A map matching approach for train positioning. Part I: Development and analysis

Abstract

This paper is devoted to the development of a map matching algorithm for a train positioning system. Due to the nonsteerable truck design associated with rail vehicles, rail track curvatures are reasonably small. Consequently, track signatures, such as curves, are difficult to identify using low-cost sensors. The algorithm proposed in this work takes full advantage of the inherited "one-dimensional" (1-D) train track profile. The core of this algorithm is correlating angular rate extracted from the map database to their corresponding measurements sensed by the yaw gyro and tachometer located on board the vehicle. The sensor fusion required consists of a yaw gyro and any speed sensor or odometer.