Autonomous Vehicle Localization Using Automotive Radar and Reflective Lane Markers
Loading...
Authors
Sacoransky, Dean
Date
Type
thesis
Language
eng
Keyword
Radar , Autonomous Vehicles
Alternative Title
Abstract
This thesis presents the formulation and experimental validation of an autonomous ground vehicle localization method that applies a low-cost, millimetre-wave radar per- ception system for the detection and tracking of roadside retro-reflectors. Radar has emerged as an essential component in the autonomous vehicle sensor suite due to its low cost, resiliency to inclement weather conditions, and its benefit in direct mea- surement of relative velocity. Many modern vehicles are equipped with pre-installed automotive radar units for Advanced Driver Assistance Systems (ADAS) features such as adaptive cruise control and collision detection systems. This work aims to leverage existing low-cost radar infrastructure by developing software-based techniques for ve- hicle localization. This approach has the potential to eliminate the need for hardware modifications, thereby enhancing the practicality, cost-effectiveness, and accessibility of the proposed localization method.
There are factors that make radar-based perception systems challenging, such as the sparsity of radar point clouds, the impact of motion distortion on radar sensor performance, and the event of missed and false detections. To overcome these chal- lenges, we propose a feature extraction and fusion scheme that uses Density Based Spatial Clustering of Applications with Noise (DBSCAN) and a Kalman filter. This system effectively tracks radar returns attributed to retro-reflectors, enabling accurate vehicle localization with respect to the road ahead.
We tested the approach via indoor experiments that made use of Continental’s ARS 408 radar, a mobile Husky A200 robot, and a Vicon motion capture system for ground truth validation. Experiments were conducted to test the localization system on curved and straight paths to resemble highway and highway-ramp driving scenarios. The results from 30 experiments showed centimetre-level localization accuracy, with an average Euclidean distance tracking error of 0.25 m for each reflector.
Description
Citation
Publisher
License
Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada
ProQuest PhD and Master's Theses International Dissemination Agreement
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.
Attribution 4.0 International
ProQuest PhD and Master's Theses International Dissemination Agreement
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.
Attribution 4.0 International