[OSM-talk] Routing over areas

Aun Johnsen lists at gimnechiske.org
Wed Apr 22 12:31:59 UTC 2015


Routing through area can be solved in 1 of 3 ways, depending on how the routing engine is built and how it handles rules.

1) route along the edge, treating the area as a way. This is probably the easiest solution
2) cross the area in straight line from entry point to exit point, be aware that you might be routed through fountains or other obstackles
3) the most advanced way is to make the routing engine aware of multi poligon shapes and try to route you the shortest way around any obstackles. This solution will cost the routing engine a lot of memory and processor time

As a work around you can create paths cutting the area to connect the common entry and exit points, this eay you get a reliable sane route independent of routing implementation

Aun Johnsen
Sent from my iPhone

> On 22. apr. 2015, at 09.00, talk-request at openstreetmap.org wrote:
> 
> Subject: Re: [OSM-talk] Routing over areas
> Message-ID:
>    <CAJvDm4Pvb5Rgt1nMJ+FV3e0Ti_Vb8my4Rr7t3i=a9aqNbS13Mw at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
> 
>> On 21 April 2015 at 15:52, Svavar Kjarrval <svavar at kjarrval.is> wrote:
>> 
>> I wonder if similar adjustments could be made for streets with a low
>> speed limit, especially where there are no pedestrian crossings nearby
>> (be it imperfect data or there aren't any). Seems crazy to suggest that
>> pedestrians and cyclers get a routing suggestion to go around to the end
>> of the road and back to the other side instead of crossing the road *in
>> the cases where a sane person would do exactly that*. Of course, it
>> might need some extra tags to prevent suggestions like that where such
>> behaviour is prohibited.
>> 
>> - Svavar Kjarrval
>> 
>>> On 20/04/15 21:25, Rob Nickerson wrote:
>>> Interesting article for any routing experts:
>>> 
>>> http://anitagraser.com/2015/04/17/routing-in-polygon-layers-yes-we-can/
>>> 
>>> Rob
> If there's a direct route across the square, it should be on the route map
> as a route - then it'd be correctly included in routing software solutions.
> 
> Mike.



More information about the talk mailing list