Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Fri, May 12, 2017 at 1:38 AM, Marc Gemis  wrote:

> ok, multiple from in a relation will solve this.
> Isn't it a problem that some "from"s do not end in some "intersection"s ?


Perhaps 3 intersection roles, with the single carriageway segment between
the dual carriageways being the third intersection role?
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Marc Gemis]

ok, multiple from in a relation will solve this.
Isn't it a problem that some "from"s do not end in some "intersection"s ?

On Fri, May 12, 2017 at 8:31 AM, Paul Johnson  wrote:
>
>
> On Fri, May 12, 2017 at 1:28 AM, Marc Gemis  wrote:
>>
>> On Fri, May 12, 2017 at 8:20 AM, Paul Johnson  wrote:
>> >
>> >> to: the collection of ways one can travel to after stopping/giving
>> >> way/waiting for traffic signal. This would include the from way so
>> >> u-turns have to obey the sign/signal as well.
>> >
>> >
>> > Yes.  At a minimum, a stop, give_way, traffic_signals or traffic_calming
>> > relation would have 1 from, 1 intersection and 1 to.
>>
>> Does that mean that for the examples on
>> https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtraffic_signals with
>> at least 1 dual carriage way, one will need 4 relations ?
>
>
> Simplest case, traffic signal impacts all movements, dual carriageway
> north-south, single-carriageway east-west, you would have 1 relation with
> four from, four to, and two intersection roles.
>

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Fri, May 12, 2017 at 1:28 AM, Marc Gemis  wrote:

> On Fri, May 12, 2017 at 8:20 AM, Paul Johnson  wrote:
> >
> >> to: the collection of ways one can travel to after stopping/giving
> >> way/waiting for traffic signal. This would include the from way so
> >> u-turns have to obey the sign/signal as well.
> >
> >
> > Yes.  At a minimum, a stop, give_way, traffic_signals or traffic_calming
> > relation would have 1 from, 1 intersection and 1 to.
>
> Does that mean that for the examples on
> https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtraffic_signals with
> at least 1 dual carriage way, one will need 4 relations ?


Simplest case, traffic signal impacts all movements, dual carriageway
north-south, single-carriageway east-west, you would have 1 relation with
four from, four to, and two intersection roles.
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Marc Gemis]

On Fri, May 12, 2017 at 8:20 AM, Paul Johnson  wrote:
>
>> to: the collection of ways one can travel to after stopping/giving
>> way/waiting for traffic signal. This would include the from way so
>> u-turns have to obey the sign/signal as well.
>
>
> Yes.  At a minimum, a stop, give_way, traffic_signals or traffic_calming
> relation would have 1 from, 1 intersection and 1 to.

Does that mean that for the examples on
https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtraffic_signals with
at least 1 dual carriage way, one will need 4 relations ?

m

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Fri, May 12, 2017 at 12:57 AM, Marc Gemis  wrote:

> So would a stop sign / give way sign /traffic signal then be mapped as
>
> stop_position: node where on the street does one have to stop/give
> way/wait for traffic signal
>

My thinking on this is stop_position isn't necessary as, unlike public
transportation, the distinction from the conflict point itself (the
intersection node(s)) isn't significant enough to bother.


> sign : node (optional) the exact location of the sign
>

I'd go with device instead, as this could potentially be a traffic signal.
Or on many intersections where a high traffic expressway crosses a minor
road, it could be multiple devices:  Traffic signals that is perpetually
flashing red for the minor way and flashing yellow for the trunk, in
addition to stop signs facing the minor directions.  This comes up a lot,
for example, on US 26 between Gresham and Government Camp, US 30 between
Portland and Warrenton, and pretty much any 70 MPH surface expressway in
Oklahoma.


> from: the way one is following to which the action has to be applied
> (is this needed ?)
>

Yes, but optional.  There could be multiple from ways.  For example, a four
way intersection for which there is a three-way stop, and one of those
directions may turn right without stopping.  Or a French example, a traffic
signal that has a "bicycles yield when turning right
"
sign.


> intersection: the node of the intersection for which the sign/signal holds
>

 Yes.

to: the collection of ways one can travel to after stopping/giving
> way/waiting for traffic signal. This would include the from way so
> u-turns have to obey the sign/signal as well.
>

Yes.  At a minimum, a stop, give_way, traffic_signals or traffic_calming
relation would have 1 from, 1 intersection and 1 to.


> is this how you see the relation ? Could it be simplified for the most
> common case that the sign/traffic signal applies to all directions one
> can travel ?
>

Yes.  A situation like an onramp meter
, onramp or similar yield sign,
or all-way stop where all possible movements for which the control or
calming applies is a single node, and all movements across that node are
affected by the calming or control, could still be mapped as just a simple
node with the current schema.


> what in case there is a turn restriction at the intersection ? Does
> the to-collection of the stop sign also have to include prohibited
> turns ?
>

No, you would use the existing turn restriction relation schema for turn
restrictions.
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Marc Gemis]

So would a stop sign / give way sign /traffic signal then be mapped as

stop_position: node where on the street does one have to stop/give
way/wait for traffic signal
sign : node (optional) the exact location of the sign
from: the way one is following to which the action has to be applied
(is this needed ?)
intersection: the node of the intersection for which the sign/signal holds
to: the collection of ways one can travel to after stopping/giving
way/waiting for traffic signal. This would include the from way so
u-turns have to obey the sign/signal as well.

In general the to collection will be all ways leaving the
intersection, except for cases where right turns do not have to obey
the traffic signal, or where a right turn is give way and left +
straight ahead is stop.

is this how you see the relation ? Could it be simplified for the most
common case that the sign/traffic signal applies to all directions one
can travel ?
what in case there is a turn restriction at the intersection ? Does
the to-collection of the stop sign also have to include prohibited
turns ?

m.

On Thu, May 11, 2017 at 8:02 PM, Paul Johnson  wrote:
> On Thu, May 11, 2017 at 10:28 AM, Jean-Marc Liotier  wrote:
>>
>> On Sun, 7 May 2017 01:57:54 -0500
>> Paul Johnson  wrote:
>>
>> > I think it's time that we seriously reconsider how stop signs, yield
>> > signs and traffic calming devices are handled in all but the most
>> > simple (all approaches to the affected node apply) cases. [..] I'm
>> > thinking it's time to start mapping this similar to how we handle
>> > enforcement and turn restrictions, ie, with relations, for all but
>> > the simplest of cases, especially since the whole forward/backward
>> > direction=* thing is nonapplicable to nodes by design.
>>
>> Do you have in mind something like
>> https://wiki.openstreetmap.org/wiki/Relation:enforcement ?
>>
>> From the points of view of edition and data modeling, I believe that it
>> is the way forward.
>
>
> Yes, precisely.  It's at least a good starting point.
>
>>
>> From the point of view of data consumers, it requires grokking
>> relations - which is currently not common. Would that be a reason not
>> to choose that method ?
>
>
> I don't think so.  I consider Osmand to be the reference implementation for
> mobile navigation based off OSM data and it definitely understands
> enforcement relations.
>
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Thu, May 11, 2017 at 10:28 AM, Jean-Marc Liotier  wrote:

> On Sun, 7 May 2017 01:57:54 -0500
> Paul Johnson  wrote:
>
> > I think it's time that we seriously reconsider how stop signs, yield
> > signs and traffic calming devices are handled in all but the most
> > simple (all approaches to the affected node apply) cases. [..] I'm
> > thinking it's time to start mapping this similar to how we handle
> > enforcement and turn restrictions, ie, with relations, for all but
> > the simplest of cases, especially since the whole forward/backward
> > direction=* thing is nonapplicable to nodes by design.
>
> Do you have in mind something like
> https://wiki.openstreetmap.org/wiki/Relation:enforcement ?
>
> From the points of view of edition and data modeling, I believe that it
> is the way forward.
>

Yes, precisely.  It's at least a good starting point.


> From the point of view of data consumers, it requires grokking
> relations - which is currently not common. Would that be a reason not
> to choose that method ?
>

I don't think so.  I consider Osmand to be the reference implementation for
mobile navigation based off OSM data and it definitely understands
enforcement relations.
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Jean-Marc Liotier]

On Sun, 7 May 2017 01:57:54 -0500
Paul Johnson  wrote:

> I think it's time that we seriously reconsider how stop signs, yield
> signs and traffic calming devices are handled in all but the most
> simple (all approaches to the affected node apply) cases. [..] I'm
> thinking it's time to start mapping this similar to how we handle
> enforcement and turn restrictions, ie, with relations, for all but
> the simplest of cases, especially since the whole forward/backward
> direction=* thing is nonapplicable to nodes by design.

Do you have in mind something like
https://wiki.openstreetmap.org/wiki/Relation:enforcement ?

From the points of view of edition and data modeling, I believe that it
is the way forward.

From the point of view of data consumers, it requires grokking
relations - which is currently not common. Would that be a reason not
to choose that method ?

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Jo]

2017-05-07 9:30 GMT+02:00 Paul Johnson :

> On Sun, May 7, 2017 at 2:25 AM, Jo  wrote:
>
>> What about a type=traffic_sign relation?
>>
>> Where traffic_sign could be stop, give_way, parking
>>
>
> I was thinking the typical highway=* tags for highway=stop,
> highway=traffic_signals and highway=give_way.
>
>
>> In case of a stop sign, we could include the sign on a node, role 'sign'.
>> The node of the intersection, maybe role 'to'. The way the vehicle is
>> approaching from, maybe role 'from'.
>>
>
> Yes, that's what I'm suggesting.
>
>
>> In case of parking it would make very clear on which ways there is
>> parking and we would have central place to keep track of the conditions
>> like "opening_hours" and tariffs, or specific requirements like permits.
>>
>
> Parking already has a very clear case of way tagging for this.
>

To map the effect on the road network, I agree. To relate the traffic signs
to those ways, we don't. I don't know if we'll be able to map all traffic
signs and include all these relations (and keep them up-to-date), but then
I also wasn't sure we´d be able to put entire countries on the map like we
did, only ust a few years ago. It would be good to have a well thought out
plan, so we can start doing it consistently.

Polyglot
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Sun, May 7, 2017 at 2:25 AM, Jo  wrote:

> What about a type=traffic_sign relation?
>
> Where traffic_sign could be stop, give_way, parking
>

I was thinking the typical highway=* tags for highway=stop,
highway=traffic_signals and highway=give_way.


> In case of a stop sign, we could include the sign on a node, role 'sign'.
> The node of the intersection, maybe role 'to'. The way the vehicle is
> approaching from, maybe role 'from'.
>

Yes, that's what I'm suggesting.


> In case of parking it would make very clear on which ways there is parking
> and we would have central place to keep track of the conditions like
> "opening_hours" and tariffs, or specific requirements like permits.
>

Parking already has a very clear case of way tagging for this.
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Jo]

What about a type=traffic_sign relation?

Where traffic_sign could be stop, give_way, parking.

We can put a traffic_sign tag on nodes, where they get the
country_code:specific_national_code like BE:C1. Several traffic signs can
have an effect on several ways and nodes of the road network, so we could
group them in such relations.

In case of a stop sign, we could include the sign on a node, role 'sign'.
The node of the intersection, maybe role 'to'. The way the vehicle is
approaching from, maybe role 'from'.

In case of parking it would make very clear on which ways there is parking
and we would have central place to keep track of the conditions like
"opening_hours" and tariffs, or specific requirements like permits.

Polyglot


2017-05-07 8:57 GMT+02:00 Paul Johnson :

> I think it's time that we seriously reconsider how stop signs, yield signs
> and traffic calming devices are handled in all but the most simple (all
> approaches to the affected node apply) cases.  This largely after having a
> protracted discussion with one person about nodes lacking direction and
> this being a big factor in turn restrictions and enforcement being handled
> by relations already (and really, the entire reason relations were
> introduced in the first place).
>
> I'm thinking it's time to start mapping this similar to how we handle
> enforcement and turn restrictions, ie, with relations, for all but the
> simplest of cases, especially since the whole forward/backward direction=*
> thing is nonapplicable to nodes by design.
>
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
>
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Paul Johnson]

On Sun, May 7, 2017 at 2:12 AM, Nicolás Alvarez 
wrote:
>
> Do you know of a case where you would have a traffic calming device
> only affecting one direction, but not already have a reason to map
> each road direction as a separate way?
>

Somewhat commonly.  Oklahoma and Texas have a strong tendency to set down
at least three sets of rumble strips on approach to a traffic light, stop,
give_way or the side way of a T intersection when the speed limit is 55 or
faster.  Oklahoma Turnpike Authority also uses rumble strips to warn of the
toll plaza on the state's only undivided turnpike.


> I agree about the signs though. Relations add complexity, but I don't
> see how else to handle that kind of directional signs...
>

This is a problem that can be solved via editors.  id and JOSM both have
excellent editors for turn restrictions that could likely be readily
extended to handle enforcement, traffic calming, stop, yield and
nonstandard traffic signal cases (and clean up traffic signal situations on
SPUI interchanges, dual carriageways and other "messy" traffic light
situations that get mapped as multiple ways intersecting but are handled by
a single assembly of traffic lights or an interlocking pair of traffic
lights as functionally a single intersection).
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Revisiting traffic control and traffic calming

[Nicolás Alvarez]

2017-05-07 3:57 GMT-03:00 Paul Johnson :
> I think it's time that we seriously reconsider how stop signs, yield signs
> and traffic calming devices are handled in all but the most simple (all
> approaches to the affected node apply) cases.  This largely after having a
> protracted discussion with one person about nodes lacking direction and this
> being a big factor in turn restrictions and enforcement being handled by
> relations already (and really, the entire reason relations were introduced
> in the first place).
>
> I'm thinking it's time to start mapping this similar to how we handle
> enforcement and turn restrictions, ie, with relations, for all but the
> simplest of cases, especially since the whole forward/backward direction=*
> thing is nonapplicable to nodes by design.

Do you know of a case where you would have a traffic calming device
only affecting one direction, but not already have a reason to map
each road direction as a separate way?

I agree about the signs though. Relations add complexity, but I don't
see how else to handle that kind of directional signs...

-- 
Nicolás

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk