Case Study on Factor Graph Optimization in Dependable GNSS Train Localization

Dependable train localization has become an important issue for GNSS (Global Navigation Satellite Systems)-based train localization in both academic research and industry localization unit development towards Next Generation Train Control (NGTC), aiming at achieving GNSS to be the dominant source for onboard train localization. RAMS (Reliability, Availability, Maintainability, and Safety) requirement of the electronic equipment for railway train control system is the basic standard. Recent advances in robust graph theories for measurement optimization have been alternative methods for accuracy improvement beyond traditional least squares and Kalman filters. This paper utilizes the factor graph optimization (FGO) framework to bring GNSS accuracy improvement comparing with weighted least squares (WLS). The accuracy improvement of the FGO is evaluated using open access data via smartphone by Google and the high-speed railway field test via GNSS receiver by our team. The result demonstrates the dependable GNSS train localization in considerable aspects in both accuracy and reliability.