Localization for autonomous driving

Accurate and reliable determination of the vehicle's location is a key requirement for autonomous driving. We tackle localization using mono-static radar in addition to radio signal measurements to base station as well as cooperative vehicles. At this page you may find our published papers and, if possible, the implementation to generate the corresponding results. Feel free to contact us!

Josef Kulmer
j.kulmer (at) gmx.net
google scholar

Cooperative localization utilizing radar and multipath delay measurements

Conventional localization systems rely on measurements to numerous base stations with known position. In these works we show that multipath propagation offers position information which circumvents the need for base stations. Long story short, localization using absolute coordinates without any base station.

It works with real data.


J. Kulmer, E. Leitinger, S. Grebien, K. Witrisal:
Anchorless cooperative tracking using multipath channel information
IEEE Transactions on Wireless Communications, 17(4):2262–2275, April 2018

In this paper, an algorithm is proposed for joint localization of at least two users. Additionally, inaccuracies in the floorplan are updated along with the user's positions.

How it works (animated gif)

Paper | Proof of concept (video .avi 9MB)

J. Kulmer and K. Witrisal
Revisiting the image source model: Towards geometry-based modeling of agent-to-agent channels
Workshop on Dependable Wireless Communications and Localization for the IoT, Graz 2017

This theoretic work illustrates the relation between multipath parameters and reflective surface orientation/position.
Paper | Slides

J. Kulmer, E. Leitinger, P. Meissner, S. Hinteregger, K. Witrisal:
Cooperative localization and tracking using multipath channel information
2016 International Conference on Localization and GNSS (ICL-GNSS), June 2016

First attempt in tracking two users without base station (it worked). We recognized that the provided floorplan needs to be quite accurate.
Paper | Slides

J. Kulmer, E. Leitinger, P. Meissner, K. Witrisal:
Cooperative Multipath-assisted Navigation and Tracking: A Low-Complexity Approach
1st EAI International Conference on Future access enablers of ubiquitous and intelligent infrastructures, 2015

Comparison between low-complexity Kalman filter and high-complexity particle filter. If there is rich multipath, the Kalman filter does the job.
Paper | Slides

Robust positioning using a single channel snapshot

We were interested to what extend a single measurement to a single base station suffices to get an position accuracy in levels of centimeters. The answer is simple, (1) temporal resolution and (2) spatial resolution in order to separate the signal reflections. We derived a low-complexity algorithm able to perform real-time positioning on a commercial available platform. Only a single base station is requried.

Yes, it works with real data.


J. Kulmer, S. Hinteregger, B. Großwindhager, M. Rath, M. Bakr, E. Leitinger, K. Witrisal
Using DecaWave UWB Transceivers for High-accuracy Multipath-assisted Indoor Positioning
IEEE ICC 2017 Workshop on Advances in Network Localization and Navigation (ANLN), May 2017

We use the DecaWave DW1000 chip to do positioning using a single measurement. A likelihood function is evaluated for getting positions from multipath delays.
Paper | Slides | Implementation and Database of recorded channel measurements using the DecaWave Chip (zip 500KB)

J. Kulmer, S. Grebien, M. Rath, and K. Witrisal
On the unimportance of phase-coherent measurements for beampattern-assisted positioning
2018 IEEE Wireless Communications and Networking Conference (WCNC), April 2018

Here, we exploit an additional spatial domain using directive antennas at the DecaWave Chip. The antennas are switched sequentially and thus, only a single DecaWave Chip is required. Very low-cost and awared.
Paper | Slides | Implementation and Database (zip 18MB)