Route changed why?

Open Source Routing Engine
1

EE647651-0497-4A87-BA25-3425563FD34C
EE647651-0497-4A87-BA25-3425563FD34C820×1013 79.5 KB

{
  "points": [
    [
      12.57840,
      48.85840
    ],
    [
      12.56120,
      48.90560
    ]
  ],
  "snap_preventions": [
    "motorway",
    "ferry",
    "tunnel"
  ],
  "details": [
    "road_class",
    "surface"
  ],
  "profile": "car",
  "locale": "en",
  "instructions": true,
  "calc_points": true,
  "points_encoded": false
}

Im wondering why the route changed. It was shorter like just 7km straight through the city, not its using the highway 12km.

2

When is the last time you remember the shorter 7km long route was chosen? It still appears as alternative: GraphHopper Maps | Route Planner

3

@easbar must be as least the 11th of september last year.

4

Generally, this can be due to changes in OSM data or data modelling on our side. Is there a specific reason why you’d prefer one route over the other and are there more cases of this kind?

5

Not sure if this is related but for some reason Google does not allow branching off B8 to the right: https://maps.app.goo.gl/jaCEjiDwazBGHmjE7. Apparently it assumes this is a one way road (indicated by a little arrow on the map), even though according to street view it is not: https://maps.app.goo.gl/ZZ1rdWuJ9Nf1LhYa8

6

I would prefer this route because its shorter, while taking just a bit more time. Would consider this as the more efficient route.

I saw the difference in google as well, but this must be a mistake.

7

You can set a preference for shorter routes using the distance_influence parameter like here

The corresponding POST request would look like this:

{
  "points": [
    [
      12.57840,
      48.85840
    ],
    [
      12.56120,
      48.90560
    ]
  ],
  "snap_preventions": [
    "motorway",
    "ferry",
    "tunnel"
  ],
  "details": [
    "road_class",
    "surface"
  ],
  "profile": "car",
  "locale": "en",
  "instructions": true,
  "calc_points": true,
  "points_encoded": false,
  "ch.disable": true,
  "custom_model": {
    "distance_influence": 30
  }
}

However, this requires disabling CH which affects the response time.