Help improving long routes

Hi, just discover this great project!

I didn’t spend a lot of time playing with it, but my use case involves computing long distance car routes in France, and I have a case that I would like to understand if I can improve.

This planning on Googlemaps finds routes that are not “in the axis”, 2 of them beeing ~800km and ~8h long.

I ran the same planning on Graphhopper online , and it’s correct.

But it’s not what I get locally. I have a very simple setup that follows the deploy instructions, using a recent france.osm.pbf file from geofabrik

Do you have a hint on what I could improve to match the results from both online sources?

  • When I use alternatives = 10, I end up with a correct route, but it’s much slower for my usage.
  • I can note a “toll” hint on Graphhopper online, maybe I miss data on highways? But if I force a point on the highway, then the first result is correct

Thanks a lot

I may have overlooked distance_influence: I thought the default profile was 100% time based, but it seems it’s not the case. I’ll give it a try with a profile with this setting to 0, and I’ll report back :saluting_face:

We still use a slightly different car profile for our online services. But you can achieve the same also with the open source project. The biggest change requires to consider a speed limit inner city. For that you use the urban_density. (For backward compatibility we still use a proprietary mechanism for this “urban_density” mechanism and we cannot release this as open source - hence the differences.)

Hi, and thanks for the quick answer! I understand the online service has clever tricks, but in the end using distance_influence = 0 did a perfect job for me, so I’m quite happy with it for now.

Have a great day,

1 Like