{"id":{"repo_id":"queens","oai_identifier":"oai:queensu.scholaris.ca:1974/29826"},"canonical_url":"https://search.dev.ndltd.org/etd/queens/oai:queensu.scholaris.ca:1974/29826","repository":{"repo_id":"queens","name":"Queens University","base_url":"https://qspace.library.queensu.ca/server/oai/request"},"display":{"title":"Path Following using Frequency Modulated Continuous Wave millimetre-wave Automotive Radar","abstract":"This thesis presents the discovery, methodology, and experimental validation of a path planning algorithm developed to only use Frequency Modulated Continuous Wave (FMCW) radar. Radar has shown to be semi-invariant to inclement weather such as snow, fog, extremely lighting conditions, and heavy rain. Traditionally, autonomous vehicle (AV) path planning algorithms use multiple introceptive and exteroceptive sensors to first determine the current position of the vehicle, then plan a path based on that position. These sensors include LiDAR, GNSS, cameras, inertial measurement units, ultrasonic, and radar. In previous research, the use of radar in these systems is to aid with the detection of obstacles, which could then be used to determine paths, not as a single modality system. This thesis presents research into using FMCW automotive radar with the aid of radar retro-reflectors to test the viability of a single modality path planning algorithm. The result is an efficient, Gated Recurrent Unit (GRU) deep neural network that has been trained on custom synthetically trained data of a vehicle following a path of retro-reflectors. The GRU network was then experimentally validated using a 60 kg skid-steer robot showing promising path planning results while in heavy fog. The results showed an average accuracy of 0.30 ± 0.15 m. When compared to LiDAR during the same testing conditions, the visual based sensor was unable to produce a viable path. As a result, this system satisfies the requirement for centimetre level accuracy for Level 3 autonomous driving in inclement weather.","abstract_html":"This thesis presents the discovery, methodology, and experimental validation of a path planning algorithm developed to only use Frequency Modulated Continuous Wave (FMCW) radar. Radar has shown to be semi-invariant to inclement weather such as snow, fog, extremely lighting conditions, and heavy rain. Traditionally, autonomous vehicle (AV) path planning algorithms use multiple introceptive and exteroceptive sensors to first determine the current position of the vehicle, then plan a path based on that position. These sensors include LiDAR, GNSS, cameras, inertial measurement units, ultrasonic, and radar. In previous research, the use of radar in these systems is to aid with the detection of obstacles, which could then be used to determine paths, not as a single modality system. This thesis presents research into using FMCW automotive radar with the aid of radar retro-reflectors to test the viability of a single modality path planning algorithm. The result is an efficient, Gated Recurrent Unit (GRU) deep neural network that has been trained on custom synthetically trained data of a vehicle following a path of retro-reflectors. The GRU network was then experimentally validated using a 60 kg skid-steer robot showing promising path planning results while in heavy fog. The results showed an average accuracy of 0.30 ± 0.15 m. When compared to LiDAR during the same testing conditions, the visual based sensor was unable to produce a viable path. As a result, this system satisfies the requirement for centimetre level accuracy for Level 3 autonomous driving in inclement weather.","abstract_has_math":false,"creators":["Greisman, Austin"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":["Hashtrudi-Zaad, Keyvan","Marshall, Joshua"],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-07-27T20:35:41Z","subjects":["Radar","mmWave","Localization","Autonomous Vehicle","FMCW","Machine Learning","GRU"],"languages":["eng"],"rights":["Attribution-NonCommercial 3.0 United States"],"rights_urls":["http://creativecommons.org/licenses/by-nc/3.0/us/"],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/1974/29826","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:contributor.supervisor","label":"Supervisor","values":["Hashtrudi-Zaad, Keyvan","Marshall, Joshua"]},{"key":"dc:creator","label":"Author","values":["Greisman, Austin"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2021-12-07T15:28:56Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2021-12-07T15:28:56Z"]},{"key":"dc:type","label":"Dc Type","values":["thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Radar","mmWave","Localization","Autonomous Vehicle","FMCW","Machine Learning","GRU"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution-NonCommercial 3.0 United States"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://creativecommons.org/licenses/by-nc/3.0/us/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/1974/29826"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This thesis presents the discovery, methodology, and experimental validation of a path planning algorithm developed to only use Frequency Modulated Continuous Wave (FMCW) radar. Radar has shown to be semi-invariant to inclement weather such as snow, fog, extremely lighting conditions, and heavy rain. Traditionally, autonomous vehicle (AV) path planning algorithms use multiple introceptive and exteroceptive sensors to first determine the current position of the vehicle, then plan a path based on that position. These sensors include LiDAR, GNSS, cameras, inertial measurement units, ultrasonic, and radar. In previous research, the use of radar in these systems is to aid with the detection of obstacles, which could then be used to determine paths, not as a single modality system. This thesis presents research into using FMCW automotive radar with the aid of radar retro-reflectors to test the viability of a single modality path planning algorithm. The result is an efficient, Gated Recurrent Unit (GRU) deep neural network that has been trained on custom synthetically trained data of a vehicle following a path of retro-reflectors. The GRU network was then experimentally validated using a 60 kg skid-steer robot showing promising path planning results while in heavy fog. The results showed an average accuracy of 0.30 ± 0.15 m. When compared to LiDAR during the same testing conditions, the visual based sensor was unable to produce a viable path. As a result, this system satisfies the requirement for centimetre level accuracy for Level 3 autonomous driving in inclement weather."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["M.A.Sc."]},{"key":"dc:title","label":"Title","values":["Path Following using Frequency Modulated Continuous Wave millimetre-wave Automotive Radar"]}]}],"canonical_facts":{"dc:contributor.department":["Electrical and Computer Engineering"],"dc:contributor.supervisor":["Hashtrudi-Zaad, Keyvan","Marshall, Joshua"],"dc:creator":["Greisman, Austin"],"dc:date.accessioned":["2021-12-07T15:28:56Z"],"dc:date.available":["2021-12-07T15:28:56Z"],"dc:description.abstract":["This thesis presents the discovery, methodology, and experimental validation of a path planning algorithm developed to only use Frequency Modulated Continuous Wave (FMCW) radar. Radar has shown to be semi-invariant to inclement weather such as snow, fog, extremely lighting conditions, and heavy rain. Traditionally, autonomous vehicle (AV) path planning algorithms use multiple introceptive and exteroceptive sensors to first determine the current position of the vehicle, then plan a path based on that position. These sensors include LiDAR, GNSS, cameras, inertial measurement units, ultrasonic, and radar. In previous research, the use of radar in these systems is to aid with the detection of obstacles, which could then be used to determine paths, not as a single modality system. This thesis presents research into using FMCW automotive radar with the aid of radar retro-reflectors to test the viability of a single modality path planning algorithm. The result is an efficient, Gated Recurrent Unit (GRU) deep neural network that has been trained on custom synthetically trained data of a vehicle following a path of retro-reflectors. The GRU network was then experimentally validated using a 60 kg skid-steer robot showing promising path planning results while in heavy fog. The results showed an average accuracy of 0.30 ± 0.15 m. When compared to LiDAR during the same testing conditions, the visual based sensor was unable to produce a viable path. As a result, this system satisfies the requirement for centimetre level accuracy for Level 3 autonomous driving in inclement weather."],"dc:description.degree":["M.A.Sc."],"dc:identifier.uri":["http://hdl.handle.net/1974/29826"],"dc:language.iso":["eng"],"dc:rights":["Attribution-NonCommercial 3.0 United States"],"dc:rights.uri":["http://creativecommons.org/licenses/by-nc/3.0/us/"],"dc:subject":["Radar","mmWave","Localization","Autonomous Vehicle","FMCW","Machine Learning","GRU"],"dc:title":["Path Following using Frequency Modulated Continuous Wave millimetre-wave Automotive Radar"],"dc:type":["thesis"]},"updated_at":"2026-07-27T20:35:41Z"}