Electrical and Information Technology

Faculty of Engineering, LTH

Denna sida på svenska This page in English


Thesis defence: Dude, Where's My Car? Cellular Navigation for Autonomous Driving

Russ Whiton
Russ Whiton


From: 2024-04-10 09:15 to 13:00
Place: E-house, E:1406.
Contact: russell [dot] whiton [at] eit [dot] lth [dot] se

Russ Whiton defends his thesis: Dude, Where's My Car? Cellular Navigation for Autonomous Driving.

Position and direction estimation is useful for numerous engineering applications for commercial, scientific, and military purposes. Technology that fuses observations of signals broadcast by Global Navigational Satellite Systems (GNSS) with inertial measurements is standard for electronic devices, but this requires at least a periodically unobstructed view of the sky to perform reliably. This has motivated the use of terrestrial radio signals as navigation references to complement or as an alternative to satellite systems, but environments with obstructions, specular reflectors, and scatterers of electromagnetic waves create challenges for any wireless navigation system.

This thesis is about how wireless signals other than those transmitted by GNSS might be used for navigation in complex propagation environments, particularly for safety-critical systems, and conversely how position and orientation information can be used to better understand electromagnetic wave propagation. The thesis is split into introductory chapters that provide broad background on the researched subjects, and five papers published in or submitted to scientific conferences and journals.\par
The first paper offers a broad analysis of the requirements for a cellular navigation system to meet the unique and particularly challenging operating requirements of an autonomous vehicle. Aspects of the problem that are not frequently addressed in literature on terrestrial positioning, particularly requirements for safety-critical operation, are included in the analysis. \par
The second, third, and fourth papers propose methods for position estimation for a passenger vehicle operating in a dense urban canyon environment, tested with a specially-designed measurement system that makes passive (opportunistic), high-resolution observations of down-link synchronization signals transmitted by commercial cellular base stations that are paired with highly accurate pose estimates. Inspired by the prominence and success of Artificial Neural Networks (ANNs) in computer vision, ANNs are used for wireless navigation. The results show that meter-level position estimates and accurate heading estimation can be achieved simultaneously when receiving only reflected and scattered signals from a single transmitter that is never within line-of-sight of the receiving antenna array, relying entirely on multipath propagation.

Finally, in the fifth paper, the link between geometry and multipath propagation is explored from a different perspective. Known navigation states are used to study and characterize multipath propagation. A method for multipath component clustering for statistical channel modeling is proposed, where knowledge about the receiver position is used to gain insight into channel statistics. The algorithm is shown to provide consistent results and to be scalable to large data sets.

Link to thesis i LU Research Portal:

Zoom link. Zoom ID: 69963010473.