Towards the Future Generation of Railway Localization Exploiting RTK and GNSS

This repository contains the datasets acquired by ETH-PBL in conjunction with Unibo and SADEL during two days of testing in October 2022 near Modena, Italy.

The data were acquired using two sensor nodes developed by ETH Zurich running a STM32L452CEU6 MCU.
Each node collected data on the motion of the train using an ST ASM330LHH automotive grade IMU as well as a u-blox ZED-F9P GNSS module fed with live RTCM-data from a closeby RTK base station provided by SADEL. The base station utilized another ZED-F9P GNSS module connected to a Raspberry Pi which transmitted the generated RTCM correction packages over a raw TCP socket.
The data was then received using a u-blox SARA-R4 cellular network module.

The track was chosen as it exposes a variety of interesting GNSS environments. Encountered environments are ranging from urban over suburban to open field environments as well as one tunnel. Due to this composition, the availability of cellular connection and thus RTK correction data was patchy but mostly stable.

The two sensor nodes were fixed to the Train Chassis, one centered in the train and the other positioned on the left side in driving orientation. Node 1 was placed on the floor in front of the driver's seat and positioned to be aligned with the center of the train in the lateral direction. A TOPGNSS TOP106 L1/L2 multi-band antenna was placed below the rear-facing windscreen also aligned with the same axis. Node 2 was mounted on a window on the left side of the train when facing in the direction of travel. This is approximately 1m above the floor and 1.4m left to the lateral center of the train. An ANN-MB00 L1/L2 antenna was attached to the outside frame of the train above the window.

This dataset is linked with the GitHub repository at Railway-Precise-Localization where the data format description and the pre-processing scripts are provided.