Fault Detector Sensitivity in Centralized Collaborative Localization

Shinsaku Kuwada, Mathieu Joerger, and Matthew Spenko

Peer Reviewed

Abstract:	Accurate localization in GNSS-denied areas is essential for autonomous ground vehicle safety. Exteroceptive sensors can achieve high-accuracy navigation, but their availability and continuity is limited in automotive environments. Collaborative localization among connected and autonomous vehicles (CAVs) can enhance navigation performance by sharing information on surrounding features, CAV-to-CAV relative pose, and collaborating CAV pose. However, CAVs are vulnerable to sensor faults including misidentifications of landmarks in LiDAR point clouds. This paper builds upon our prior research designing collaborative fault detectors and integrity monitors, which provide probabilistic bounds on CAV pose estimation errors in the presence of nominal sensor errors and undetected faults. Two collaborative approaches are developed using Centralized Extended Kalman Filter (CEKF) and Decorrelation Minimum Variance (DMV). Both approaches improve CAV pose estimation accuracy as compared to non-collaborative navigation. CEKF achieves higher accuracy than DMV. In this paper, we identify the conditions under which CEKF integrity performance exceeds that of DMV; we also explain cases where, using an innovation-based detector under a tight false alert risk requirement, the integrity risk can become larger for CEKF versus DMV. Index Terms—Integrity, Connected Autonomous Vehicle, Kalman Filter, Discorrelated Minimum Variance, fault detection
Published in:	2025 IEEE/ION Position, Location and Navigation Symposium (PLANS) April 28 - 1, 2025 Salt Lake Marriott Downtown at City Creek Salt Lake City, UT
Pages:	173 - 179
Cite this article:	Kuwada, Shinsaku, Joerger, Mathieu, Spenko, Matthew, "Fault Detector Sensitivity in Centralized Collaborative Localization," 2025 IEEE/ION Position, Location and Navigation Symposium (PLANS), Salt Lake City, UT, April 2025, pp. 173-179.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In