using Michelin's road classification (was: Routing in Liège)

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

using Michelin's road classification (was: Routing in Liège)

André Pirard-2
On 2016-09-13 18:21, Marc Gemis wrote:
Hallo,

I was contacted by a mapper from Germany with whom I worked on turn:lanes.
He has to following question, can someone with local knowledge inform
us about the road classifications ? I have the impression a lot of
streets are indeed residential. Feel free to reply in French, I'll
translate it to English for him.


[snipped]
Now I want ask you about another problem.
Coming from here
http://forum.mapfactor.com/discussion/comment/13515#Comment_13515
I checked Liege to find out the mapping of roads there:
http://overpass-turbo.eu/s/imn
My guess is, that unclassified is used wrong there and that is the
reason for strange routings. My opinion is that unclassified as the
lowest kind of connecting roads do not end at city borders and have or
need common connection to same or higher class inside of towns or
villages. For me routers should avoid residentials and lower as much
as possible. Do you have any idea to check and correct this in Liege
to make routing better?

If there are any questions, please ask.

Regards
Michael aka hurdygurdyman
In order to produce good routing using the main roads (1), ...
why not adopt in OSM the same classification as Michelin (2)?
They should know something about routing, shouldn't they?

Michelin



BE ID/type
Nxx
Nxxx
other
many houses
rare houses
OSM
primary
secondary
tertiary
residential
unclassified

Inside town, the main advantage is that streets are then classified as either secondary/tertiary or residential/unclassified
according to whether they should be used or not for routes from town place to place.
Please note that when a street is promoted to tertiary status, the fact that it contains houses gets disregarded
(and hence I wonder if it's a good idea to consider houses  for road classification rather than using an additional residential attribute (yes, I know it's the way "we" do it)).

I am willing to explain JOSM users how to compare OSM and Michelin easily.
Please send me a private reply to the next message "Michelin info" to get it.
I would do a part of Liège myself if it's organized by someone distributing the tasks.

Cheers

André.

(1) several de-contributors claim loudly that OSM routing is an every contributors' hoax; I tend to agree with them for routing finer than main roads given the complexity of making a no-turn relation, the general misunderstanding of access restriction rules ("bicycle=yes" alone), etc.; but I hate the "we don't do it like that" answers without any constructive remark towards my goal when I suggest improvements.  The "dedicated" subject alone makes whole chapters just because the "dedicated" concept is not an (already existing) access restriction but a reason for one ("but "we" do it like that").

(2) I suppose that Michelin doesn't mind just that since they updated OSM themselves in their 2012 experiment.



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: using Michelin's road classification (was: Routing in Liège)

dieterdreist

2016-09-15 17:02 GMT+02:00 André Pirard <[hidden email]>:
Please note that when a street is promoted to tertiary status, the fact that it contains houses gets disregarded
(and hence I wonder if it's a good idea to consider houses  for road classification rather than using an additional residential attribute (yes, I know it's the way "we" do it)).


when you write "houses", is this about "buildings" or a subset of residential buildings (aka "houses")?
Generally, I think we are doing it more or less like Michelin in your table. Where do you see differences? Clearly the classification of a road should not change just because it enters a settlement.

Cheers,
Martin

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

using Michelin's road classification (was: Routing in Liège)

André Pirard-2
In reply to this post by André Pirard-2
Hi Marc,

Could you please forward/translate this reply to Michael and anyone who believe that Liège (and other places?) contain main road classification errors that produce bad routing.  One-way streets and the like will be another subject.
I suggested below to compare the OSM classification with that of Michelin.
Martin Koppenhoefer replied down below that Michael is wrong because "we" (they) "do more or less like Michelin". But he doesn't say if it is more or less and how much.
He asks "Where do you see differences?", meaning that he did not try to find them.

Consequently, I have made my own Michelin compatible overpass query (thanks to the author of the original) that is using the same colors as Michelin (1).

At very first (short) sight, I am surprised that Martin did not spot that the ref=N3 road that's going north-west is primary (in red) mostly but is interrupted by secondary segments (in yellow) Rue de Hesbaye and Rue Eugène Houdret.  As well as in Rue Louis Jamme to connect Place Delcour to primary N90 and primary N610 (2). N610 should in theory be secondary but I completely agree with Michelin to make it debatable and make it primary like the Namur road should.

I did not investigate further (I'm short sighted indeed) but I suggest that anyone contesting an OSM route compared it with the same routing by Michelin, tried to find an explanation by comparing my overpass with the Michelin map (my "Michelin info" message helps the wise JOSM users too), and asked people with local knowledge if they know better than Michelin.

Last point is what source:???=Michelin ??? to use to prevent a StijnRR or like arbitrarily destructing well thought out tagging without notifying the author. I suggest source:highway=https://viamichelin.be/web/Cartes-plans 2016 2016.

Kéne afêre à Lîdje and I hope that this work will be useful elsewhere too,

Cheers

André.

(1) As white would not fit for residential, I used grey.  I'm surprised indeed that no unclassified roads turn up in blue but that doesn't affect routing.
(2) why the heck do those people adore splitting?

On 2016-09-15 17:02, André Pirard wrote:
On 2016-09-13 18:21, Marc Gemis wrote:
Hallo,

I was contacted by a mapper from Germany with whom I worked on turn:lanes.
He has to following question, can someone with local knowledge inform
us about the road classifications ? I have the impression a lot of
streets are indeed residential. Feel free to reply in French, I'll
translate it to English for him.


[snipped]
Now I want ask you about another problem.
Coming from here
http://forum.mapfactor.com/discussion/comment/13515#Comment_13515
I checked Liege to find out the mapping of roads there:
http://overpass-turbo.eu/s/imn
My guess is, that unclassified is used wrong there and that is the
reason for strange routings. My opinion is that unclassified as the
lowest kind of connecting roads do not end at city borders and have or
need common connection to same or higher class inside of towns or
villages. For me routers should avoid residentials and lower as much
as possible. Do you have any idea to check and correct this in Liege
to make routing better?

If there are any questions, please ask.

Regards
Michael aka hurdygurdyman
In order to produce good routing using the main roads (1), ...
why not adopt in OSM the same classification as Michelin (2)?
They should know something about routing, shouldn't they?

Michelin



BE ID/type
Nxx
Nxxx
other
many houses
rare houses
OSM
primary
secondary
tertiary
residential
unclassified

Inside town, the main advantage is that streets are then classified as either secondary/tertiary or residential/unclassified
according to whether they should be used or not for routes from town place to place.
Please note that when a street is promoted to tertiary status, the fact that it contains houses gets disregarded
(and hence I wonder if it's a good idea to consider houses  for road classification rather than using an additional residential attribute (yes, I know it's the way "we" do it)).

I am willing to explain JOSM users how to compare OSM and Michelin easily.
Please send me a private reply to the next message "Michelin info" to get it.
I would do a part of Liège myself if it's organized by someone distributing the tasks.

Cheers

André.


On 2016-09-15 17:26, Martin Koppenhoefer wrote:
Generally, I think we are doing it more or less like Michelin in your table. Where do you see differences? Clearly the classification of a road should not change just because it enters a settlement.

Cheers,
Martin





_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: using Michelin's road classification (was: Routing in Liège)

dieterdreist


sent from a phone

Il giorno 16 set 2016, alle ore 21:51, André Pirard <[hidden email]> ha scritto:

At very first (short) sight, I am surprised that Martin did not spot that the ref=N3 road that's going north-west is primary (in red) mostly but is interrupted by secondary segments (in yellow) Rue de Hesbaye and Rue Eugène Houdret.


I was referring to osm in general and not to your particular situation which might have problems or not. Generally, a primary should be connected with other primaries (or higher road classes) and constitute a consistent network, but there might also be (few) exceptions to this general rule.


cheers,
Martin 

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: [OSM-talk-be] using Michelin's road classification (was: Routing in Liège)

joost
In reply to this post by André Pirard-2
I did not investigate further (I'm short sighted indeed) but I suggest that anyone contesting an OSM route compared it with the same routing by Michelin, tried to find an explanation by comparing my overpass with the Michelin map (my "Michelin info" message helps the wise JOSM users too), and asked people with local knowledge if they know better than Michelin.

Last point is what source:???=Michelin ??? to use to prevent a StijnRR or like arbitrarily destructing well thought out tagging without notifying the author. I suggest source:highway=https://viamichelin.be/web/Cartes-plans 2016 2016.


You can't just use Michelin as a source. It is copyrighted data. The site explicitly states you are only allowed to use their content for personal use. 
 So source:classification=michelin is probably a good idea, makes it easier to revert.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: using Michelin's road classification (was: Routing in Liège)

Richard Fairhurst
In reply to this post by André Pirard-2
André Pirard wrote:
> Last point is what source:???=Michelin ??? to use to prevent a
> StijnRR or like arbitrarily destructing well thought out tagging
> without notifying the author. I suggest
> source:highway=https://viamichelin.be/web/Cartes-plans 2016 2016.

No, you must not copy from copyrighted maps, which includes Michelin's.

Please confirm that you have not added, and are not going to add, any data (including classification judgements) from Michelin maps, otherwise I guess we'll have to ask the Data Working Group to suspend your OSM account and revert your edits.

Richard
Reply | Threaded
Open this post in threaded view
|

Re: Routing in Liège (consulting Michelin)

André Pirard-2
On 2016-09-17 20:46, Richard Fairhurst wrote:
André Pirard wrote:
Last point is what source:???=Michelin ??? to use to prevent a 
StijnRR or like arbitrarily destructing well thought out tagging 
without notifying the author. I suggest
source:highway=https://viamichelin.be/web/Cartes-plans 2016 2016.
No, you must not copy from copyrighted maps, which includes Michelin's.

Please confirm that you have not added, and are not going to add, any data
(including classification judgements) from Michelin maps, otherwise I guess
we'll have to ask the Data Working Group to suspend your OSM account and
revert your edits.

Richard
Of course not, I did not use that Michelin map as a source and I won't.
It would be stupid because it's coarse compared to other allowed sources.
And everything of it and much more is in OSM already.
Even a road classification does not interestingly exist in it (1).
This Subject: and suggested source= are inappropriate and misleading.

The only thing I did is to notice that Michelin runs the N3 in red from Ans to La Dérivation, compare it to my turbopass map, discover that OSM has a gap in this primary and post the URLs of a few streets containing ref=N3 and highway=secondary.  You have read that, haven't you, I posted their URLs in this thread.
Now if I hope that you won't say that this is copying from Michelin and that you would send to prison someone who would correct those mistakes and that you would put the mistakes back in OSM.

I have been extremely shocked by what you said after my attempt to help.

Beside mapping the boundaries of South Belgium and other major works, I have done excellent mapping, I very often correct a huge number of houses and roads, almost all, misplaced by 3-5-+ meters, I help mappers, I even help JOSM and others to improve their software for a better OSM.
Revert my edits? 
Remove the Walloon borders?  Remove the other many things? Remove my humanitarian tagging?  Put houses and roads back to the wrong place?  etc...
Very rarely a word of thanks, except Marc in this thread, close friends, some developers and the humanitarians.
Always reproaches and destruction.
Even the DWG vandals messed up the marvelous job of the boundaries of Wallonia.

That total lack of consideration for what I'm doing made me decide to stop contributing to OSM.

André.

(1) On that map, Michelin unhelpfully draws the equivalents of both OSM's secondary and tertiary in yellow.
Tertiary can't be distinguished from secondary.
The only thing that can be done with it is to build a route to compare with an OSM and so to have hints to modify the  OSM route.
This said, I tried on Michelin the route Beaufays-Oupeye that I know well.
Definitely, their route #2 27 km via E25 is much better than their recommended #1 26 km via E40.
And within #2, Bd de Douai -> Poincaré is better than quai des Ardennes, it is even the signposted route!
So, Michelin is not even always a good adviser!


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging