Routing Problem - roundabout with bypass

[franco...@gmail.com]

In Landsberg, Germany, there is a giant roundabout that connects the Motorway A96 with the Trunk road B17.
http://www.openstreetmap.org/#map=17/48.04811/10.83800
When you travel coming from the north (Augsburg) to the west (Lindau) there is a bypass lane that ships you around the roundabout.
Osmand does not take this lane, it tells me to "keep left" then "enter roundabout" and "leave at 1st exit".
This works, but it's neither the fastest, nor the shortest way.

Using an OSM based map on my Garmin device for reference, I get routed through the bypass lane.
So I think the problem does not lie in the OSM data, but in the routing logic of osmand.

[Harry van der Wolf]

It is indeed strange. The inner ring is a primary road, the bypass lanes are motorway links.

Primary roads have average speed of 65 and priority 1.05.

motorway links have average speed of 110 and priority of 1.2

OsmAnd should use the bypass unless you use the routing option "Shortest way" or set the option "Avoid motorways".

Please check that.

Harry

2013/11/26 <franco...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "Osmand" group.
To unsubscribe from this group and stop receiving emails from it, send an email to osmand+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[franco...@gmail.com]

Hi Harry,

thanks for the fast reply.
I didn't change any of the default settings, but I just took  a look:
Navigation is set to "fastest route",
there is no avoidance set - everything is allowed.

BTW, If I had "Avoid motorways" ON, Osmand shouldn't have taken me on the A96 at all.

Yes, it really should take the bypass but it doesn't.

Ciao,
  Franco

[Stephan75]

What is the result with other OSM based routing programs? see https://code.google.com/p/osmand/wiki/FAQ#Routing_in_Osmand_from_start_to_destination_point_is_wrong

What is routing result in Mapfactor Navigator free or be-on-road?

[Harry van der Wolf]

2013/11/27 Stephan75 <der.steph...@googlemail.com>

What is the result with other OSM based routing programs? see https://code.google.com/p/osmand/wiki/FAQ#Routing_in_Osmand_from_start_to_destination_point_is_wrong

What is routing result in Mapfactor Navigator free or be-on-road?

Be-on-road does exactly the same as OsmAnd. It routes you over the roundabout and not over the bypass lane.

Harry

[Nelson A. de Oliveira]

OSRM doesn't use the bypass too: http://osrm.at/5JS

Maybe there is something wrong on the OSM data? (unconnected ways maybe)

[Nelson A. de Oliveira]

Nor GraphHopper uses the bypass:
http://graphhopper.com/maps/?point=48.050722%2C10.836055&point=48.045913%2C10.829971&locale=en-US

[Harry van der Wolf]

The problem is the max speed and the way the roundabout is created. The bypass has a maxspeed of 60 and the to/from links also have a maxspeed of 60.

The rounadabout is actually tagged as "meadow" having relations being tagged as routes, all primary. This means that a default primary route uses the average speed of 65 kmph. This means that the roundabout is faster (theoretically).

That is the reason all navigation apps and webservices use the "faster" roundabout instead of the bypass lane.

Someone should change the (multi)polygon to a highway and make the inner part meadow and not the other way around.

Harry

2013/11/27 Nelson A. de Oliveira <nao...@gmail.com>

Nor GraphHopper uses the bypass:
http://graphhopper.com/maps/?point=48.050722%2C10.836055&point=48.045913%2C10.829971&locale=en-US

[Franco Bez]

Actually there is a traffic sign for a speed limit of 60km/h on the roudabout itself.
Unluckily the OSM Database is in read only mode due to service operations.
I will check the tags and set a speed limit to the roundabout as soon as possible.

Thanks,
  Franco

[Nelson A. de Oliveira]

On Wed, Nov 27, 2013 at 5:01 PM, Harry van der Wolf <hvd...@gmail.com> wrote:
> The rounadabout is actually tagged as "meadow" having relations being tagged
> as routes, all primary. This means that a default primary route uses the
> average speed of 65 kmph. This means that the roundabout is faster
> (theoretically).

There are indeed two overlapping ways (the meadow and the highway),
but it's tagged with maxspeed=60 too (so they (bypass and roundabout)
all have the same maxspeed).
I don't know what is the tagging convention in Germany, but for me
(and for JOSM's validator), this primary roundabout should be changed
to motorway_link.

[Nelson A. de Oliveira]

On Wed, Nov 27, 2013 at 5:12 PM, Franco Bez <franco...@gmail.com> wrote:
> Actually there is a traffic sign for a speed limit of 60km/h on the
> roudabout itself.
> Unluckily the OSM Database is in read only mode due to service operations.
> I will check the tags and set a speed limit to the roundabout as soon as
> possible.

It's already tagged with maxspeed=60

[Rodolfo]

Both options have maxspeed=60, but the first exit of the roundabout (shortest distance from the split!) has maxspeed=none, so it may be preferred for calculating the fastest route. I'm not sure if the motorroad=yes (for the roundabout) is used by the routing services.

[Franco Bez]

Wouldn't it be a good idea to have a default "maxspeed" in the program code for *-link roads of maybe 60km/h ?
So in cases like this, where there is no traffic sign that actually limits the speed on the link (and this is why maxspeed is not set in OSM) you have a reasonable default value.
Wouldn't this avoid problems like not taking the bypass ?
Also it should have no measurable affect on the calculated "time to arrival" as link roads are usually short.

[Harry van der Wolf]

2013/11/28 Franco Bez <franco...@gmail.com>

Wouldn't it be a good idea to have a default "maxspeed" in the program code for *-link roads of maybe 60km/h ?

No, it would not in my opinion. Motorway links connecting one motorway to another are mostly faster. Next to that: if there is no maxspeed set in OSM it doesn't mean that it is 60 km/h on that road.

The example that occurred now is simple one that I have never encountered as it mostly works OK. A link way should have a slightly lower value than the main way, but not 60 km/h.

Harry

[Franco Bez]

Hi Harry,

Am Donnerstag, 28. November 2013 23:18:17 UTC+1 schrieb Harry van der Wolf:

2013/11/28 Franco Bez <franco...@gmail.com>

Wouldn't it be a good idea to have a default "maxspeed" in the program code for *-link roads of maybe 60km/h ?

No, it would not in my opinion. Motorway links connecting one motorway to another are mostly faster.

 But mostly the links are quite curvy. Some crazy guys might try them with 100 but most people will go 60 to 80.

Next to that: if there is no maxspeed set in OSM it doesn't mean that it is 60 km/h on that road.

That's true, but for other roads a default is used for calculation also.

I do not know about motorways, here in Germany they are unlimited, but for sure there will be a default speed used for calculation, and for sure this won't be 290 or more ?

The example that occurred now is simple one that I have never encountered as it mostly works OK. A link way should have a slightly lower value than the main way, but not 60 km/h.

I think the value should be a lot slower than on the main road.
On a motorway for instance I might go 140, if I took the exit just slightly slower it would not be a good idea.
 

Harry

Anyway, we seem to have different opinions here, but this is not a problem for me.

Maybe there is a better solution.

Anything that works is fine for me ;-)

As long as my example is the only one worldwide I consider it a minor Problem.

Ciao,
  Franco

[Rodolfo]

>> "Wouldn't it be a good idea to have a default "maxspeed" in the program code for *-link roads of maybe 60km/h ?"

I don't think so, but even if there was a default maxspeed for *-links, this default would not be used in this case, because all ways here have a maxspeed  tag.

[Harry van der Wolf]

Hi Franco,

2013/11/28 Franco Bez <franco...@gmail.com>

Hi Harry,

I do not know about motorways, here in Germany they are unlimited, but for sure there will be a default speed used for calculation, and for sure this won't be 290 or more ?

No. I think it will be approximately 250 :)

Please take a look at the routing.xml (https://github.com/osmandapp/OsmAnd-resources/blob/master/routing/routing.xml). It contains the default speeds that OsmAnd uses for the several road types or options you set (shortes/fastest, avoid ..).

 

The example that occurred now is simple one that I have never encountered as it mostly works OK. A link way should have a slightly lower value than the main way, but not 60 km/h.

I think the value should be a lot slower than on the main road.
On a motorway for instance I might go 140, if I took the exit just slightly slower it would not be a good idea.

No of course not. But when driving safely you adapt your speed to the circumstances. If the road sign says you are allowed to drive 120 but if the road itself, traffic or weather conditions don't allow it, I hope you don't continue driving at 120.

In Nederland we have motorway exits and ramps with different speeds depending on the curvature of the exit or ramp.

You can have exits/ramps that still allow to a large extend to drive 120, some allow you to drive 90, some 80 or 70, and some indeed even 60. These speeds are, under normal conditions, safe speeds. And I assume it is not different in other countries. So why set some default too low speed for a *-link

Harry

[Franco Bez]

Hi Harry,
thanks for the link to the xml-file.
It's the answer to some questions I was thinking about.

--------------------------------
I see that in reality we two mostly argee about what speeds are sensible.

--------------------------------
Back to the actual issue.

If a slower default speed is not a slolution, are there any other ideas on how to fix the bypass problem?
Or will I simply have to live with it ?

Ciao,
  Franco

[sympa]

If a slower default speed is not a slolution, are there any other ideas on how to fix the bypass problem?
Or will I simply have to live with it ?

Even if speeds are the same and a bypass is only slightly shorter, it is often advisable to stay on the same road instead of taking a minor shortcut. There are different reasons to do this:

- merging carries some risk and effort, which sould be traded against a 2 second time difference
- the shortcut road may be in a traffic jam
- the shortcut may be wrongly mapped, for example a service road
- the road type may be  or misinterpreted if no speed limit is set
- the shortcut may be of very poor quality

As of the last case, I was zigzaging down a secondary road in Tenerife to some town. Between the legs of the zig-zag were residential roads with very steep incline. But Osmand wanted us to take these because they were twice as short.

For a solution? There should be some penalty on merging or changing road name and/or type.
And the map should carry tags on the roads indicating:

- use only local access if destination is on this street or within X km
- for local and intermediate access, if destination s within Y km
- use always

It might even be implemented on road type: if I am driving from Belgium to Spain, I want to take the ring around Paris even if there is a good shortcut through the center. Thus, less roads should carry some penalty if they are far from source or destination. Maybe this even makes routing of long trips easier?

[Franco Bez]

When I switch on "precise routing" I get routed through the bypass lane, just as it should be.

[Franco Bez]

with the new routing, introduced in Version 1.7 this issue is fixed.

Now I get routed through the bypass lane.

Good work.

[Franco Bez]

Now it's broken again :-(
Something must have changed quite recently.

I tried OsmAnd+ 1.7.4 - here it's still OK

OsmAnd Nightly #1952D from this morning - here it's broken.
Also yesterdays OsmAnd~ Nightly - broken.

Both on the same machine, same map.

I attached some screenshots.

[Franco Bez]

Just tested tonights nightly #1959D - still broken

[Franco Bez]

#1964D - still broken:-(

[Max]

Yes, one or several of these commits are causing this:

https://github.com/osmandapp/Osmand/commit/fb83b64ac456728c3c28cd3e445bdea89d654951
https://github.com/osmandapp/OsmAnd-core/commit/95ccb80c2183c2211e3bdc95aa5f030d96bb8ea8
https://github.com/osmandapp/OsmAnd-resources/commit/0a15af5e3cdfecfbf73c882a6251bdc2d591ba02

But I don't know, if the old behaviour was right or the new behaviour is right.

The two alternative routes have maxspeed=60:
https://www.openstreetmap.org/way/25000228
https://www.openstreetmap.org/way/151334285
https://www.openstreetmap.org/way/151332842

Except of this small section between
https://www.openstreetmap.org/node/1641566441
and
https://www.openstreetmap.org/node/479099922
maybe this makes the difference, because it has maxspeed=none:
https://www.openstreetmap.org/way/39905378

But the change of "CAR_SHORTEST_DEFAULT_SPEED" should only affect "shortest route", not "fastest route"?
So I don't understand...

Regards,
Max

[Rodolfo]

Just to compare how other routing engines handle this. See screenshot.
YOURS and MapQuest use OSM data. (Of course Google and Bing use their own maps), but all 4 take the bypass using fastest route and shortest route.

[Franco Bez]

Hi Rodolfo and Max,

and GARMIN using a OSM-based map also routes through the bypass.
It's both faster and shorter to take the bypass, and also it's the designated way - the Sign hanging over the bypass lane reads "A96 Lindau"".
Routing through the roundabout here is definitely a bug.

Ciao,
  Franco

[Franco Bez]

Hi Max,

I changed the maxspeed of this section to 60 - infact all the exits from the roundabout are limited to 60.

With the next bayern-map we will see whether this helps or not.

Ciao,
 Franco

Am Montag, 14. April 2014 07:01:36 UTC+2 schrieb Max

[Franco Bez]

Now I just dowloaded the new bayern-map.
Setting max-speed=60 on all the exits seems to have done the trick.
Let's hope this bug remains fixed from now on.

Ciao,
 Franco