Cascade-structured Visual-Inertial Navigation System for Aerial Imaginary

Seongho Nam, Taeyun Kim, Juhyun Oh, and Sangwoo Lee

Peer Reviewed

Abstract: A robust and precise navigation algorithm is imperative for the autonomous navigation of unmanned aerial vehicles (UAV). While most navigation algorithms rely on the global positioning system (GPS) and inertial navigation system (INS), vulnerabilities such as jamming and spoofing necessitate alternative solutions for GPS-denied environments. In robotics, visual-inertial navigation systems (VINS), such as visual simultaneous localization and mapping (SLAM) or visual odometry (VO), have been widely studied recently. However, the application of VINS has primarily been explored in ground vehicles and small drones. High-altitude UAVs present unique challenges, including feature tracking difficulties and limitations in low-cost inertial measurement units (IMUs). This paper introduces a factor graph optimization (FGO)-based VINS algorithm designed for precise navigation in high-altitude aerial imagery environments. The algorithm adopts a cascade structure, strategically leveraging the existing INS/GPS navigation solution. Contributions include proposing a relative pose factor instead of IMU preintegration, and introducing a novel constraining factor graph nodes method based on GPS availability to enhance performance by categorizing and fixing accurate nodes during optimization, mitigating navigation errors in GPS outage scenarios. The proposed algorithm excels in robustness during GPS outage scenarios, and simplicity in utilizing existing INS/GPS solutions with minimal modification.
Published in: Proceedings of the ION 2024 Pacific PNT Meeting
April 15 - 18, 2024
Hilton Waikiki Beach
Honolulu, Hawaii
Pages: 143 - 153
Cite this article: Nam, Seongho, Kim, Taeyun, Oh, Juhyun, Lee, Sangwoo, "Cascade-structured Visual-Inertial Navigation System for Aerial Imaginary," Proceedings of the ION 2024 Pacific PNT Meeting, Honolulu, Hawaii, April 2024, pp. 143-153.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In