Center for Applied Research on Supply Chain Services at Fraunhofer IIS
Center for Applied Research on Supply Chain Services at Fraunhofer IIS

Rail siding identification through intelligent geolocalisation - SIDING

Development of private railway sidings for freight loading

SIDING - Schienenanschluss-Identifikation durch intelligente Geolokalisierung
© AdobeStock.com

Sidings are the sources and sinks in the rail network and are therefore essential for the climate-friendly expansion of rail freight transport. Every company that relocates to a siding makes a significant contribution to increasing the share of rail freight transport.

Companies that want to shift their freight transport to rail in the long term are faced with the question when looking for a location: At which locations in Germany is a connection to the rail freight network possible?

This is precisely where the problem lies: there is no official data on the number of private sidings, i.e. sidings not owned by DB, and their location. In the past, these were listed as works sidings, sidings or privately owned sidings. According to an expert estimate by the Association of German Transport Companies (VDV), only 50-70% of private sidings are recorded in the network. This means that companies are faced with a high level of manual effort when searching for suitable land or properties including sidings.

Comprehensive identification of unregistered sidings with AI

The aim of the »SIDING« project is the nationwide mapping of private railway sidings to support the establishment of companies in the industrial and logistics sectors and for the strategic expansion and reactivation of rail infrastructure.

The sidings are identified in a two-stage process. Firstly, the geographically relevant study areas are defined, which are characterised by their proximity to the rail network and commercial or industrial land use. In the second step, the sidings are identified using artificial intelligence methods. Methodologically, automated pattern recognition of satellite images is being developed for this purpose. Using learning and test data, researchers are training a neural network that adapts existing image recognition methods for use in rail infrastructure. The identification results are validated in the project in the Bavarian region.

Automated geodata processing system

The result will be an automated data processing system that uses satellite imagery and georeferenced data to automatically identify railway sidings.

At the end of the project in summer 2023, a list of all private railway sidings in Germany will be made available to the public, along with their spatial location. The research results are also published on the research platform »L.Immo online« of the Working Group for Supply Chain Services, which regularly analyses the most important German logistics locations and visualises them interactively as a geoinformation system.

The SIDING data processing system will also be available to other geographical application areas following the project.