OpenTrafficRouting

OpenTrafficRouting turns the OpenTrafficDB graph into fast, traffic-aware journeys. It is the routing and navigation brain that sits on top of the data layer, answering the question every mapping product ultimately has to answer: what is the best way from A to B, right now?

The challenge

Commercial routing APIs are costly, opaque and hard to tune — you cannot see or influence how they weigh roads, and every route is a metered call. Building your own is non-trivial: you need efficient path-finding over a very large graph, and you need to fold constantly-changing traffic conditions into the cost model without recomputing the world on every request. Doing that quickly, at scale, from your own infrastructure, is where most self-hosted attempts stall.

The approach

I built a routing engine that reads the OpenTrafficDB tile graph directly and computes routes whose edge costs reflect live traffic, so travel times are realistic rather than free-flow ideals. Around the core algorithm sits the supporting machinery a real service needs:

• Its own map-tile serving so clients can render the network without a third party.
• Background services that keep traffic and derived data fresh.
• A clean REST API and a multi-language web frontend that navigation clients build against.

The outcome

Together with OpenTrafficDB it forms a fully independent, self-hosted alternative to commercial routing stacks — one where the cost model, the data and the infrastructure are all under my control, and where an Android navigation client can be pointed at an API I own end to end.