Resilient Multi-Sensor Navigation for Autonomous Vehicles: Leveraging Onboard Motion Sensors, Perception Systems, and 3D Digital Maps
Loading...
Authors
Abdelmoneem, Eslam Mounier
Date
2025-05-30
Type
thesis
Language
eng
Keyword
Multi-sensor fusion , Navigation , Positioning , Inertial Sensors , LiDAR , Radar , 3D Digital Maps
Alternative Title
Abstract
Autonomous vehicles (AVs) are expected to transform the future of mobility by improving transportation efficiency, reducing human error, and expanding accessibility. To ensure safe and efficient operation of core guidance and control functionalities, AVs require continuous, high-accuracy, and resilient navigation capabilities amid diverse and challenging environmental and operational conditions.
This thesis aims to advance the design of accurate, robust, and cost-effective navigation solutions for land-based AVs. To that end, it proposes several complementary methodologies targeting key challenges in sensor error mitigation, fault resilience, and multi-sensor integration, with a focus on ensuring the continuity and reliability of the navigation solution.
The research begins by addressing inertial sensor errors through two complementary approaches: a Fast Orthogonal Search (FOS)-based technique for online modeling of all Inertial Measurement Unit (IMU) channels and a supervised Deep Learning (DL)-based method focused on gyroscope error correction. A key innovation lies in the use of an inverse mechanization algorithm to generate synthetic ground truth data, enabling effective error modeling as reflected in the improved performance of standalone inertial navigation. To enhance resilience against sensor faults, a multi-IMU navigation system was developed, integrating a bank of Kalman Filters (KFs) with a Differential Evolution (DE)-based optimization strategy. This system not only improved navigation accuracy in nominal conditions but also demonstrated robust performance through rapid fault detection and isolation.
To provide alternative aiding information, particularly in Global Navigation Satellite System (GNSS)-challenged environments, the use of high-accuracy Three-Dimensional (3D) digital city maps in conjunction with Light Detection and Ranging (LiDAR) technology was proposed. An Extended Kalman Filter (EKF) was employed to fuse this information with Onboard Motion Sensors (OBMS), achieving sustained high-precision positioning while preserving computational efficiency. Additionally, a multi-modal odometry system was developed to leverage the complementary characteristics of LiDAR and Radio Detection and Ranging (RADAR), maintaining reliable navigation in highly dynamic environments, particularly in scenarios lacking external corrections.
All proposed methods were validated using real-world data collected across diverse urban and indoor driving scenarios. Collectively, the contributions of this thesis represent a significant advancement toward resilient and high-precision navigation systems, enabling more practical, scalable, and cost-effective AV operations.
Description
Citation
Publisher
License
Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada
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-NonCommercial-NoDerivatives 4.0 International
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-NonCommercial-NoDerivatives 4.0 International