On Wed, Jun 17, 2015 at 09:00:37AM +0000, Janko Mihelić wrote: > If you ask me, they are all in their infancy. Non of these routing services > even route right. In a turn restriction the "via" role can be a way. > Neither OSRM, ORS or GraphHopper knows how to restrict that, and that's > IMHO one of the crucial parts of a routing engine. > > When one of them starts routing right, than we can talk about picking a > winner service. Right now only MapQuest knows how to route.
There is no such thing as a perfect routing engine. Every engine can be brought to its knees with some cluefully choosen endpoints. I am thankfull that we have the multitude of routing engines. At work i am using OSRM for calculating distance for telecoms infrastructure for which i am tuning the OSM dataset and calculate like 500K routes in a couple of minutes. On the mobile i am switching between OSMand and Mapfactor Navigator, but all of them refuse to route into a track even if its grade2 and your destination is at that track -> FAIL. Routing home from work with mapquest gives me a route which is not that bad but can be beaten by 20% less travel time from a local. Thats just because everybody in this town knows that certain traffic lights are synced in a tricky manner. You cant put this into OSM. When looking at the bike route the route of mapquest is 1.5miles longer (4.6 -> 6.2miles) than the Car route although there are cycle tracks in the map along ALL roads taken. There is even a shorter route using tracks in the wood but mapquest does not use them for bikes it seems. So a quick test with carefully chosen endpoints produces less then optimal solutions for ANY routing engine. Flo -- Florian Lohoff [email protected] We need to self-defense - GnuPG/PGP enable your email today!
signature.asc
Description: Digital signature
_______________________________________________ talk mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk
