Function to compute travel time

[fzenoni]

After our previous discussion I started making comparisons between dodgr and OSRM. And I have to say that this package is most impressive and shows an amazing potential.

However, being more interested in travel time than distance, is there a function similar to dodgr_dists that could take care of this? I couldn't find it myself. If it doesn't exist, would it be a good plan for an enhancement?

[mpadge]

That's great to hear! I'd love it if you could devise and structured and reproducible benchmark of dodgr and OSRM. Even better if you could PR straight into the benchmark vignette (with rendered version here). Let me know if you'd be interested in doing that.

As for your question ... this time a more affirmative answer. There is no such function at the moment, but there definitely should and will be, also as part of #59. The basic infrastructure is all there, because it's primarily just a linear conversion, BUT the missing bit is traffic lights, which are part of the more general issue that dodgr can not route using nodal properties. It currently only uses edge properties (as weighted distances). Incorporating nodal properties in routing calculations is very straightforward, I just have to implement it. This issue can serve to get that done, and I've already put it in the project board, so it should be implemented fairly soon.

[mpadge]

Some useful insights on google travel times here

[fzenoni]

Let me know if you'd be interested in doing that.

It sounds like a great proposal. I am currently on project and I can't commit on extra tasks, but I will certainly let you know if things evolve.

[fzenoni]

In conclusion it looks like dodgr is focusing on providing the shortest route, not the fastest. In comparison, OSRM offers some flexibility in that respect (but not during query-time, currently one needs to recompile the map, see Project-OSRM/osrm-backend#4007).

In the studies I made, fastest route is more interesting, especially for motorized vehicles that suffer of accessibility penalties.

I will nevertheless keep a close watch on dodgr, and I'm looking forward to the next steps.

[mpadge]

Time-based routing started as demonstrated in #72. Two remaining aspects:

• Wait times at traffic lights
• Turn angles

Those are now relatively straightforward and will be implemented soon.

[mpadge]

And actually closing here and moving to duplicate at #88