Re: [Tagging] Sidewalk Tagging for Routing

[John Eldredge]

Nashville, TN has a population of over 600,000 people, but, outside the 
central core of the city, less than half the streets have sidewalks. The 
primary reason for this is that most of the city's growth has been in the 
last 50 years, and the planners assumed that everyone would be traveling by 
car, not by public transportation or on foot.


--
John F. Eldredge -- j...@jfeldredge.com
"Darkness cannot drive out darkness; only light can do that. Hate cannot 
drive out hate; only love can do that." -- Martin Luther King, Jr.




On November 24, 2015 9:57:42 AM Clifford Snow  wrote:


On Tue, Nov 24, 2015 at 5:24 AM, John Willis  wrote:


Jaywalking will get you a bored policeman giving you a ticket in Tokyo.

They pride themselves in people who follow the rules and wait for
crosswalks and such.



Seattle is one of the few cities in the US that actively hands out
jaywalking tickets. We warn all newcomers not to jaywalk. Moving from
Chicago ment having to force myself not to jaywalk. In Chicago, even if
there was a cop directing traffic everyone crossed when and where they
wanted. Traffic be dammed.



However the sidewalk "grid" disappears very quickly (most unclassified and
residential streets have no sidewalks), but there are "green" zones on the
sides of some narrow roads (like narrow residential roads and back alleys
of Tokyo) where the shoulder of the road is expected to be used for
pedestrian access (not a full sidewalk).

This makes mapping the sidewalks quite easy in Japan - they disappear
quite quickly once you leave secondary roads. For Places where they they go
on and on and on - like huge planned housing communities in California -
it's a good question!



The city in question is Seattle. Most of Seattle's residential streets have
sidewalks as do most large cities in the US.





--
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch



--
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-25 22:18 GMT+01:00 Philip Barnes :

> > People with vision impairments or wheelchairs can't - so directing
> > them to crosswalks with kerb cuts/slopes and assisted signals
> > (sounds, etc) sounds like the proper thing to do.
>
> I agree, but it should not be done at the expense of pedestrian
> mapping.
>
> Using the existing style of highway attributes should be the way to
> achieve this, not adding parallel ways which do not exist on the
> ground.
>


the sidewalks/pavements DO exist on the ground. Adding tags to the main
highway way / street (i.e. splitting for everything) becomes unhandy as
soon as you add a lot of details (extreme  fragmentation of the main
highway way), and it is unsuitable for things like maxspeed on the
pavement/cycleway, surface, geometric details (shape) of the pavement,
things on the pavement like bollards, width reductions, etc.



>
> We should not be changing the database just because the existing system
> is too comlicated for new mappers.



I'm not sure if the current system is too complicated, but if I were
convinced it was and there was a simpler solution for the same problem, I'd
not hesitate to change the current system.

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[John Willis]

Javbw

> On Nov 26, 2015, at 6:18 AM, Philip Barnes  wrote:
> 
> adding parallel ways which do not exist on the
> ground.

Sidewalks are parallel ways with very different access restrictions that exist 
on the ground - so I'm not sure what you mean. I can easily see them and their 
crosswalks and bridges in arial imagery. 

Sidewalks accompany a road - but cars on sidewalks and people in the center of 
roads is a very bad thing, right? 

Frontage roads often accompany train tracks, but the frontage road is not an 
attribute of the track. Both sidewalks and frontage roads have different 
routing, different grades of construction, disappear and reappear, and have 
different access restrictions. They are *separate*. 

Roads and sidewalks often have heavy barriers between the road and the 
sidewalk. A kerb is a very big barrier to a wheelchair. A fence, bollard, 
hedges, *planter boxes, walls, bridge pilings, guardrails, and other barriers 
exist on urban areas to physically separate the road from the sidewalk, but for 
certain people (wheelchairs) a kerb can be just as effective a barrier as a 
wall. 

[*I need to create that barrier value]

Sidewalks can end when a road continues. Sidewalks can have width restrictions 
and hazards (poles, walls, etc) that affect the sidewalk but not the road. The 
sidewalk can have different routing (into a park or under a bridge with steps) 
rather than using the same paths as the road.  They may not share intersections 
with a road (bypass over it via footbridge). Sidewalks can have different 
surfaces, which change much more rapidly than the road does. 

What if it's different on one side of the road from the other ? We will need to 
keep tagging the road with "foo:sidewalk:left=*" and "foo:sidewalk:right=*" to 
keep everything straight?

All of this could be done by adding tags to the road, but wow! The segmentation 
that would occur! I don't care so much, but people running relations would. How 
would I show where a crosswalk or footbridge intersects - which is a separate 
way from the road but not the sidewalk??? 

What a giant, complex pile of newbie-nuking tag-salad! 

Or we can draw a way where the  sidewalk is and dump all that tag complexity 
onto it, completely separate from the road and it's attributes - and the 
sidewalk on the other side, let alone the poles and barriers and vending 
machines and bus stops and driveways and kerbs and all the other stuff - beyond 
buildings and driveways - that is easily mappable (and is being mapped) in 
urban imagery that exists somewhere around the sidewalk and the road.

It belongs on a separate way where imagery is good enough and when mapping time 
allows. 
Sidewalk=left/right/both on a road is a great stop-gap until better imagery is 
available and better mapping can be done - but by no means is sufficient in any 
dimension whatsoever for proper foot routing in urban envrons, which is where 
like 90% of foot routing is going to take place. Getting that last 500m from 
the train station to the destination can be a totally and completely different. 
Proper routing of each depends on proper tagging of each separate route, for 
ease of use, safety, and access restriction reasons. 

Javbw. 
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[John Willis]

> On Nov 26, 2015, at 6:09 AM, Philip Barnes  wrote:
> 
> Wheelchair routing would need to used sloped curbs, in most cases in a
> residential area these will be driveways

If they are in a residential area with sidewalks, then the sidewalks and their 
junctions with the roads are mappable. And then you will be able to get within 
a 1/2 block of the location with error free routing. 

If it is a very long road with a sidewalk on one side, driveways that intersect 
will share a node with the sidewalk and with the road, and can be used for the 
last 100m of routing, or a suggestion to cross the street in a safe location if 
there is no shared node between the road and the sidewalk for 50-100m around 
the destination on the other side of the street. 

If there is no sidewalk, then we are just mapping the road or a rough 
*highway=trail* unsuitable for wheelchairs or kids on bikes (now that 
highway=path is hopelessly confused with other tags in some regions [like 
Japan] - another thing that adds to this giant mess) to indicate that there is 
no proper sidewalk (and like almost all residential roads in Japan) you will be 
traveling in the road to your destination. 

If we want very accurate pedestrian routing with kerb slope and tactile paving 
tags, we need very accurate separate sidewalk and driveway mapping. Otherwise 
the routing is very similar to car or cycle mapping. 

Perhaps we need a highway=kerb_cut that is mappable or barrier=kerb_cut to map 
the extent (way) or intersection with a footpath (node) of where the sidewalk 
ends and crossings (a unlabeled crossing or a controlled marked one) begins . 
Barrier=kerb can be put on a footpath without cuts where it meets a road but 
has no cuts (common) as a barrier for cyclists and wheelchairs. Barrier=* has 
several different "openings" in the barriers. Having curb cut might be a good 
addition of we are going to be mapping kerb as ways on the road - mapping the 
long cuts for pedestrian crossings (and eventually driveways and other places 
where the curb is sloped to allow vehicle access.) 

I think Barrier kerb should default access to motive heckle=no foot=yes 
bicycle=dismount and wheelchair=no 

Javbw. 
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Wed, 2015-11-25 at 10:27 +0900, John Willis wrote:
> 
> > On Nov 25, 2015, at 9:14 AM, Philip Barnes 
> > wrote:
> > 
> > Remember
> > normal pedestrians can cross wherever they want
> 
> 
> People with vision impairments or wheelchairs can't - so directing
> them to crosswalks with kerb cuts/slopes and assisted signals
> (sounds, etc) sounds like the proper thing to do. 

I agree, but it should not be done at the expense of pedestrian
mapping.

Using the existing style of highway attributes should be the way to
achieve this, not adding parallel ways which do not exist on the
ground.

We should not be changing the database just because the existing system
is too comlicated for new mappers.

> 
> As long as we map the crosswalks and map a sidewalk's route across
> the road (regardless of if there is a marked signal controlled
> crosswalk) then it should be possible to route people across the
> streets where there is a road that crosses a street w/o a crosswalk,
> right? 
> 
> It would be better to have people walk 1/2 a block longer and use the
> crosswalks then having them try to jump across a road. 
I have no concept of how far a block is, but you seem to be assuming
that crosswalks exist everywhere. 

Phil (trigpoint)

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

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Tue, 2015-11-24 at 16:51 -0800, Clifford Snow wrote:
> 
> On Tue, Nov 24, 2015 at 4:14 PM, Philip Barnes 
> wrote:
> > They are not matching reality, can cause long detours and poor
> > routing
> > unless the mapper provides a lot of connections to the road.
> > Remember
> > normal pedestrians can cross wherever they want.
> > 
> That is true, but what we want to give someone in a wheelchair is a
> route that they can safely take. So at a minimum, the way must be
> connect, where appropriate, at intersections and other recognized
> crossings.  A recognized crossing would be one with markings for
> pedestrians or from local knowledge, that it is safe to cross the
> street. 
Outside of areas with heavy traffic on main roads or with high numbers
of pedestrians marked crossings do not exist, pedestrians normally
cross wherever they need to, obviously avoiding danger spots such as
blind corners.

Wheelchair routing would need to used sloped curbs, in most cases in a
residential area these will be driveways.


> >  
> > 
> > 
> > There is also the simple rendering issue, roads are already wide
> > and
> > are very close or clipping buildings. The sidewalk if mapped in
> > position is likely to be hidden under the road.
> > 
> I'm hoping that a wheelchair map would draw the sidewalks and
> minimize the streets.  That is a rendering issue. As we like to say,
> don't map for the renderer. 
You are proposing to change the database, other maps will be affected
too. If a separate footway is drawn, it will render alongside the road
on the default style, which will be misleading as there is not a
separate way.


> 
> > 
> > > Setting aside the newbie friendly issue, how do you map a
> > crosswalk
> > > in the middle of a street?
> > Add a node where the drop kerb is and map which side it belongs to.
> > 
> If I understand correctly, at a street crossing, the tags would be
> 
> highway=crossing
> kerb:(left/right/both)=lowered 

I would have thought only marked crossings should be tagged in this
way, otherwise they are just the mappers opinion.

> > 
> > 
> > > How do you map kerb slopes when the the slope is in the corner
> > the
> > > intersection?
> > I assume by kerb slope you mean a drop kerb? They are never in the
> > corner. Remember the way, when editing, represents the centreline
> > so
> > they are a few metres into the joining street. This also give
> > turning
> > vehicles space to stop for pedestrians.
> > 
> Unfortunately, at least around here, they often place the wheelchair
> ramp in the corner. My guess is it cost less money. As I said
> earlier, using the corner results in 4 instead of 8 ramps on a
> typical intersection. A picture of one of these ramps might help.
> I'll see if I can dig one up.

As I said the corner is not the point where the ways meet, in OSM terms
it is at least a lanes width from the that point, a node would work if
a suitable router was produced.

Phil (trigpoint)

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

Re: [Tagging] Sidewalk Tagging for Routing

[Paul Johnson]

On Tue, Nov 24, 2015 at 6:14 PM, Philip Barnes  wrote:

> On Tue, 2015-11-24 at 15:22 -0800, Clifford Snow wrote:
> >
> > On Tue, Nov 24, 2015 at 2:12 PM, Tobias Knerr 
> > wrote:
> > > Sure, but the sidewalk attribute is essential for other, much more
> > > basic
> > > use cases that separate ways fail to serve.
> > >
> > Can you elaborate on why separate ways fail to serve?
>
> They are not matching reality, can cause long detours and poor routing
> unless the mapper provides a lot of connections to the road. Remember
> normal pedestrians can cross wherever they want.
>

Sounds like something any reasonable router should be able to take into
account and recalculate against based on position.  I can rodeo to the
other side of an expressway and a couple freeways to hang a U-turn (with
varying degrees of physical possibility or legal acceptability) in my
region across the grass median but that doesn't necessarily mean that's a
smart move that should be suggested to the average person with no regional
experience.  Likewise, I didn't know jaywalking wasn't a thing in terms of
legal concepts in the UK until this thread and would be more prone to
expect a more comfortable route to where other road users would explicitly
expect me to take such an action.

Point is, the data should be on the assumption of a complete greenhorn or
someone who doesn't necessarily understand how getting around a particular
locale works is using the map.


> There is also the simple rendering issue, roads are already wide and
> are very close or clipping buildings. The sidewalk if mapped in
> position is likely to be hidden under the road.
>

I'm not seeing how this is a data problem.  It still routes, right?
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-25 8:59 GMT+01:00 Mateusz Konieczny :

> "unless the mapper provides a lot of connections to the
> road" - or data is properly used.



can you elaborate on this? How can you use data properly if important
information is missing (which footway belongs to which road, how are they
separated or not)?

Cheers,
Martin

PS: There's the area relation idea in the wiki, which allows you to connect
footways and the main road, and add barriers (or absence of barriers) in
between them either as geometry or as tags.
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Mateusz Konieczny]

On Wed, 25 Nov 2015 00:14:48 +
Philip Barnes  wrote:

> On Tue, 2015-11-24 at 15:22 -0800, Clifford Snow wrote:
> > 
> > On Tue, Nov 24, 2015 at 2:12 PM, Tobias Knerr 
> > wrote:
> > > Sure, but the sidewalk attribute is essential for other, much more
> > > basic
> > > use cases that separate ways fail to serve.
> > > 
> > Can you elaborate on why separate ways fail to serve?
> 
> They are not matching reality, can cause long detours and poor routing
> unless the mapper provides a lot of connections to the road. Remember
> normal pedestrians can cross wherever they want.

"Remember normal pedestrians can cross wherever they want." - Again,
that is not true for all places.

"unless the mapper provides a lot of connections to the
road" - or data is properly used.

> There is also the simple rendering issue, roads are already wide and
> are very close or clipping buildings. The sidewalk if mapped in
> position is likely to be hidden under the road.

And why it is supposed to be relevant?

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

Re: [Tagging] Sidewalk Tagging for Routing

[John Willis]

> On Nov 25, 2015, at 9:14 AM, Philip Barnes  wrote:
> 
> Remember
> normal pedestrians can cross wherever they want


People with vision impairments or wheelchairs can't - so directing them to 
crosswalks with kerb cuts/slopes and assisted signals (sounds, etc) sounds like 
the proper thing to do. 

As long as we map the crosswalks and map a sidewalk's route across the road 
(regardless of if there is a marked signal controlled crosswalk) then it should 
be possible to route people across the streets where there is a road that 
crosses a street w/o a crosswalk, right? 

It would be better to have people walk 1/2 a block longer and use the 
crosswalks then having them try to jump across a road. 

Perhaps there is some radius we can use -if there is no junction within 100m of 
the goal point which allows the walkway to cross the road, then suggest that 
the walker attempt to cross the road if they feel safe. Routes on 
bridges/embankments/tunnels or separated by a barrier would continue to be 
routed normally. 

Javbw


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

Re: [Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

On Tue, Nov 24, 2015 at 4:14 PM, Philip Barnes  wrote:

> They are not matching reality, can cause long detours and poor routing
> unless the mapper provides a lot of connections to the road. Remember
> normal pedestrians can cross wherever they want.
>

That is true, but what we want to give someone in a wheelchair is a route
that they can safely take. So at a minimum, the way must be connect, where
appropriate, at intersections and other recognized crossings.  A recognized
crossing would be one with markings for pedestrians or from local
knowledge, that it is safe to cross the street.

>
>

> There is also the simple rendering issue, roads are already wide and
> are very close or clipping buildings. The sidewalk if mapped in
> position is likely to be hidden under the road.
>

I'm hoping that a wheelchair map would draw the sidewalks and minimize the
streets.  That is a rendering issue. As we like to say, don't map for the
renderer.


> > Setting aside the newbie friendly issue, how do you map a crosswalk
> > in the middle of a street?
> Add a node where the drop kerb is and map which side it belongs to.
>

If I understand correctly, at a street crossing, the tags would be

highway=crossing
kerb:(left/right/both)=lowered

>
>
> > How do you map kerb slopes when the the slope is in the corner the
> > intersection?
> I assume by kerb slope you mean a drop kerb? They are never in the
> corner. Remember the way, when editing, represents the centreline so
> they are a few metres into the joining street. This also give turning
> vehicles space to stop for pedestrians.
>

Unfortunately, at least around here, they often place the wheelchair ramp
in the corner. My guess is it cost less money. As I said earlier, using the
corner results in 4 instead of 8 ramps on a typical intersection. A picture
of one of these ramps might help. I'll see if I can dig one up.

It wouldn't be accurate, but using the same logic as above,
(highway=crossing, kerb:(left/right/both)=lowered) would work.

>From taginfo, kerb=lowered seems to be the most popular usage and is
documented [1]. sidewalk:right:sloped_curb:end [2] is also documented but
has limited usage.

[1] https://wiki.openstreetmap.org/wiki/Wheelchair_routing
[2] https://wiki.openstreetmap.org/wiki/Proposed_features/kerb

You have convinced me that the attribute method is optimal.

Thanks,
Clifford

-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Tue, 2015-11-24 at 15:22 -0800, Clifford Snow wrote:
> 
> On Tue, Nov 24, 2015 at 2:12 PM, Tobias Knerr 
> wrote:
> > Sure, but the sidewalk attribute is essential for other, much more
> > basic
> > use cases that separate ways fail to serve.
> > 
> Can you elaborate on why separate ways fail to serve?

They are not matching reality, can cause long detours and poor routing
unless the mapper provides a lot of connections to the road. Remember
normal pedestrians can cross wherever they want.

There is also the simple rendering issue, roads are already wide and
are very close or clipping buildings. The sidewalk if mapped in
position is likely to be hidden under the road.
>  
> > 
> > > I don't think relations are the right answer either. It isn't
> > easy to
> > > teach new people how to add relations. We should be able to
> > identify
> > > footpaths adjacent to roads by their spatial characteristics.
> > 
> > But we CAN'T, at least not in any reliable fashion. "Just draw a
> > way
> > somewhere near the road and magic computers sort out the mess" is,
> > frankly, not a viable option.
> > 
> > There has to be some explicit connection between sidewalks and
> > their
> > roads. And as soon as you take this into account, the separate
> > sidewalks
> > stop being newbie friendly at all. They only seem that way because
> > you
> > pretend the main challenge does not exist.
> > 
> Setting aside the newbie friendly issue, how do you map a crosswalk
> in the middle of a street? 
Add a node where the drop kerb is and map which side it belongs to.


> How do you map kerb slopes when the the slope is in the corner the
> intersection? 
I assume by kerb slope you mean a drop kerb? They are never in the
corner. Remember the way, when editing, represents the centreline so
they are a few metres into the joining street. This also give turning
vehicles space to stop for pedestrians.

Phil (trigpoint)


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

Re: [Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

On Tue, Nov 24, 2015 at 2:12 PM, Tobias Knerr  wrote:

> Sure, but the sidewalk attribute is essential for other, much more basic
> use cases that separate ways fail to serve.
>

Can you elaborate on why separate ways fail to serve?


>
> > I don't think relations are the right answer either. It isn't easy to
> > teach new people how to add relations. We should be able to identify
> > footpaths adjacent to roads by their spatial characteristics.
>
> But we CAN'T, at least not in any reliable fashion. "Just draw a way
> somewhere near the road and magic computers sort out the mess" is,
> frankly, not a viable option.
>
> There has to be some explicit connection between sidewalks and their
> roads. And as soon as you take this into account, the separate sidewalks
> stop being newbie friendly at all. They only seem that way because you
> pretend the main challenge does not exist.
>

Setting aside the newbie friendly issue, how do you map a crosswalk in the
middle of a street? How do you map kerb slopes when the the slope is in the
corner the intersection? I realize that both are possible, but the
complexity of the tagging required increases significantly over a separate
way.

-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Richard]

On Wed, Nov 25, 2015 at 12:09:23AM +0900, johnw wrote:

> having the man_made=bridge share layers with the roads and sidewalks does 
> work for all but a handful of bridges (I like that tag) - but assuming the 
> bridge is a single layer really makes things difficult for large/iconic/odd 
> bridges. 

see 
https://wiki.openstreetmap.org/wiki/Proposed_features/Simplify_man_made%3Dbridge_mapping#Multi-level_bridges

Richard


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

Re: [Tagging] Sidewalk Tagging for Routing

[Tobias Knerr]

On 24.11.2015 17:21, Clifford Snow wrote:
> Thanks everyone for the input. As much as I like the concept of using
> the sidewalk attribute to the road, it doesn't seem like it is all that
> useful for adding kerb slope.

Sure, but the sidewalk attribute is essential for other, much more basic
use cases that separate ways fail to serve.

> I don't think relations are the right answer either. It isn't easy to
> teach new people how to add relations. We should be able to identify
> footpaths adjacent to roads by their spatial characteristics. 

But we CAN'T, at least not in any reliable fashion. "Just draw a way
somewhere near the road and magic computers sort out the mess" is,
frankly, not a viable option.

There has to be some explicit connection between sidewalks and their
roads. And as soon as you take this into account, the separate sidewalks
stop being newbie friendly at all. They only seem that way because you
pretend the main challenge does not exist.

> Should we depreciate the sidewalk=left/right/both/none tag in favor of
> drawing separate footpaths?

No.

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

Re: [Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

Thanks everyone for the input. As much as I like the concept of using the
sidewalk attribute to the road, it doesn't seem like it is all that useful
for adding kerb slope. It wasn't pointed out, but one additional problem
with using the sidewalk attribute is that a new mapper will likely not look
at the tags on the road and draw in a footpath, duplicating the existing
one.

I don't think relations are the right answer either. It isn't easy to teach
new people how to add relations. We should be able to identify footpaths
adjacent to roads by their spatial characteristics.

Should we depreciate the sidewalk=left/right/both/none tag in favor of
drawing separate footpaths? The problem with the highway attribute feature
is as soon as the there is a special circumstance the editor is obligated
to draw a separate footpath. For example, a crossing in the middle of the
street. They happen all the time.

Mapping sidewalks as an attribute to an existing road is easy. I did my
little town. But adding in wheelchair access, seems to make that method
look very difficult. Looking to the future, more communities will (should)
want to map wheelchair access.

Clifford

On Tue, Nov 24, 2015 at 7:54 AM, Philip Barnes  wrote:

> On Tue Nov 24 15:39:45 2015 GMT, johnw wrote:
> >
> > > On Nov 24, 2015, at 11:32 PM, Martin Koppenhoefer <
> dieterdre...@gmail.com> wrote:
> > >
> > > the road, they are quite capable of deciding which side to walk, where
> > > to cross and whether it is simply easier to walk on the road.
> >
> >
> > TL;DR:
> > Although I care more about the rendering than the routing, the routing
> in this situation needs to be very explicit in places because sidewalks are
> crazy in some countries. Simple routing may work in most of the globe, but
> there are some cities/countries where pedestrian access conventions require
> very explicit mapping and routing.
> >
> > ~~
> >
> > I can imagine that this is a non-issue in a lot of places, as I grew up
> where getting to the other side of the street was not such a big deal, nor
> was what side of the street you were on a big deal when looking for routing
> information.
> >
> > But there are plenty of places - or at least city centers -  where it is
> a big deal.
> >
> > There are plenty of situations out here in small Japanese towns (let
> alone Tokyo) where which side of the street you are on is very important
> for routing, and it is impossible to change sides in a convenient/safe spot
> after a decision has been made. Some have dead-end sidewalks (that end in
> walls and a narrow shoulder)  that put you into extremely hazardous
> situations (being right next to traffic against a wall/guardrail - which
> they do all the time), or put you onto walkways onto bridges where you
> cannot take walkways that lead away from the opposite side, or there are
> access stairs to the pedestrian walkways that run under the bridge that
> only connect on one side of the bridge (leading to 300 meters or more of
> walking around the longer route).  as most Japanese towns are in valleys or
> near water, there are tons of bridges everywhere - some with really weird
> pedestrian routing restrictions.
> >
> > given the absolute psychotic nature of Japanese sidewalks throughout the
> whole country, explicit sidewalk mapping is a very good thing. I was just
> in Tokyo this evening - and I used 6 different kinds of footpaths - two
> were sidewalks with severe access restrictions to keep people out of the
> street (large steel pipe fencing along the road to keep people from
> Jaywalking) except at crosswalks, down to one way alleys with the green
> paths.  choosing the correct side of the street makes a difference for
> certain ped access bridges to other buildings, to subway and train
> entrances (which by no means are accessible from the other side - steel
> pipe barrier and all), or lead to completely different layers and tunnels
> depending on your side of the road.
> >
> You are describing situations where there are separate sidewalks, you are
> mentioning fences and other restrictions. In these cases I would map
> sidewalks as separate ways.
>
> Both methods, separate ways or as highway tags, are fine and should be
> used as appropriate.
>
> Phil (trigpoint)
> --
> Sent from my Jolla
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
>



-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

On Tue, Nov 24, 2015 at 5:24 AM, John Willis  wrote:

> Jaywalking will get you a bored policeman giving you a ticket in Tokyo.
>
> They pride themselves in people who follow the rules and wait for
> crosswalks and such.
>

Seattle is one of the few cities in the US that actively hands out
jaywalking tickets. We warn all newcomers not to jaywalk. Moving from
Chicago ment having to force myself not to jaywalk. In Chicago, even if
there was a cop directing traffic everyone crossed when and where they
wanted. Traffic be dammed.

>
> However the sidewalk "grid" disappears very quickly (most unclassified and
> residential streets have no sidewalks), but there are "green" zones on the
> sides of some narrow roads (like narrow residential roads and back alleys
> of Tokyo) where the shoulder of the road is expected to be used for
> pedestrian access (not a full sidewalk).
>
> This makes mapping the sidewalks quite easy in Japan - they disappear
> quite quickly once you leave secondary roads. For Places where they they go
> on and on and on - like huge planned housing communities in California -
> it's a good question!
>

The city in question is Seattle. Most of Seattle's residential streets have
sidewalks as do most large cities in the US.





-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Tue Nov 24 15:39:45 2015 GMT, johnw wrote:
> 
> > On Nov 24, 2015, at 11:32 PM, Martin Koppenhoefer  
> > wrote:
> > 
> > the road, they are quite capable of deciding which side to walk, where
> > to cross and whether it is simply easier to walk on the road.
> 
> 
> TL;DR:
> Although I care more about the rendering than the routing, the routing in 
> this situation needs to be very explicit in places because sidewalks are 
> crazy in some countries. Simple routing may work in most of the globe, but 
> there are some cities/countries where pedestrian access conventions require 
> very explicit mapping and routing. 
> 
> ~~
> 
> I can imagine that this is a non-issue in a lot of places, as I grew up where 
> getting to the other side of the street was not such a big deal, nor was what 
> side of the street you were on a big deal when looking for routing 
> information. 
> 
> But there are plenty of places - or at least city centers -  where it is a 
> big deal. 
> 
> There are plenty of situations out here in small Japanese towns (let alone 
> Tokyo) where which side of the street you are on is very important for 
> routing, and it is impossible to change sides in a convenient/safe spot after 
> a decision has been made. Some have dead-end sidewalks (that end in walls and 
> a narrow shoulder)  that put you into extremely hazardous situations (being 
> right next to traffic against a wall/guardrail - which they do all the time), 
> or put you onto walkways onto bridges where you cannot take walkways that 
> lead away from the opposite side, or there are access stairs to the 
> pedestrian walkways that run under the bridge that only connect on one side 
> of the bridge (leading to 300 meters or more of walking around the longer 
> route).  as most Japanese towns are in valleys or near water, there are tons 
> of bridges everywhere - some with really weird pedestrian routing 
> restrictions. 
> 
> given the absolute psychotic nature of Japanese sidewalks throughout the 
> whole country, explicit sidewalk mapping is a very good thing. I was just in 
> Tokyo this evening - and I used 6 different kinds of footpaths - two were 
> sidewalks with severe access restrictions to keep people out of the street 
> (large steel pipe fencing along the road to keep people from Jaywalking) 
> except at crosswalks, down to one way alleys with the green paths.  choosing 
> the correct side of the street makes a difference for certain ped access 
> bridges to other buildings, to subway and train entrances (which by no means 
> are accessible from the other side - steel pipe barrier and all), or lead to 
> completely different layers and tunnels depending on your side of the road. 
> 
You are describing situations where there are separate sidewalks, you are 
mentioning fences and other restrictions. In these cases I would map sidewalks 
as separate ways.

Both methods, separate ways or as highway tags, are fine and should be used as 
appropriate. 

Phil (trigpoint)
-- 
Sent from my Jolla
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[johnw]

> On Nov 24, 2015, at 11:32 PM, Martin Koppenhoefer  
> wrote:
> 
> the road, they are quite capable of deciding which side to walk, where
> to cross and whether it is simply easier to walk on the road.


TL;DR:
Although I care more about the rendering than the routing, the routing in this 
situation needs to be very explicit in places because sidewalks are crazy in 
some countries. Simple routing may work in most of the globe, but there are 
some cities/countries where pedestrian access conventions require very explicit 
mapping and routing. 

~~

I can imagine that this is a non-issue in a lot of places, as I grew up where 
getting to the other side of the street was not such a big deal, nor was what 
side of the street you were on a big deal when looking for routing information. 

But there are plenty of places - or at least city centers -  where it is a big 
deal. 

There are plenty of situations out here in small Japanese towns (let alone 
Tokyo) where which side of the street you are on is very important for routing, 
and it is impossible to change sides in a convenient/safe spot after a decision 
has been made. Some have dead-end sidewalks (that end in walls and a narrow 
shoulder)  that put you into extremely hazardous situations (being right next 
to traffic against a wall/guardrail - which they do all the time), or put you 
onto walkways onto bridges where you cannot take walkways that lead away from 
the opposite side, or there are access stairs to the pedestrian walkways that 
run under the bridge that only connect on one side of the bridge (leading to 
300 meters or more of walking around the longer route).  as most Japanese towns 
are in valleys or near water, there are tons of bridges everywhere - some with 
really weird pedestrian routing restrictions. 

given the absolute psychotic nature of Japanese sidewalks throughout the whole 
country, explicit sidewalk mapping is a very good thing. I was just in Tokyo 
this evening - and I used 6 different kinds of footpaths - two were sidewalks 
with severe access restrictions to keep people out of the street (large steel 
pipe fencing along the road to keep people from Jaywalking) except at 
crosswalks, down to one way alleys with the green paths.  choosing the correct 
side of the street makes a difference for certain ped access bridges to other 
buildings, to subway and train entrances (which by no means are accessible from 
the other side - steel pipe barrier and all), or lead to completely different 
layers and tunnels depending on your side of the road. 


Javbw___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 16:09 GMT+01:00 johnw :

> Speaking of layers & bridges..




actually, layers and bridges do not pose a problem, as long as it's all the
same bridge. One object is ok. In the case of the spiral bridge you might
need a relation to make it clear.

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[johnw]

> On Nov 24, 2015, at 9:20 PM, Martin Koppenhoefer  
> wrote:
> 
> yes, it would eventually create problems in case e.g. a bridge spirals around 
> itself (guess improbable case) similar to spiral stairs (maybe these wouldn't 
> be called bridges but ramps). It worked for all bridges I have mapped so far


I have been on all these bridges, and walked the walkways mentioned. 

- a “loop bridge” it goes around 720 degrees. (two full turns). it is not a 
ramp - it is a feature of the trunk road.

https://goo.gl/maps/7U2vg4TBmaG2 

The rainbow bridge in Tokyo. there is tollway traffic, roadway traffic, a tram, 
and two separate walking paths with viewpoints and elevators in the bridge 
itself (not the  approaches). it’s two layers, unless you count the 2 
additional layers for the walkways and elevators in the support towers to 
ground level below the road bed level(s). 


Speaking of layers & bridges..

-  Double decker motorway bridges crossing the Arakawa river in Tokyo 

https://goo.gl/maps/64xyvf4LH6k  (where 6 
meets C2 downtown)


- Naruto Bridge

https://goo.gl/maps/K37HDTsATvq 

I know some bridges have road traffic going in different directions on 
different levels, but it's considered the same bridge. 

This bridge has a full observation deck and view point under the roadway (only 
from the west shore to he western tower, not the full distance) for people to 
pay to walk out and see the Natural whirlpools that form from the inland sea 
tide rush over a submerged ridge (visible in the aerial imagery). 

it is the bridge - but there is a whole facility down there *in the bridge* 
below the main road bed for pedestrian use, connected to a building onshore 
that sells admission to the walkway. 

having the man_made=bridge share layers with the roads and sidewalks does work 
for all but a handful of bridges (I like that tag) - but assuming the bridge is 
a single layer really makes things difficult for large/iconic/odd bridges. 

Javbw___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Marc Gemis]

There is already a page on the wiki, especially section [1]. Those
properties are used by the wheelchair navigation tool mentioned at the
top of that page.

regards

m

[1] 
http://wiki.openstreetmap.org/wiki/Wheelchair_routing#Sidewalks_and_properties

On Tue, Nov 24, 2015 at 3:02 PM,   wrote:
> On Tue Nov 24 13:40:38 2015 GMT, Marc Gemis wrote:
>> On Tue, Nov 24, 2015 at 1:48 PM, Philip Barnes  wrote:
>> >
>> > Sidewalks, unless they are physically separated in some way are an
>> > integral part of the highway. Sidewalk tags can allow a router to, maybe
>> > prefer, roads with sidewalks but there a lot of cases where this would
>> > just be plain annoying. A pedestrian simply needs to be told to follow
>> > the road, they are quite capable of deciding which side to walk, where
>> > to cross and whether it is simply easier to walk on the road.
>> >
>>
>>
>> Then we do not have an answer for Clifford's original question: how do
>> you tag sidewalk characteristics for wheelchair users: such as kerb,
>> slope, width, surface ? Especially when tagging is done by novices ?
>>
> I was thinking that, not easy but important.  We do need to do this, just not 
> so that it breaks normal pedestrian routing.
>
> I guess the things that need mapping are:
> drop kerbs - Probably a highway node,  sidewalk:drop_kerb=left/right/both. We 
> will need to map driveway drop kerbs too as these will be crossing point too.
>
> width - very important as many sidewalks are singlefile for walking.
> sidewalk:left:width=0.6, or 1.5 etc
>
>  kerb height - Again important,  different heights are barriers for different 
> people, different wheelchairs.  sidewalk:left:kerb_height=10 cm.
>
> Phil (trigpoint)
> --
> Sent from my Jolla
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging

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

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 13:48 GMT+01:00 Philip Barnes :

> A pedestrian simply needs to be told to follow
> the road, they are quite capable of deciding which side to walk, where
> to cross and whether it is simply easier to walk on the road.
>


that's partly true, but there are groups of people who could benefit from
more detailed descriptions: the blind, people in wheelchairs, routers that
try to avoid less safe situations (e.g. for kids), and generally, knowing
about the actual details for pedestrians will improve the routing for every
pedestrian, because a better route can be found (less true in rural
settings with less alternatives, obviously).

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[phil]

On Tue Nov 24 13:40:38 2015 GMT, Marc Gemis wrote:
> On Tue, Nov 24, 2015 at 1:48 PM, Philip Barnes  wrote:
> >
> > Sidewalks, unless they are physically separated in some way are an
> > integral part of the highway. Sidewalk tags can allow a router to, maybe
> > prefer, roads with sidewalks but there a lot of cases where this would
> > just be plain annoying. A pedestrian simply needs to be told to follow
> > the road, they are quite capable of deciding which side to walk, where
> > to cross and whether it is simply easier to walk on the road.
> >
> 
> 
> Then we do not have an answer for Clifford's original question: how do
> you tag sidewalk characteristics for wheelchair users: such as kerb,
> slope, width, surface ? Especially when tagging is done by novices ?
> 
I was thinking that, not easy but important.  We do need to do this, just not 
so that it breaks normal pedestrian routing.

I guess the things that need mapping are: 
drop kerbs - Probably a highway node,  sidewalk:drop_kerb=left/right/both. We 
will need to map driveway drop kerbs too as these will be crossing point too. 

width - very important as many sidewalks are singlefile for walking.
sidewalk:left:width=0.6, or 1.5 etc

 kerb height - Again important,  different heights are barriers for different 
people, different wheelchairs.  sidewalk:left:kerb_height=10 cm.

Phil (trigpoint)
-- 
Sent from my Jolla
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Marc Gemis]

On Tue, Nov 24, 2015 at 1:48 PM, Philip Barnes  wrote:
>
> Sidewalks, unless they are physically separated in some way are an
> integral part of the highway. Sidewalk tags can allow a router to, maybe
> prefer, roads with sidewalks but there a lot of cases where this would
> just be plain annoying. A pedestrian simply needs to be told to follow
> the road, they are quite capable of deciding which side to walk, where
> to cross and whether it is simply easier to walk on the road.
>


Then we do not have an answer for Clifford's original question: how do
you tag sidewalk characteristics for wheelchair users: such as kerb,
slope, width, surface ? Especially when tagging is done by novices ?


regards

m

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

Re: [Tagging] Sidewalk Tagging for Routing

[John Willis]

> On Nov 24, 2015, at 6:40 PM, Richard Fairhurst  wrote:
> 
> Jaywalking is a North American concept.

Jaywalking will get you a bored policeman giving you a ticket in Tokyo. 

They pride themselves in people who follow the rules and wait for crosswalks 
and such. 

However the sidewalk "grid" disappears very quickly (most unclassified and 
residential streets have no sidewalks), but there are "green" zones on the 
sides of some narrow roads (like narrow residential roads and back alleys of 
Tokyo) where the shoulder of the road is expected to be used for pedestrian 
access (not a full sidewalk). 

This makes mapping the sidewalks quite easy in Japan - they disappear quite 
quickly once you leave secondary roads. For Places where they they go on and on 
and on - like huge planned housing communities in California - it's a good 
question! 

Javbw
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Tue, 2015-11-24 at 09:01 +0100, Marc Gemis wrote:
> In Belgium several primary/secondary roads go through town centers
> (e.g. [1], [2]), where the maximum speed is reduced to 50 km/h (same
> as for residential roads).  Most people will cross the street where it
> is most suitable (especially near [2] where the road is even pretty
> narrow), not necessarily at marked pedestrian crossings.
> I guess Jaywalking is pretty common in Belgium.
> 
> And what about the obligation to walk on the left side of the road
> when there is no sidewalk ? Should a router take that into account and
> let you walk to the next road crossing to switch sides ? Would be
> pretty neat.

It would be very annoying. 

The rule, or recommendation as its not always the sensible thing to do,
is to walk facing oncoming traffic, its the right in the UK. Walkers
will decide what is sensible depending on the situation. 

It is not a routers job to dictate rules but to simply guide the user in
the right direction.

Phil (trigpoint)


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

Re: [Tagging] Sidewalk Tagging for Routing

[Colin Smale]

 

On 2015-11-24 13:20, Martin Koppenhoefer wrote: 

> 2015-11-24 12:43 GMT+01:00 Colin Smale :
> 
>> One issue with dual carriageways (and now I think about it, also railway 
>> lines) is about generalisations at certain zoom levels. If you zoom out 
>> beyond a certain level, both halves of a DC (or the individual tracks of a 
>> railway) would be better modelled as a single line. The renderer/consumer 
>> needs an algorithm, and/or hints from the tagging, to know what belongs to 
>> what.
> 
> yes, generalization would be nice sometimes for zoomed out levels, but a 
> relation would not help much. You still have to judge which part can be 
> omitted and which not (or how to combine the two (or more) into one), at 
> least for situations where both parts are not strictly parallel.
> 
> See here how it works not too bad like it is done now:
> http://www.openstreetmap.org/#map=16/42.1322/13.1322
> http://www.openstreetmap.org/#map=15/42.1314/13.1323

I see how the casing on the lines gets suppressed where they overlap,
but you can see the name being rendered individually on the two halves.
And I suspect the shields with the road numbers are more numerous than
they need to be. That casing-suppression looks a bit wierd sometimes
where two roads are geometrically adjacent WITHOUT being logically
related, for example here (the north-south road right in the middle of
the map, where it is labelled "Canterbury Way"). Two parallel but
unrelated roads, of different classes, are being blended into one: 

http://www.openstreetmap.org/#map=15/51.4347/0.2426 

>> They can provide certainty where otherwise complex heuristics would be 
>> needed which may or may not work in all cases. Two streets with the same 
>> name isn't good enough on its own IMHO. The two halves of a dual carriageway 
>> may have different names and still be the same road.
> 
> Given that there is no such unambiguous thing in the real world like "the 
> same road", you'd always have to judge based on a common definition. There 
> will likely be arguments for both ways of looking at it: same road or 
> different roads. A relation could make it clear how the mapper saw it, but 
> I'd already question the concept on a more general level: what is "the same 
> road"?

The definition of "the same road" will probably depend on the specific
use case. The use case may be rendering, to produce a nice-looking map
on the screen or on paper, or it MIGHT be navigation-related. My idea is
to remove the constituents of the relation and replace them with a
simplified version, which looks/behaves more suitably for the given zoom
level. Keeping all the roads properly connected will be a helluva job,
so it might not be viable as a simplification of a routing network, but
giving the resulting turn directions based on the simplified model may
have some benefits; not sure about this though. 

//colin ___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Philip Barnes]

On Mon, 2015-11-23 at 19:14 -0500, Bryan Housel wrote:
> Current preference seems to be map sidewalks as separate ways tagged as 
> `highway=footway + footway=sidewalk`.
> 
> 1. As you mentioned, you can use sidewalk-specific tags (slope, surface) 
> without affecting the adjacent highway.
> 2. The sidewalk and the road really are separate features.
> 3. New users like to trace and they are going to map it like this anyway.
>  
> I don’t really buy the “mapping sidewalks as part of the road makes routing 
> easier" argument, but I don’t work on routing software so maybe there really 
> is something complicated about routing that I am unaware of.
> 
-1 000 000

Sidewalks, unless they are physically separated in some way are an
integral part of the highway. Sidewalk tags can allow a router to, maybe
prefer, roads with sidewalks but there a lot of cases where this would
just be plain annoying. A pedestrian simply needs to be told to follow
the road, they are quite capable of deciding which side to walk, where
to cross and whether it is simply easier to walk on the road.

Mapping sidewalks separately will result in routing to junctions, which
are the worse place for crossing the road, or am I expected to connect
the sidewalk to the road every metre or so to allow real world
pedestrian routing?

Phil (trigpoint)


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

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 12:43 GMT+01:00 Colin Smale :

> One issue with dual carriageways (and now I think about it, also railway
> lines) is about generalisations at certain zoom levels. If you zoom out
> beyond a certain level, both halves of a DC (or the individual tracks of a
> railway) would be better modelled as a single line. The renderer/consumer
> needs an algorithm, and/or hints from the tagging, to know what belongs to
> what.
>


yes, generalization would be nice sometimes for zoomed out levels, but a
relation would not help much. You still have to judge which part can be
omitted and which not (or how to combine the two (or more) into one), at
least for situations where both parts are not strictly parallel.

See here how it works not too bad like it is done now:
http://www.openstreetmap.org/#map=16/42.1322/13.1322
http://www.openstreetmap.org/#map=15/42.1314/13.1323

or you'd want to see actually 2 separate ways still in z14:
http://www.openstreetmap.org/#map=14/42.4821/13.6116

or even in zoom 10:
http://www.openstreetmap.org/?mlat=49.6009&mlon=11.3557#map=10/49.6014/11.3551




> On the subject of man_made=bridge, that relies on geometry to associate
> the parts. That loose association only works because bridges are
> straight(ish).
>


yes, it would eventually create problems in case e.g. a bridge spirals
around itself (guess improbable case) similar to spiral stairs (maybe these
wouldn't be called bridges but ramps). It worked for all bridges I have
mapped so far. It is a very intuitive and easy approach, leads to very
simple mapping results (typically just a rectangle of the bridge outline),
i.e. it is easily adoptable and maintainable, is supported by even the more
simple editing programs.

You can call that a loose association because it "only" relies on spatial
properties, but I bet it is more sustainable and reliable than any solution
that requires mappers to explicitly search for a relation and add parts to
it when they modify some road. And it stores the outline of the bridge
you'd need anyway if you wanted to add some significant detail. More
complex cases can still be modelled with a bridge relation anyway, but
these are very rare compared to the rest of bridges.



> No need for relations where there is no need for them? Sounds a bit like a
> circular definition. If a relation is the right way to model reality, we
> should also not run away from them.
>

yes, of course. Routes, multipolygons, turn restrictions, boundaries, etc.
There are lots of good reasons for relations in our current data model.



> They can provide certainty where otherwise complex heuristics would be
> needed which may or may not work in all cases. Two streets with the same
> name isn't good enough on its own IMHO. The two halves of a dual
> carriageway may have different names and still be the same road.
>

Given that there is no such unambiguous thing in the real world like "the
same road", you'd always have to judge based on a common definition. There
will likely be arguments for both ways of looking at it: same road or
different roads. A relation could make it clear how the mapper saw it, but
I'd already question the concept on a more general level: what is "the same
road"?

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Colin Smale]

 

One issue with dual carriageways (and now I think about it, also railway
lines) is about generalisations at certain zoom levels. If you zoom out
beyond a certain level, both halves of a DC (or the individual tracks of
a railway) would be better modelled as a single line. The
renderer/consumer needs an algorithm, and/or hints from the tagging, to
know what belongs to what. Often a DC is very recognisable as two
one-ways in opposite directions with the same name/ref and "more or less
parallel", but I imagine it's that last bit which can cause problems if
the roads are not reasonably straight. 

On the subject of man_made=bridge, that relies on geometry to associate
the parts. That loose association only works because bridges are
straight(ish). 

No need for relations where there is no need for them? Sounds a bit like
a circular definition. If a relation is the right way to model reality,
we should also not run away from them. They can provide certainty where
otherwise complex heuristics would be needed which may or may not work
in all cases. Two streets with the same name isn't good enough on its
own IMHO. The two halves of a dual carriageway may have different names
and still be the same road. 

//colin 

On 2015-11-24 12:17, Martin Koppenhoefer wrote: 

> 2015-11-24 11:45 GMT+01:00 Colin Smale :
> 
>> So has the "street" relation just been born? It could solve some other 
>> puzzles as well: dual carriageways, cycle tracks, bridges...
> 
> bridges have been solved by introducing a revolutionary concept: a dedicated 
> object for an explicit bridge. It remains a mystery why it took us 10 years 
> ;-) 
> Have a look at man_made=bridge, doesn't even require relations.
> 
> The "street relation" has not only been born, it has already died, at least 
> almost. Like its cousing, the associatedstreet street (which seems more live 
> than the street relation). Honestly, I don't understand what these relations 
> could solve in conjunction with dual carriageways, could you explain?
> 
> Generally we should avoid relations where they are not necessary. No need for 
> a relation to say these street parts do share the same name. Simply put the 
> same name on them. 
> 
> Cheers, 
> Martin 
> 
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
 ___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 11:45 GMT+01:00 Colin Smale :

> So has the "street" relation just been born? It could solve some other
> puzzles as well: dual carriageways, cycle tracks, bridges...



bridges have been solved by introducing a revolutionary concept: a
dedicated object for an explicit bridge. It remains a mystery why it took
us 10 years ;-)
Have a look at man_made=bridge, doesn't even require relations.

The "street relation" has not only been born, it has already died, at least
almost. Like its cousing, the associatedstreet street (which seems more
live than the street relation). Honestly, I don't understand what these
relations could solve in conjunction with dual carriageways, could you
explain?

Generally we should avoid relations where they are not necessary. No need
for a relation to say these street parts do share the same name. Simply put
the same name on them.

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Gerd Petermann]

It already exists for that but is rarely used, probably because of

the ususal problems:

- Sometimes roads are split to give one part a different name,

the type=street relations are typically ignored.

- Since it is not often used mappers will forget to add a new sidewalk

to such a relation.

Not that much a problem with the relation itself, more one of the

editors which could handle these cases better than they do now.


Gerd


Von: Colin Smale 
Gesendet: Dienstag, 24. November 2015 11:45
An: tagging@openstreetmap.org
Betreff: Re: [Tagging] Sidewalk Tagging for Routing






So has the "street" relation just been born? It could solve some other puzzles 
as well: dual carriageways, cycle tracks, bridges...

On 2015-11-24 11:40, Martin Koppenhoefer wrote:

2015-11-24 11:30 GMT+01:00 Gerd Petermann 
mailto:gpetermann_muenc...@hotmail.com>>:

And this problem is not easy to solve by programs.

Up to now I don't know any efficient algo that allows to

find the nearest named road that goes parallel

to a selected unnamed way. It would be of great help

here.


we should ask contributors to add this information explicitly, also because an 
algorithmic method will never solve all cases.

Cheers,
Martin

___
Tagging mailing list
Tagging@openstreetmap.org<mailto:Tagging@openstreetmap.org>
https://lists.openstreetmap.org/listinfo/tagging
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Colin Smale]

 

So has the "street" relation just been born? It could solve some other
puzzles as well: dual carriageways, cycle tracks, bridges... 

On 2015-11-24 11:40, Martin Koppenhoefer wrote: 

> 2015-11-24 11:30 GMT+01:00 Gerd Petermann :
> 
>> And this problem is not easy to solve by programs. 
>> 
>> Up to now I don't know any efficient algo that allows to 
>> 
>> find the nearest named road that goes parallel  
>> 
>> to a selected unnamed way. It would be of great help 
>> 
>> here.
> 
> we should ask contributors to add this information explicitly, also because 
> an algorithmic method will never solve all cases.
> 
> Cheers, 
> Martin 
> 
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
 ___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 11:30 GMT+01:00 Gerd Petermann :

> And this problem is not easy to solve by programs.
>
> Up to now I don't know any efficient algo that allows to
>
> find the nearest named road that goes parallel
>
> to a selected unnamed way. It would be of great help
>
> here.
>


we should ask contributors to add this information explicitly, also because
an algorithmic method will never solve all cases.

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Gerd Petermann]

And this problem is not easy to solve by programs.

Up to now I don't know any efficient algo that allows to

find the nearest named road that goes parallel

to a selected unnamed way. It would be of great help

here.


Gerd



Von: Martin Koppenhoefer 
Gesendet: Dienstag, 24. November 2015 11:20
An: Tag discussion, strategy and related tools
Betreff: Re: [Tagging] Sidewalk Tagging for Routing


2015-11-24 10:40 GMT+01:00 Richard Fairhurst 
mailto:rich...@systemed.net>>:
Another issue with routing along pavements ("sidewalks") as separate ways is
the name tag. IME pavement-mappers rarely add the street name to the
pavement/sidewalk, but in fact the name applies to the pavement/sidewalk as
much as to the bicycle/car lanes. "Walk along unnamed footway then turn left
onto unnamed footway" is a less helpful direction than "Walk along Broad
Street then turn left onto Cornmarket Street". (The ref, on the other hand,
usually applies only to the bicycle/car lanes.)



very true. The same issue often occurs with cycleways (omitted names).

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Martin Koppenhoefer]

2015-11-24 10:40 GMT+01:00 Richard Fairhurst :

> Another issue with routing along pavements ("sidewalks") as separate ways
> is
> the name tag. IME pavement-mappers rarely add the street name to the
> pavement/sidewalk, but in fact the name applies to the pavement/sidewalk as
> much as to the bicycle/car lanes. "Walk along unnamed footway then turn
> left
> onto unnamed footway" is a less helpful direction than "Walk along Broad
> Street then turn left onto Cornmarket Street". (The ref, on the other hand,
> usually applies only to the bicycle/car lanes.)
>



very true. The same issue often occurs with cycleways (omitted names).

Cheers,
Martin
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Richard Fairhurst]

John Willis wrote:
> Perhaps we can have a routing engine at will interpret 
> a sidewalk with residential road junctions as being 
> along a residential road and route for Jay Walking. 
> [...]
> I would rather the router always error on the side of 
> crosswalks

Jaywalking is a North American concept. Here in the UK you can walk wherever
you like, whether the road is residential, an A (primary/trunk) or B
(secondary) road or whatever - anywhere apart from a motorway or somewhere
else with an explicit prohibition. Please don't suggest that routers should
export this people-hostile custom to other countries!

http://www.vox.com/2015/1/15/7551873/jaywalking-history


Another issue with routing along pavements ("sidewalks") as separate ways is
the name tag. IME pavement-mappers rarely add the street name to the
pavement/sidewalk, but in fact the name applies to the pavement/sidewalk as
much as to the bicycle/car lanes. "Walk along unnamed footway then turn left
onto unnamed footway" is a less helpful direction than "Walk along Broad
Street then turn left onto Cornmarket Street". (The ref, on the other hand,
usually applies only to the bicycle/car lanes.)

Richard




--
View this message in context: 
http://gis.19327.n5.nabble.com/Sidewalk-Tagging-for-Routing-tp5860841p5860877.html
Sent from the Tagging mailing list archive at Nabble.com.

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

Re: [Tagging] Sidewalk Tagging for Routing

[Marc Gemis]

On Tue, Nov 24, 2015 at 9:38 AM, Gerd Petermann
 wrote:
> One of the problems that is not well solved is the switch
> between both tagging schemes. When I start to draw
> extra ways for the sidewalk, where do i stop this and how
> do I connect the sidewalk with the main road?

same problem exists for cycleways (cycleway=track on main road or
separate way with highway=cycleway)
In those cases I draw a virtual cyclepath from the node on the main
road to the parallel cyclepath.

Another problem I have is how to connect a cyclepath/sidewalk on a
T-junction. Do I add a short cyclepath, or do I connect the crossing
street with the cyclepath ?

m

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

Re: [Tagging] Sidewalk Tagging for Routing

[Gerd Petermann]

As a co-developer of mkgmap, a software that convert OSM data
to Garmin map format,  I've looked at this problem quite 
often now. I think routing itself is not the big problem here,
unless we talk about the additional resources needed for the 
additional ways and junctions. 

Both tagging schemes have pros and cons, none is making
it easy for the mapper when it comes to map complex crossings.

Still, I tend to think that the mapping of extra ways for the 
sidewalks is the solution that requires less training and 
less "abstraction capabilities". 
One of the problems that is not well solved is the switch
between both tagging schemes. When I start to draw
extra ways for the sidewalk, where do i stop this and how
do I connect the sidewalk with the main road?

Gerd



Von: Marc Gemis 
Gesendet: Dienstag, 24. November 2015 05:22
An: Tag discussion, strategy and related tools
Betreff: Re: [Tagging] Sidewalk Tagging for Routing

On Tue, Nov 24, 2015 at 1:14 AM, Bryan Housel  wrote:
> I don’t really buy the “mapping sidewalks as part of the road makes routing 
> easier" argument, but I don’t work on routing software so maybe there really 
> is something complicated about routing that I am unaware of.

The problem with separate ways, is that none of the current routers
will tell you that you have to cross the street to reach a house on
the other side of the road. Most likely, they will let you walk till
the next street crossing and let you walk back then. There are some
proposals to solve this problem, but no implementation so far AFAIK.

Try cycle routing on osm.org in an area with a lot of separate
cycleways (e.g. The Netherlands) and you'll see what I mean.

regards

m

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

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

Re: [Tagging] Sidewalk Tagging for Routing

[Paul Johnson]

On Tue, Nov 24, 2015 at 2:01 AM, Marc Gemis  wrote:

> And what about the obligation to walk on the left side of the road
> when there is no sidewalk ? Should a router take that into account and
> let you walk to the next road crossing to switch sides ? Would be
> pretty neat.
>

Not sure how feasible it is for routers to handle such cases, but it does
make me wonder how to tag a situation like this one, which has been in
place for 30+ years now.
https://www.google.com/maps/@45.4981432,-122.7872102,3a,26.1y,318.53h,81.24t/data=!3m6!1e1!3m4!1sgjNPr6k_Wvp7U43x1_-0oQ!2e0!7i13312!8i6656
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Marc Gemis]

In Belgium several primary/secondary roads go through town centers
(e.g. [1], [2]), where the maximum speed is reduced to 50 km/h (same
as for residential roads).  Most people will cross the street where it
is most suitable (especially near [2] where the road is even pretty
narrow), not necessarily at marked pedestrian crossings.
I guess Jaywalking is pretty common in Belgium.

And what about the obligation to walk on the left side of the road
when there is no sidewalk ? Should a router take that into account and
let you walk to the next road crossing to switch sides ? Would be
pretty neat.

[1] http://osm.org/go/0EpBawFQ?m=
[2] http://osm.org/go/0EjrPCeh?m=

On Tue, Nov 24, 2015 at 8:28 AM, John Willis  wrote:
>
>
>> On Nov 24, 2015, at 1:22 PM, Marc Gemis  wrote:
>>
>> The problem with separate ways, is that none of the current routers
>> will tell you that you have to cross the street to reach a house on
>> the other side of the road. Most likely, they will let you walk till
>> the next street crossing and let you walk back then.
>
> Perhaps we can have a routing engine at will interpret a sidewalk with 
> residential road junctions as being along a residential road and route for 
> Jay Walking.
>
> Otherwise, being routed to the nearest crosswalk to cross to the other side 
> the street sounds like "working as intended". Jay walking a secondary road 
> will get you killed (or at least I will slap you as You step off the curb in 
> front of me when I'm on my bike), and routing disabled people (where curb 
> cuts may be necessary) means a crosswalk is the only place where they should 
> be routed to cross.
>
> I would rather the router always error on the side of crosswalks, and let the 
> person decide to Jaywalk if they feel like it.
>
> Javbw.
>
>
>
>
> ___
> Tagging mailing list
> Tagging@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging

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

Re: [Tagging] Sidewalk Tagging for Routing

[Warin]

On 24/11/2015 6:28 PM, John Willis wrote:



On Nov 24, 2015, at 1:22 PM, Marc Gemis  wrote:

The problem with separate ways, is that none of the current routers
will tell you that you have to cross the street to reach a house on
the other side of the road. Most likely, they will let you walk till
the next street crossing and let you walk back then.

Perhaps we can have a routing engine at will interpret a sidewalk with 
residential road junctions as being along a residential road and route for Jay 
Walking.

Otherwise, being routed to the nearest crosswalk to cross to the other side the street 
sounds like "working as intended". Jay walking a secondary road will get you 
killed (or at least I will slap you as You step off the curb in front of me when I'm on 
my bike), and routing disabled people (where curb cuts may be necessary) means a 
crosswalk is the only place where they should be routed to cross.

I would rather the router always error on the side of crosswalks, and let the 
person decide to Jaywalk if they feel like it.




I believe it is legal to cross the road normal to its path at least in 
Australia and in the UK!

If you cross at some angle (other than normal) then it is 'Jay Walking'.

Going to the next council cross street in some parts of Australia may be a very 
long walk ... may be 150 miles in any direction.


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

Re: [Tagging] Sidewalk Tagging for Routing

[John Willis]

> On Nov 24, 2015, at 1:22 PM, Marc Gemis  wrote:
> 
> The problem with separate ways, is that none of the current routers
> will tell you that you have to cross the street to reach a house on
> the other side of the road. Most likely, they will let you walk till
> the next street crossing and let you walk back then.

Perhaps we can have a routing engine at will interpret a sidewalk with 
residential road junctions as being along a residential road and route for Jay 
Walking. 

Otherwise, being routed to the nearest crosswalk to cross to the other side the 
street sounds like "working as intended". Jay walking a secondary road will get 
you killed (or at least I will slap you as You step off the curb in front of me 
when I'm on my bike), and routing disabled people (where curb cuts may be 
necessary) means a crosswalk is the only place where they should be routed to 
cross. 

I would rather the router always error on the side of crosswalks, and let the 
person decide to Jaywalk if they feel like it. 

Javbw.




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

Re: [Tagging] Sidewalk Tagging for Routing

[Marc Gemis]

On Tue, Nov 24, 2015 at 1:14 AM, Bryan Housel  wrote:
> I don’t really buy the “mapping sidewalks as part of the road makes routing 
> easier" argument, but I don’t work on routing software so maybe there really 
> is something complicated about routing that I am unaware of.

The problem with separate ways, is that none of the current routers
will tell you that you have to cross the street to reach a house on
the other side of the road. Most likely, they will let you walk till
the next street crossing and let you walk back then. There are some
proposals to solve this problem, but no implementation so far AFAIK.

Try cycle routing on osm.org in an area with a lot of separate
cycleways (e.g. The Netherlands) and you'll see what I mean.

regards

m

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

Re: [Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

On Mon, Nov 23, 2015 at 4:14 PM, Bryan Housel  wrote:

> I don’t really buy the “mapping sidewalks as part of the road makes
> routing easier" argument, but I don’t work on routing software so maybe
> there really is something complicated about routing that I am unaware of.


I believe Mapbox is working on routing sidewalks. Maybe they can help out.

Clifford


-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Re: [Tagging] Sidewalk Tagging for Routing

[Bryan Housel]

Current preference seems to be map sidewalks as separate ways tagged as 
`highway=footway + footway=sidewalk`.

1. As you mentioned, you can use sidewalk-specific tags (slope, surface) 
without affecting the adjacent highway.
2. The sidewalk and the road really are separate features.
3. New users like to trace and they are going to map it like this anyway.
 
I don’t really buy the “mapping sidewalks as part of the road makes routing 
easier" argument, but I don’t work on routing software so maybe there really is 
something complicated about routing that I am unaware of.



> On Nov 23, 2015, at 6:03 PM, Clifford Snow  wrote:
> 
> I'm working with a local non-profit to tag sidewalks for wheelchair [1] 
> accessibility. Tagging sidewalks is usually done by either drawing in a new 
> way and often parallel to roads or tagging the highway with 
> sidewalk=left/right/both/none. Either of these produce acceptable results for 
> normal walking. My questions concern tagging the wheelchair slope. Tagging 
> slope when the sidewalk is tagged with highway=footway, footway=sidewalk is 
> pretty straight forward. Novices can be easily trained to tag sidewalks this 
> way. One of the issues with this arrangement is routing the other is it adds 
> to the number of features. 
> 
> The other method, tagging sidewalks as features of highway=* would appear to 
> make routing much easier. Adding in tags for the kerb (curb) slope start to 
> increase the complexity. A typical tag for slope is 
> sidewalk:right:sloped_curb:start. iD can handle this tag by manually entering 
> the tag, but it increases the complexity to teach new mappers. There is also 
> the problem when a city decides to save money by adding the slop the the 
> corner of the intersection. Instead of 8 cuts, they only have four. Taginfo 
> only has 175 slopes [2]  tagged. 
> 
> My goal is to provide a simple method to train novices to add sidewalks and 
> kerb ramps. 
> 
> I would be interested to hear your thoughts on how to approach this problem.
> 
> Clifford


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

[Tagging] Sidewalk Tagging for Routing

[Clifford Snow]

I'm working with a local non-profit to tag sidewalks for wheelchair [1]
accessibility. Tagging sidewalks is usually done by either drawing in a new
way and often parallel to roads or tagging the highway with
sidewalk=left/right/both/none. Either of these produce acceptable results
for normal walking. My questions concern tagging the wheelchair slope.
Tagging slope when the sidewalk is tagged with highway=footway,
footway=sidewalk is pretty straight forward. Novices can be easily trained
to tag sidewalks this way. One of the issues with this arrangement is
routing the other is it adds to the number of features.

The other method, tagging sidewalks as features of highway=* would appear
to make routing much easier. Adding in tags for the kerb (curb) slope start
to increase the complexity. A typical tag for slope
is sidewalk:right:sloped_curb:start. iD can handle this tag by manually
entering the tag, but it increases the complexity to teach new mappers.
There is also the problem when a city decides to save money by adding the
slop the the corner of the intersection. Instead of 8 cuts, they only have
four. Taginfo only has 175 slopes [2]  tagged.

My goal is to provide a simple method to train novices to add sidewalks and
kerb ramps.

I would be interested to hear your thoughts on how to approach this problem.

Clifford

[1] https://wiki.openstreetmap.org/wiki/Wheelchair_routing
[2]
http://taginfo.openstreetmap.org/keys/sidewalk%3Aright%3Asloped_curb%3Astart

-- 
@osm_seattle
osm_seattle.snowandsnow.us
OpenStreetMap: Maps with a human touch
___
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging