[OSRM-talk] Huge difference in needed time computation OSRM / Google Maps

[Emil Tin]

The car profile currently uses a smallish penalty for signals. Turn penalty is implemented in OSRM and used in the bicycle profile, but not yet in the car profile. 

For bicycles, the turn penalty (which increases with angle) provides better routes, because they have fewer turns and are only marginally longer.



-----Oprindelig meddelelse-----
Fra: Rudolf Mayer [mailto:rudolf.mayer at gmail.com] 
Sendt: 28. marts 2014 11:11
Til: osrm-talk at openstreetmap.org
Emne: Re: [OSRM-talk] Huge difference in needed time computation OSRM / Google Maps

Hi Emil!

Thanks for pointing this out!

Indeed the limit to 2/3 of the max speed seems a bit too conservative, that is, at least outside of cities; inside city limits you could argue that it is ok, at least as long as there is no turn penalties and traffic light penalties implemented; and you generally can always expect a bus or slow traffic.. :-)

I am still wondering whether also the number of lanes is considered in the example. I think that other routers consider this, and if a road has more than 1 lane per direction, the speed reduction is less severe, or at least that road is preferred over other roads that are single lanes per direction.

greetings
rudi

[Emil Tin], On 2014-03-28 09:33:
>
>
> Hi,
> The following issue might contribute to inflated travel time:
> https://github.com/DennisOSRM/Project-OSRM/pull/969
>
>
> Emil Tin
> City of Copenhagen
>
>
> -----Oprindelig meddelelse-----
> Fra: Rudolf Mayer [mailto:rudolf.mayer at gmail.com]
> Sendt: 24. marts 2014 13:30
> Til: OSRM-talk at openstreetmap.org
> Emne: [OSRM-talk] Huge difference in needed time computation OSRM / 
> Google Maps
>
> Hi everyone!
>
> I tried to compute the following route:
>
> http://osrm.at/?hl=en&loc=48.208845,16.372329&loc=48.276104,16.791565
>
> And get this as a proposed result:
>
> Route Description
> (B3 - L2)
> Distance:	44.4 km
> Duration:	1 h 15 min
>
>
> This seems very wrong, though, Google maps estimates
> (http://goo.gl/maps/EPt14)
>
> a.) a likely more optimal route, preferring bigger streets
>
> b.) estimates 45,1 km, 47 Minutes. Almsot 30 mins difference in a 47 minutes route seems way off...
>
>
> I am not sure what this is caused by, the map data seems ok for this route.
>
>
> Besides this huge time difference, one particullarily interesting 
> detail in the routing is at http://osrm.at/6PV, where OSRM prefers a 
> short cut through a much smaller street (tertiary vs primary) that has 
> much more crossings, and http://osrm.at/6PW would be the much preferred rout..
>
> Any ideas why this is happening?
>
> greetings
> rudi
>
> _______________________________________________
> OSRM-talk mailing list
> OSRM-talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/osrm-talk
>
> _______________________________________________
> OSRM-talk mailing list
> OSRM-talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/osrm-talk
>


_______________________________________________
OSRM-talk mailing list
OSRM-talk at openstreetmap.org
https://lists.openstreetmap.org/listinfo/osrm-talk