Set-Membership Position Estimation With GNSS Pseudorange Error Mitigation Using Lane-Boundary Measurements

Bento, Luis Conde; Bonnifait, Philippe; Nunes, Urbano J. · 2019 · Crossref

DOI: 10.1109/tits.2018.2808542

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of achieving high-confidence, accurate positioning for autonomous vehicles, particularly in environments where standalone low-cost Global Navigation Satellite System (GNSS) receivers are insufficient due to nonlinearities and error susceptibility. The authors propose a novel algorithm called Lane Boundary Augmented Set-Membership GNSS Positioning (LB-ASGP). Unlike classical estimation methods that may converge to local optima or fail to guarantee solution integrity, LB-ASGP utilizes set-membership theory and interval analysis to provide a guaranteed enclosure of the true vehicle position within a characterized confidence domain. The method fuses raw GNSS pseudoranges with lane-boundary measurements derived from local perception systems and 2D road network maps, eliminating the need for vehicle-to-everything (V2X) communication. The methodology relies on set inversion via interval analysis (SIVIA) to solve the nonlinear positioning equations. The algorithm models GNSS pseudorange errors—including ionospheric, tropospheric, and multipath delays—as bounded intervals based on a chosen integrity risk. It incorporates geometric constraints from geo-referenced lane boundaries, specifically the perpendicular distance to lane markings and the transverse unit vector. The LB-ASGP algorithm operates in two stages: first, it computes a solution set using only GNSS data (SGP); second, it refines this set by applying lane-boundary constraints to reduce the search space, particularly in the direction perpendicular to the lane. This approach ensures that no valid solution is missed within the initial search box, providing rigorous bounds on satellite-specific errors. The performance of LB-ASGP was evaluated through simulations using the ISR-TRAFSIM environment and GPSoft for GNSS emulation, as well as field experiments. The simulations modeled urban road networks with realistic error characteristics, including ionospheric delays averaging 4 meters. Results demonstrated that LB-ASGP significantly reduces positioning errors in the cross-track direction (perpendicular to the lane) compared to the standard Set Membership GNSS Positioning (SGP) algorithm and iterative least squares methods. The algorithm successfully bounded the solution set with high confidence, effectively mitigating GNSS pseudorange errors. Additionally, the method provided an estimate of satellite-specific errors along the transverse direction, further enhancing localization integrity. The significance of this work lies in its ability to provide guaranteed positioning accuracy using inexpensive sensors without relying on external communication infrastructure. By leveraging set-membership algorithms, the approach offers a robust alternative to probabilistic methods, ensuring that the true position is contained within the estimated domain up to a defined risk level. This capability is critical for the safety and reliability of autonomous driving systems, particularly in scenarios where communication links are unavailable or GNSS signals are degraded. The study confirms that integrating lane-boundary constraints into set-membership frameworks can achieve sufficient accuracy for driverless cars, complementing other sensor data like LIDAR.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• lane positioning