[jillesvangurp]

I looked into the gtfs (gtfs.org) a while ago. This is a data feed format (protobuf based) that a lot of sites like this use for both schedules and real time updates that is pretty widely adopted. Most gtfs location feeds require getting an API key. But there are a handful of public ones that share actual location updates without that.

This is also how e.g. Google maps and others integrate schedule information for public transport as well.

Building a map from the real time updates is relatively straight foward once you have access.

[dan_sbl]

GTFS is widely adopted, "standardized", but also quite a wild west when it comes to the actual details. I've been building http://mobility-bot.com/ as a spare time project for two years, parsing the alerts GTFS feeds from a handful of providers, and for just about every one of them, I have to do something in the pipeline to clean up the provided data - fixing unicode issues, remapping route and agency IDs, rewriting headers and descriptions to not be identical, etc. You also have to watch and capture the data over time to make sure you have a corpus to generalize parsing and cleanup rules from.

Most of these projects made the smart choice to focus on a single or just a small handful of related providers. As soon as you're trying to aggregate, the problem becomes a lot trickier.