Unified multi-object tracking and sensing with mmWave radar

Pearce, Andre

Unified multi-object tracking and sensing with mmWave radar

Pearce, Andre

Permalink

Publication Type:: Thesis
Issue Date:: 2025

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download thesisAdobe PDF (21.17 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Pearce, Andre
dc.date.accessioned	2026-01-29T03:35:41Z
dc.date.available	2026-01-29T03:35:41Z
dc.date.issued	2025
dc.identifier.uri	http://hdl.handle.net/10453/192530
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	Millimetre-wave (mmWave) radars have increasingly become more popular for short-range object tracking due to their high resolution and robustness in various environmental conditions. Although the literature on single object tracking with mmWave radar is well-established, multi-object tracking remains a developing field. The current leading implementations for mmWave multi-object tracking typically operate under assumptions that can either limit their adaptability, performance, or accuracy. The research presented in this thesis aims to enhance the capability and reliability of mmWave multiobject tracking systems, by unifying traditional tracking methodologies with advanced mmWave radar sensing. To ultimately achieve this, the thesis addresses several key challenges. Firstly, the feasibility of improving mmWave multiobject tracking performance through environmental sensing is explored. A novel tracking algorithm is developed to leverage the high-resolution of the mmWave radar to define regional trajectory analysis patterns, improving object detection and tracking performance. Secondly, an approach towards reducing the challenges associated with labelling mmWave radar data is presented, fundamentally achieved through a sensor fusion architecture. This proposed approach dramatically improves the accessibility and feasibility of constructing large scale datasets that can be used to train accurate mmWave radar deep learning systems. Thirdly, the research proposes a generalised framework for a joint mmWave radar sensing and tracking system, incorporating our novel mmWave Convolutional LSTM Autoencoder (mmCLAE) for rain-induced noise reduction. This framework is designed to be inherently adaptable to various tracking scenarios and environmental conditions. The unified framework is validated through an implementation that estimates rainfall with mmWave radar, while jointly incorporating the impact of the rainfall and the noise reduction capabilities of mmCLAE to improve the overall tracking performance. The results of the research conducted demonstrate a promising approach to yield significant improvements in tracking performance, adaptability, and robustness, compared to existing traditional multi-object tracking architectures. The research showcases mmWave multi-object tracking systems that can accurately track multiple objects, even in challenging conditions where objects intermittently leave the radar's field of view or the radar data frames are highly congested with complex noise profiles. The findings of this research have significant implications for a variety of differing applications, including autonomous vehicles, robotics and surveillance systems. This thesis contributes to the advancement of mmWave radar technology by providing a comprehensive study on enhancing the capabilities of mmWave radar systems for multiobject tracking. The unified tracking and sensing system proposed offers a foundation for future research towards more advanced and reliable tracking systems.	en_US.UTF-8
dc.format	Thesis (PhD)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/192530/1/thesis.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	© 2025 Andre Pearce
dc.rights	au.edu.uts.lib/nph
dc.title	Unified multi-object tracking and sensing with mmWave radar	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

Millimetre-wave (mmWave) radars have increasingly become more popular for short-range object tracking due to their high resolution and robustness in various environmental conditions. Although the literature on single object tracking with mmWave radar is well-established, multi-object tracking remains a developing field. The current leading implementations for mmWave multi-object tracking typically operate under assumptions that can either limit their adaptability, performance, or accuracy. The research presented in this thesis aims to enhance the capability and reliability of mmWave multiobject tracking systems, by unifying traditional tracking methodologies with advanced mmWave radar sensing. To ultimately achieve this, the thesis addresses several key challenges. Firstly, the feasibility of improving mmWave multiobject tracking performance through environmental sensing is explored. A novel tracking algorithm is developed to leverage the high-resolution of the mmWave radar to define regional trajectory analysis patterns, improving object detection and tracking performance. Secondly, an approach towards reducing the challenges associated with labelling mmWave radar data is presented, fundamentally achieved through a sensor fusion architecture. This proposed approach dramatically improves the accessibility and feasibility of constructing large scale datasets that can be used to train accurate mmWave radar deep learning systems. Thirdly, the research proposes a generalised framework for a joint mmWave radar sensing and tracking system, incorporating our novel mmWave Convolutional LSTM Autoencoder (mmCLAE) for rain-induced noise reduction. This framework is designed to be inherently adaptable to various tracking scenarios and environmental conditions. The unified framework is validated through an implementation that estimates rainfall with mmWave radar, while jointly incorporating the impact of the rainfall and the noise reduction capabilities of mmCLAE to improve the overall tracking performance. The results of the research conducted demonstrate a promising approach to yield significant improvements in tracking performance, adaptability, and robustness, compared to existing traditional multi-object tracking architectures. The research showcases mmWave multi-object tracking systems that can accurately track multiple objects, even in challenging conditions where objects intermittently leave the radar's field of view or the radar data frames are highly congested with complex noise profiles. The findings of this research have significant implications for a variety of differing applications, including autonomous vehicles, robotics and surveillance systems. This thesis contributes to the advancement of mmWave radar technology by providing a comprehensive study on enhancing the capabilities of mmWave radar systems for multiobject tracking. The unified tracking and sensing system proposed offers a foundation for future research towards more advanced and reliable tracking systems.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/192530