Robust Localization of Autonomous Vehicles Using GNSS, IMU and 3D LiDAR in the Presence of GNSS Outage
Sabbih, Rehan Tahir (2024)
2024
Automaatiotekniikan DI-ohjelma - Master's Programme in Automation Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2024-12-02
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-2024112110394
https://urn.fi/URN:NBN:fi:tuni-2024112110394
Tiivistelmä
Localization is a foundational aspect of autonomous vehicle operation, crucial for ensuring smooth and effective navigation. However, environmental uncertainties and sensor limitations, such as GNSS outages, can compromise its reliability.
This thesis investigates robust localization strategies for autonomous vehicles using GNSS, IMU, and 3D LiDAR while facing GNSS outage challenges. It explores various fusion architec-tures, ultimately selecting the dual factor graph-based Graph MSF after comparative analysis. It further discusses the importance of fusing accurate LiDAR odometry during GNSS outage by experimentally comparing full fusion with a setup lacking LiDAR odometry, and quantifying the impact on Absolute Pose Error (APE) and Relative Pose Error (RPE).
Using similar performance evaluation metrics, the thesis examines factors influencing locali-zation accuracy during GNSS outages, specifically addressing the effects of IMU intrinsic cal-ibration, GNSS heading accuracy, sensor misalignment, and sensor data integrity using Graph MSF. These insights provide valuable guidance for diagnosing localization issues. Finally Graph MSF’s performance is evaluated under various GNSS outage and jumping scenarios to assess its practical viability and computational resource efficiency. Results indicate that while Graph MSF delivers seamless localization during GNSS outages, its precision depends on factors such as the duration of the outage, the quality of odometry fused, and sensor calibra-tion accuracy. For applications requiring very high precision over extended GNSS outages, improved odometry or more advanced fusion architectures are necessary.
In summary, this thesis sheds light on robust localization during GNSS outages, discusses suitable architectures in the literature, and examines the factors influencing localization accu-racy using the selected fusion architecture along with its own performance analysis. Hence, it provides a comprehensive overview of the challenges in achieving robust localization during GNSS outages.
This thesis investigates robust localization strategies for autonomous vehicles using GNSS, IMU, and 3D LiDAR while facing GNSS outage challenges. It explores various fusion architec-tures, ultimately selecting the dual factor graph-based Graph MSF after comparative analysis. It further discusses the importance of fusing accurate LiDAR odometry during GNSS outage by experimentally comparing full fusion with a setup lacking LiDAR odometry, and quantifying the impact on Absolute Pose Error (APE) and Relative Pose Error (RPE).
Using similar performance evaluation metrics, the thesis examines factors influencing locali-zation accuracy during GNSS outages, specifically addressing the effects of IMU intrinsic cal-ibration, GNSS heading accuracy, sensor misalignment, and sensor data integrity using Graph MSF. These insights provide valuable guidance for diagnosing localization issues. Finally Graph MSF’s performance is evaluated under various GNSS outage and jumping scenarios to assess its practical viability and computational resource efficiency. Results indicate that while Graph MSF delivers seamless localization during GNSS outages, its precision depends on factors such as the duration of the outage, the quality of odometry fused, and sensor calibra-tion accuracy. For applications requiring very high precision over extended GNSS outages, improved odometry or more advanced fusion architectures are necessary.
In summary, this thesis sheds light on robust localization during GNSS outages, discusses suitable architectures in the literature, and examines the factors influencing localization accu-racy using the selected fusion architecture along with its own performance analysis. Hence, it provides a comprehensive overview of the challenges in achieving robust localization during GNSS outages.