This repository contains the results of the Master Project 2022/23 "Dispatching Trains from the Cloud". It is divided into a few subprojects:
code_generation: A program that given a JSON file from the interlocking exporter generates interlocking code in Rust using thetrack_elementcrate.grpc_control_station: A frontend for the generated interlocking that can be used with the EULYNX Live Lab UI.ui: A web-based application to generate interlockings from OpenRailwayMap data.
The subprojects all contain a README with local setup instructions. To run the project as whole, you can use the provided dockerfile by running the following commands from the project root:
docker build -t instant-interlocking .
docker run --net host instant-interlockingNote that the --net host is mandatory as we are starting the interlocking instances on new ports.
You can reach the UI via http://localhost:5000
Even though digital tools are available, most railway planning still happens with analog methods. Digital planning data generated from publicly available data sources could serve as a basis for new building projects. This project explores a pipeline that transforms geodata from public sources into planning data and generates executable interlocking code from this planning data.
This project presents a pipeline for generating executable interlocking code from planning data.
The pipeline starts by extracting planning information from geodata (e.g. OpenRailwayMap). This planning
data (which includes information such as where tracks, points, and signals are located) is then used to
generate possible driveways and enriched with the information what states track elements have to be in
for a driveway to be set. We then use this enriched planning data to create executable Rust code that
uses the track_element crate.
Beyond this step, two independent projects explored the viability of using WebAssembly to test and deploy the generated code.
yaramo is a python library used to extract the planning data from a source file, generate driveways and other information and finally export topology, placement and route json files used by the interlocking.
When we started this project, there have been a few previous projects and repositories around digital rail that were developed at the OSM chair at HPI Potsdam. These were mainly centered around the usage of PlanPro as a digital rail topology data format and its usage for simulation. For our use case, we needed to extend these existing projects, e.g. to enrich routes with their maximum speed. This was very cumbersome, as there was no common python representation between the projects and each needed to be adapted at several points. This is also due to the fact that PlanPro is a XML-based format, which makes it suitable for data exchange, but nearly impossible as a object representation in a programming language, especially because there are a lot of indirections. That's why we decided to develop a common python representation for the topology data and adapt the existing tools to work with it. This allowed us to easily extend the model, develop new tools and flexibilize the use of the tools along our pipeline. In the list of tools below, tools that have been mainly developed by us and weren't pre-existing are marked with an asterisk.
yaramo works as an exchange format where there can be an importer turning other formats into yaramo and one or more exporters turning the yaramo object into their specific formats. Moreover, there the yaramo topology can by modified to do things with the planning data, like finding driveways.
The following projects operate on the yaramo library. Some of them are part of the instant-interlocking pipeline.
Importers
Modifiers
Exporters
The instant-interlocking pipeline uses the ORM-Importer to retrieve the planning data. Both the RailwayRoute-Generator and the VacancySection-Generator are used to enrich the yaramo object with relevant vacancy section and driveway data. Finally, the Interlocking-Exporter is used to generate the three json files which are the used to configure the interlocking and the UI.