[GraphHopper] Unreachable addresses by car

Peter graphhopper at gmx.de
Wed Dec 17 08:00:53 UTC 2014


Hi David,

more precise:

> prepare.minOnewayNetworkSize = 1
> prepare.minNetworkSize = 1

this both can be 200 and will remove all disconnected roadnetworks with
less than 200 nodes.

> index.highResolution=10000
> index.maxRegionSearch=10000

this is not so good. For maximum speed use e.g.

> index.highResolution=100
> index.maxRegionSearch=4

For max 'fault tolerance' use:

> index.highResolution=1000
> index.maxRegionSearch=8

which will search approx 1000m around the query AND try around this
location until it reaches roughly 8*1000m!

Regards,
Peter



On 16.12.2014 23:08, David Ödling wrote:
> Looks good, but having played around with the settings I still do not
> get them to work. My current config looks as follows:
>
> prepare.chWeighting=fastest
> prepare.minOnewayNetworkSize = 1
> prepare.minNetworkSize = 1
> graph.flagEncoders=car
> index.highResolution=10000
> index.maxRegionSearch=10000
> routing.defaultAlgorithm=dijkstrabi
>
> I believe I’m doing something wrong as the demo-api manages all the
> routes. 
>
> I’’m thankful for all suggestions! 
>
> //
>
> David
>
>> 16 dec 2014 kl. 15:08 skrev Peter <graphhopper at gmx.de
>> <mailto:graphhopper at gmx.de>>:
>>
>> See if this answer helps:
>> http://stackoverflow.com/a/26433076/194609
>>
>> Peter
>>
>> On 16.12.2014 15:03, David Ödling wrote:
>>>
>>> Hi!
>>> I have encountered a problem in finding routes between a few number
>>> of addresses. Most work well, but it seems like some addresses on
>>> one way streets and in areas with restricted access it fails to find
>>> a route. I have implemented a quick-fix for this by adjusting the
>>> coordinates in a outgoing  spiral until a route is found. But this
>>> is not a good solution in densely populated areas, so,  is there any
>>> better solution?
>>>
>>> Hopefully, I’m simply doing something wrong by getting this problem,
>>> as I do not get the same problem on the test-server:
>>> https://graphhopper.com/maps/?point=10%20Lilla%20Nygatan%2C%2011129%2C%20Stockholm%2C%20Sweden&point=10A%20Arsenalsgatan%2C%20103%2027%2C%20Stockholm%2C%20Sweden&layer=Lyrk
>>> <https://graphhopper.com/maps/?point=10%20Lilla%20Nygatan,%2011129,%20Stockholm,%20Sweden&point=10A%20Arsenalsgatan,%20103%2027,%20Stockholm,%20Sweden&layer=Lyrk>
>>>
>>> In essence my setup looks like this:
>>>
>>> Init:
>>>
>>>         hopper = new CustomGraphHopper(settings);
>>>         hopper.setPreciseIndexResolution(1000);
>>>         hopper.forDesktop();
>>>         hopper.setInMemory(true);
>>>         hopper.setOSMFile(osmPath);
>>>         hopper.setGraphHopperLocation("GraphData");
>>>         hopper.setEncodingManager(new EncodingManager("car"));
>>>         hopper.importOrLoad();
>>>
>>> Req: 
>>>
>>>         GHRequest req = new
>>> GHRequest(graph.getNode(from).address.getLatitude(),
>>> graph.getNode(from).address.getLongitude(),
>>> graph.getNode(to).address.getLatitude(),
>>> graph.getNode(to).address.getLongitude())
>>>             .setWeighting("custom")
>>>             .setLocale("SE").setLocale("sv_SE")
>>>             .setVehicle("car").setAlgorithm("dijkstrabi");
>>>         GHResponse rsp = hopper.route(req);
>>>
>>>
>>> Thanks in advance, 
>>>
>>> David
>>>
>>>
>>> _______________________________________________
>>> GraphHopper mailing list
>>> GraphHopper at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/graphhopper
>>
>> _______________________________________________
>> GraphHopper mailing list
>> GraphHopper at openstreetmap.org <mailto:GraphHopper at openstreetmap.org>
>> https://lists.openstreetmap.org/listinfo/graphhopper
>
>
>
> _______________________________________________
> GraphHopper mailing list
> GraphHopper at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/graphhopper

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/graphhopper/attachments/20141217/9a323f84/attachment.html>


More information about the GraphHopper mailing list