Relative Train Localization for Cooperative Maneuvers using GNSS Pseudoranges and Geometric Track Information

Cooperative maneuvers between trains like platooning or dynamic merging and splitting of train sets have the potential to increase efficiency and flexibility in railway operations. However, such maneuvers are safety critical and require continuously accurate relative position and velocity information. This information cannot be provided by infrastructure based train control systems like the European Train Control System, which is limited both in accuracy and update rate. Since the braking distance of trains often exceeds the visible track length, line-of-sight sensors alone are not sufficient for merging and splitting in all situations. We therefore present a Bayesian estimation algorithm that jointly estimates absolute and relative train positions and velocities based on GNSS pseudo-range measurements and track map information for increased positioning performance. By transmitting GNSS pseudorange measurements between vehicles, common measurement errors can be eliminated, yielding better relative positioning accuracy. In simulations we show that the proposed algorithm achieves reduced absolute and relative positioning errors compared to algorithms using preprocessed GNSS positions and velocities as measurement input.