4th European Conference on Mobile Robots (ECMR 2009), pages 247--252, Mlini/Dubrovnik, Croatia, September 2009. KoREMA, Zagreb, Croatia. [ details | pdf ]
[9] Philipp Vorst, Bin Yang, and Andreas Zell. Loop closure and trajectory estimation with long-range passive RFID in densely tagged environments. In 14th International Conference on Advanced Robotics (ICAR 2009), pages 1--6, Munich, Germany, June 22-26 2009. [ details | pdf ]
[10] Philipp Vorst, Sebastian Schneegans, Bin Yang, and Andreas Zell. Self-localization with RFID snapshots in densely tagged environments. In Proceedings of the 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2008), pages 1353--1358, Nice, France, September 22-26 2008. [ DOI | details | pdf ]
[11] Timo Schairer, Christian Weiss, Philipp Vorst, Jürgen Sommer, Christian Hoene, Wolfgang Rosenstiel, Wolfgang Straßer, Andreas Zell, Georg Carle, Patrick Schneider, and Anette Weisbecker. Integrated scenario for machine-aided inventory using ambient sensors. In 4th European Workshop on RFID Systems and Technologies (RFID SysTech 2008), number 209 in ITG-Fachbericht, Freiburg, Germany, June 10-11 2008. VDE Verlag. [ details | pdf ]
[12] Philipp Vorst, Jürgen Sommer, Christian Hoene, Patrick Schneider, Christian Weiss, Timo Schairer, Wolfgang Rosenstiel, Andreas Zell, and Georg Carle. Indoor positioning via three different RF technologies. In 4th European Workshop on RFID Systems and Technologies (RFID SysTech 2008), number 209 in ITG-Fachbericht, Freiburg, Germany, June 10-11 2008. VDE Verlag. [ details | pdf ]
[13] Philipp Vorst and Andreas Zell. Semi-autonomous learning of an RFID sensor model for mobile robot self-localization. In Bruno Siciliano, Oussama Khatib, and Frans Groen, editors, European Robotics Symposium 2008, volume 44/2008 of Springer Tracts in Advanced Robotics, pages 273--282. Springer Berlin/Heidelberg, February 2008. [ DOI | details | pdf ]
[14] Sebastian Schneegans, Philipp Vorst, and Andreas Zell. Using RFID snapshots for mobile robot self-localization. In Proceedings of the 3rd European Conference on Mobile Robots (ECMR 2007), pages 241--246, Freiburg, Germany, September 19-21 2007. [ details | pdf ]

Contact

Philipp Vorst, Tel. (+49/0)7071-70441, philipp.vorst at uni-tuebingen.de
Artur Koch, Tel. (+49/0)7071-70441, artur.koch at uni-tuebingen.de

Karsten Bohlmann, Tel. (+49/0)7071-77176, karsten.bohlmann at uni-tuebingen.de

Further details about the research cooperation AmbiSense can be found on the AmbiSense web page.