I am involved with the OpenTripPlanner project, which is a Java trip planning application that also uses the RAPTOR algorithm! It’s used in cities all over the world, with the biggest deployment being ENTUR’s in Norway, which covers the entire country. I believe all trip planning apps in Norway use this deployment.

It supports many features and has a very active developer community.

Before Citymapper existed, there was OneBusAway, a Ph.D. student project at the University of Washington.

It still exists and powers millions of transit rider trips every day all around the world in Seattle, Washington DC, New York City, Poznan Poland, Buenos Aires Argentina, Adelaide Australia, and who knows where else.

If you’re interested in hacking on something like Citymapper, or setting up an OBA server for your own city, you can find everything you need on our GitHub organization: https://github.com/OneBusAway

That includes docker images, an iOS app and a trip planner framework, android app, Sveltekit web app, and even a next generation OBA server written in Go.

As far as the data to power this, you can get GTFS for every US transit agency from https://mobilitydatabase.org/

(nb I’ve been involved in the OBA project since 2012)

During university, we've built OptiTravel (https://github.com/denysvitali/optitravel) to do something similar. We couldn't use Google Maps APIs (project requirement), so we wrote a custom routing algorithm based on A* and I've created a Rust server to host GTFS data (https://github.com/denysvitali/gtfs-server) à la Transitland (https://transit.land/).

Performance wasn't great since everything had to run locally and do network roundtrips, but it found routes in my hometown that Google Maps didn't show. Pretty cool discovering hidden connections in the transit network and being able to customize your own params (https://github.com/denysvitali/optitravel/blob/master/src/ma...)

Very nice! I've also used the Rail Data Marketplace (terrible name) to build an app that uses the live departure board data (https://apps.apple.com/gb/app/traintrack-uk/id6504205950), and it's great the data is freely available.

Whilst HN is on the case, does anyone know of an API (inside Rail Data Marketplace or elsewhere) that acts as a journey planner for UK national rail, i.e. you can input source + destination stations and it will output journey options?

Surprised not to find Motis and Transitous cited in the article.

Check out api.tfl.gov.uk which should have everything you need for London in a more structured format.

Why are the table and the description of the RAPTOR algorithm in the article images rather than text?

Citymapper et al are excellent. The big issue I find is that half the time TFL seems to completely ignore its own bus schedule (at least in my area). You rush for a bus and it leaves 2mins before schedule (and the next one also turns up early). I would rather longer waits and better predictability.