Sensor data fusion for inertial navigation of trains in GPS-dark areas

The motivation of the presented work is to develop a robust navigation system for accurate localization of trains on railway tracks in the cases where the GPS-based navigation is not temporally available. As the final solution of the train locator naturally takes into consideration the satellite-based navigation, the satellite signal needs not to be available all along the railway. The presented contribution describes an approach to preprocessing and fusing of additional train onboard sensors - the odometer and accelerometer, all targeted to serve as optional and temporary substitution for GPS navigation. The suggested solution is exploring a rule-based substitutions and mutual calibrations of the used sensors depending on actual conditions. The given approach relies on the GPS system only to provide position calibrations and serves a reference method for evaluation of the presented results. Suggested solutions have been experimentally tested with real-world data gathered with locomotive onboard inertial sensors.