[Indoor] Indoor Routing Observations
Volker Krause
vkrause at kde.org
Thu Mar 7 16:52:10 UTC 2024
Thanks Antoine, I suspected you had faced all of this already :)
For the implicit walls I agree with your conclusion, and we also came to the
same result in the discussion in yesterday's meeting:
- assume indoor=corridor has no implicit walls, ie. routing-wise it's the same
as indoor=area.
- fix TU Dresden (and other affected buildings using indoor=corridor for fully
walled corridors with mapped doors) by using indoor=room + room=corridor
instead.
- add explicit outer building walls where needed (tagged as indoor=wall).
Regarding counters/reception desks I didn't know about the reception desk
proposals yet. It would seem like something useful to have when mapping e.g.
hotels or event venues. It's a bit orthogonal to tagging counters though I
think, one describes physical structure the other one semantics (which might
apply to a room, a counter, a desk, an area, etc). So I guess we want both?
Example 1:
a way tagged as indoor=room + amenity=fast_food with one edge tagged as
indoor=counter
Example 2:
a way or node tagged as amenity=reception + indoor=counter within an
indoor=area hotel lobby
Technically this isn't indoor-specific, but I'm not sure if there's a better
tag than indoor for this?
Regards,
Volker
On Dienstag, 5. März 2024 15:36:51 CET Antoine Riche via indoor wrote:
> Hi Volker.
>
> Thanks for raising these interesting questions. We actually faced these
> issues when mapping Paris railway stations. Here are my two pennies.
>
> > Implicit corridor walls
>
> Actually https://wiki.openstreetmap.org/wiki/Tag:indoor%3Dcorridor
> defines indoor=corridor as a "passageway space without walls". This
> seems a little bit contradictory with "Enclosed walkway area" as
> described in https://wiki.openstreetmap.org/wiki/Simple_Indoor_Tagging,
> however I'm not so sure that "enclosed" should be understood as closed
> by walls on all sides here.
>
> Considering that indoor=corridor have implicit walls would imply to use
> indoor=area for areas that really feel like corridors though they do not
> have a door at ends. Mapping corridor walls explicitly offers more
> flexibility, and they are not required anywhere the corridor is adjacent
> to an idoor=room.
>
> > (Implicit) Outer building walls
>
> Again not considering walls are implicit offer more flexibility. I can
> think of a railway station building
> (https://www.openstreetmap.org/way/30174513) adjacent to a corridor
> (https://www.openstreetmap.org/way/643954662) that runs below the rails
> and platforms. There is no door between the building and the corridor,
> considering implicit walls on the building outline would not work here.
>
> > Service counters
>
> The tag amenity=reception_desk is in use.
> https://wiki.openstreetmap.org/wiki/Tag%3Aamenity%3Dreception_desk
> explains that the proposal was rejected several times, mostly due to the
> name of the tag. Counter sounds like a better name. Also the definition
> as "the location of a receptionist
> <https://en.wikipedia.org/wiki/receptionist> who is employed by an
> organization to receive or greet any visitors, patients, or clients" is
> perhaps a bit restrictive, this could be widened to any location where a
> person can make a request to a professionnal.
>
> Cheers,
> Antoine.
>
> Le 26/02/2024 à 18:01, Volker Krause via indoor a écrit :
> > Hello,
> >
> > I've been working on indoor routing recently and hit a few challenges. We
> > discussed those at the Hack Weekend in Karlsruhe as well, but there isn't
> > really an elegant solution for any of the following yet:
> >
> >
> > # Implicit corridor walls
> >
> > indoor=corridor and indoor=room are defined to have implicit walls, unlike
> > e.g. indoor=area (seehttps://wiki.openstreetmap.org/wiki/
> > Simple_Indoor_Tagging#Modelling_the_different_indoor_elements).
> >
> > However the predominant use of indoor=corridor isn't really in line with
> > that.
> >
> > Example 1:https://www.openstreetmap.org/relation/3845325
> >
> > This is entirely open to the north end leading into the main hall, closed
> > on the south (with multiple doors) and on the sides it's either walls,
> > adjacent shops (also see service counters below) or ways up to the
> > platforms.
> >
> > By the strict SIT interpretation neither the main hall nor the platforms
> > nor the shop service counters would be accessible.
> >
> > Example 2:https://www.openstreetmap.org/way/453091694
> >
> > Open on short-edge ends, intuitively a "corridor", but by strict SIT
> > interpretation all connections to staircases or adjacent halls would be
> > closed.
> >
> > Example 3:https://www.openstreetmap.org/way/870541411
> >
> > This strictly follows SIT. All edges of the corridor either have doors or
> > are actually walled (adjacent to rooms or the outer building wall).
> >
> >
> > For rendering this is mostly subtle, however for routing this is very
> > problematic. With strict SIT interpretation many train stations are
> > practically inaccessible. And when not using implicit walls you end up
> > with
> > routes through the building walls in example 3 (due to the next point).
> >
> >
> > # (Implicit) Outer building walls
> >
> > indoor=area has no implicit walls but can occur with one or more edges
> > aligned to an outer building wall.
> >
> > Example 1:https://www.openstreetmap.org/relation/6038767
> >
> > There are walls where this area aligns with the building outline on the
> > north and south sides, ie. routing through those edges should not be
> > possible (apart from the doors).
> >
> > Example 2:https://www.openstreetmap.org/relation/11878686
> >
> > Same situation here. However the building outline (https://
> > www.openstreetmap.org/way/23290301) does not match the actual outer
> > building walls on the ground floor here, with the parking spaces on
> > https:// www.openstreetmap.org/way/27743604X being on the outside and
> > the building only extending to the full outline on higher levels.
> >
> > Example 3:https://www.openstreetmap.org/relation/10142437
> >
> > Here the area is open towards the outside. Assuming there's an implicit
> > wall on the building outline would make this inaccessible for routing.
> >
> >
> > For rendering this again is relatively subtle, but for routing that's a
> > crucial difference.
> >
> >
> > # Service counters
> >
> > Less severe, but also occurring in many train stations is the fact that
> > amenities are often tagged as indoor=room (and thus implicitly walled) but
> > without any tagged doors, making it impossible to route into them. In many
> > cases the fix here is adding the missing doors. However there are also
> > amenities (bakeries, fast food stands, etc) that serve over a counter on
> > an
> > edge of the room polygon and don't actually allow you to enter. That
> > counter is where routing should lead you to ideally, but how can that be
> > tagged?
> >
> > Example:https://www.openstreetmap.org/way/402790810
> >
> > Free-standing room, service counter at the north-facing edge.
> >
> >
> > Any thoughts on how to deal with those issues?
> >
> > Thanks,
> > Volker
> >
> > PS: Please remember the quarterly OSM Indoor online meetup on March 6th
> > 18:00 CET athttps://osmvideo.cloud68.co/user/tob-2uf-drl-eal!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 195 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.openstreetmap.org/pipermail/indoor/attachments/20240307/96823f83/attachment.sig>
More information about the indoor
mailing list