Closed LafarM closed 11 years ago
There is not much to correct on our end. The alternative that is suggested already gives you the route you want.
Very funny. I notify this situation because this is problem. Such mistake in the routing causing that users do not want to use this service. Please think about it again.
First thing in testing new nawigation for is checking nawigation on well known places. This is one of them. You want to have routing service or good routing service?
It is a data issue, not a routing issue. Nothing on that tertiary road indicates that it may be a bad choice.
one thing that i think improved routes suggestions for bicycle a lot was to use a turn penalty function, so that routes with less turns are preferred. the car profiles doesn't use turn penalties at the moment. adding one would be easy, maybe worth considering?
I think that you should think again about this problem. I know that route calculated by service is shortest about 900m, calculated time is the same but please believe me that every time I use mu route. My route is faster and safer and probably cheaper. Maybe you should add specjal rating for roads and streets, for example: as default value all road should have the same value and after notification as this my you should change this value. Next time during route calculation this route should has worse score and other correct route should be choose.
P.S. I have other ideas to improve the routing in Poland by this service. Where I can describe my ideas? This is not issue.
@emiltin Turn penalties do not solve the problem at hand.
@LafarM Have you looked at the data? There is nothing in the data that supports your claims. The problem should be fixed in the data. If there was anything in the data that we can use, then we can respect that in the speed profile.
Tag maxspeed is added on all roads used on both routes. In my route are used roads with a higher priority. Roads used in route calculated by service have lower prioryty. On these roads I have to go slowly because there are tractors and I have no possibility to overtake because they are too short lengths of straight road between the curves. The situation is even worse in the winter, as indicated by the service roads are often covered by snow.
@DennisOSRM LafarM said the route osrm picks (A) has more turns, which is a reasons it's slower to follow. so turn penalties could be relevant? A has 5 sharp turns and slight, while B has only 1 sharp, and 2 slight turns.
as far is i recall, turns between intersections are also taken into account. this stretch of A seems to have many bends: http://osrm.at/57Z
@emiltin you have right.
@DennisOSRM Maybe service should add some time for crossroads, sharp turns, etc. during route calculation.
My other idea for correction route calculation in Poland. From October 30th to November 2nd extra time should added to roads and streets next to the cemetery, If You wish I can write more detailed my ideas.
The routing service -as of now- will not incorporate any special rules that cannot be infered from the OSM map data. Otherwise, you will end up with a huge description file that contains a plethora of exceptions and is impossible to manage.
A better approach might be thinking about how such additional information can be introduced to the OSM data in a generic way (i.e. applicable for many different cases) , so that it can in turn be processed by OSRM with only little changes to the LUA.
Also, your concrete idea should definitely be handled differently. You don't want to add exception for all roads that can become congested due to regular events. If you go down that road, you would have to add extra time on arterial roads during rush hours, around schools in the morning and after they end, around stadions at the weekends, and so on. For this problem you want a different kind of routing engine that can process previous traffic situations and make predictions for the future. But this is totally different story (not even taking into account how to aggregate this data).
@DennisSchiefer I use OsmAnd. I thought that OSRM create service which will be able to set routes with high precision. However, Now it seems to me that the service would give me a little more than offline calculate routes through OsmAnd. Maybe I'm wrong.
Sorry for my unnecessary comments.
Hello,
I think that service should make smal corection in routing for below route. This route is correct only for local drivers which know this road very well. They use this route only in good weather conditions because this road is narrow and has many bends. Please correct this in service.
Route calculated by service http://map.project-osrm.org/?hl=pl&loc=50.363590,22.671665&loc=50.436863,22.745102&z=13¢er=50.393582,22.718697&alt=0&df=0&re=0&ly=588398340
Correct route with my modification http://map.project-osrm.org/?hl=pl&loc=50.363590,22.671665&loc=50.381475,22.741971&loc=50.436863,22.745102&z=13¢er=50.393582,22.718353&alt=0&df=0&re=0&ly=588398340
Best regards, Rafał Michalski