Design real-time data public transport routing architecture (client and server side) #5332

New issue

Open

opened 2023-06-13 15:11:18 +00:00 by dikkechill · 0 comments

dikkechill commented

2023-06-13 15:11:18 +00:00

(Migrated from github.com)

Ticket to design real-time transport routing architecture and related subtasks

OrganicMaps mostly functions offline and requires a connection for monthly map updates.

Currently public transport routing is in an experimental stage and uses GTFS static data.

The GTFS files are downloaded with a python script
GTFS files are converted with the gtfs_converter tool into JSON
The map generator combines the public transport JSON data with OSM data in .mwm files

With the current approach public transport data can only update with monthly map downloads.
However, routes and schedules change more often, trips are cancelled last minute or unforeseen events happen that cause delays.
OrganicMaps doesn't have this information, which leads to a sub-optimal user experience with possibly incorrect routing advice.

Describe the solution you would like

OrganicMaps can calculate the most optimal route based on the most recent available public transport information.

Ideally the solution allows the OrganicMaps client to update:

more frequent than monthly (originally sourced from GTFS static feed, basically CSV based)
real-time (e.g. 30 second interval) updates (originally sourced from GTFS real time feed, protobuf based)

To do this:

Server collects schedules from different sources. Efficient implementation in golang, rust or C++ is preferred for easier scaling and maintenance.
Different formats are converted to the universal one that can be read by Organic Maps and other clients to allow the following functionality:
- Clients are periodically updating schedule info, that is cached on them and can be used offline, if there is no connection. With the connection real-time info is available.
- Clients build routes from this schedule info offline.
- To scale efficiently, this format should be compact and easily cacheable on HTTP CDN.

Describe alternatives you have considered

Use or cooperate with existing (open source) projects:
- One interesting project is https://travic.app with details on the architecture by @patrickbr
- Navitia open source project
- OpenTripPlanner open source routing with OSM and GTFS (inc. realtime) data
Acquire data from external (commercial) provider
Raises issues with data ownership and costs
- E.g. navitia.io data
Use external routing service(s)
Raises issues with privacy and connectivity
- E.g. Public Transport Enabler

Additional context

Curated list of public transport projects https://github.com/CUTR-at-USF/awesome-transit
Google transit developer docs

_Ticket to design real-time transport routing architecture and related subtasks_ ## Is your feature request related to a problem? Please describe. OrganicMaps mostly functions offline and requires a connection for monthly map updates. Currently public transport routing is in an [experimental stage](https://github.com/organicmaps/organicmaps/blob/master/docs/EXPERIMENTAL_PUBLIC_TRANSPORT_SUPPORT.md) and uses GTFS static data. * The GTFS files are downloaded with a python script * GTFS files are converted with the gtfs_converter tool into JSON * The map generator combines the public transport JSON data with OSM data in .mwm files With the current approach public transport data can only update with monthly map downloads. However, routes and schedules change more often, trips are cancelled last minute or unforeseen events happen that cause delays. OrganicMaps doesn't have this information, which leads to a sub-optimal user experience with possibly incorrect routing advice. ## Describe the solution you would like OrganicMaps can calculate the most optimal route based on the most recent available public transport information. Ideally the solution allows the OrganicMaps client to update: 1. more frequent than monthly (originally sourced from [GTFS static feed](https://gtfs.org/schedule/), basically CSV based) 2. real-time (e.g. 30 second interval) updates (originally sourced from [GTFS real time feed](https://gtfs.org/realtime/), protobuf based) To do this: * Server collects schedules from different sources. Efficient implementation in golang, rust or C++ is preferred for easier scaling and maintenance. * Different formats are converted to the universal one that can be read by Organic Maps and other clients to allow the following functionality: * Clients are periodically updating schedule info, that is cached on them and can be used offline, if there is no connection. With the connection real-time info is available. * Clients build routes from this schedule info offline. * To scale efficiently, this format should be compact and easily cacheable on HTTP CDN. ## Describe alternatives you have considered - Use or cooperate with existing (open source) projects: - One interesting project is https://travic.app with [details on the architecture](http://ad-publications.informatik.uni-freiburg.de/theses/Master_Patrick_Brosi_2014.pdf) by @patrickbr - [Navitia open source project](https://github.com/hove-io/navitia) - [OpenTripPlanner](https://github.com/opentripplanner/OpenTripPlanner/) open source routing with OSM and GTFS (inc. realtime) data - Acquire data from external (commercial) provider Raises issues with data ownership and costs - E.g. [navitia.io data](https://navitia.io/en/nos-donnees/) - Use external routing service(s) Raises issues with privacy and connectivity - E.g. [Public Transport Enabler](https://github.com/schildbach/public-transport-enabler) ## Additional context - Curated list of public transport projects https://github.com/CUTR-at-USF/awesome-transit - [Google transit developer docs](https://developers.google.com/transit)

👍 2

This repo is archived. You cannot comment on issues.