The REAKT DATA project (https://reakt.sh/) includes visualizing rail vehicle positions on the track based on a set of position data from GNNS trackers. These position data are sparse and include errors and variation, yet it can be safely assumed that the rail vehicle will not leave the track in normal operation. Hence, this topic will investigate an algorithm to process the data and project the position of the vehicles onto the track.

!image-20240226113213-1.png

Goals

• Develop and implement an algorithm for computing a best-effort projection of a vehicle position on a track
• Evaluate the algorithm (and potential variants of the algorithm) with the real-world data collected on the Malente-Lütjenburg track

Consiterations for the Algorithm

• Errors in GNNS data
• Gaps in GNNS data (missing network)
• Different accuracy of tracking devices (Smartphones vs. dedicated Trackers)
• Showing estimates of expected movement or potential location areas
• Handling inconsistencies
  • Unrealistic position jumps
  • Passing other vehicles (quickly) on single-lane track
  • Vehicles changing direction without stopping

Scope

Bachelor's Thesis

Related Work/Literature

https://reakt.sh/ *

https://github.com/kieler/RailTrail/wiki/Data-processing#computation-of-vehicle-position-speed-and-heading

Involved Languages/Technologies

Python

Supervised by

Alexander Schulz-Rosengarten
als@informatik.uni-kiel.de