[OSM-talk-ie] Traffic Islands, pedestrian areas and footways for routing purposes

Conor movinghouse at gmail.com
Tue Dec 8 15:04:56 UTC 2015


Hi all,

Colm and myself raised an issue relating to pedestrianised areas and
footways a month ago and there was no response at the time. This
issues are ongoing and I'm wondering if anyone has an opinion a month
later? I'll include Colm's and my own original messages below. Thanks

=========

Hi,
I'm wondering what the best way is to deal with traffic islands - the
parts of roads that aren't roadway / carriageway and aren't lateral
footways / footpaths. Sometimes they are unmapped, sometimes they are
mapped as pedestrian areas and/or pedestrian ways .
On the road, they generally come in one of three fashions:1. No
pedestrian access / no meaningful pedestrian use, e.g. Dorset Street
Lower in Dublin, where  there are trees every 10 meters that block the
way. Not mapped, other than as two separate roadways. Should these be
mapped at all?
https://www.google.ie/maps/@53.3599777,-6.261141,3a,75y,44.35h,85.42t/data=!3m7!1e1!3m5!1s0YTaJATDMDakjhiLjRa__w!2e0!6s%2F%2Fgeo2.ggpht.com%2Fcbk%3Fpanoid%3D0YTaJATDMDakjhiLjRa__w%26output%3Dthumbnail%26cb_client%3Dmaps_sv.tactile.gps%26thumb%3D2%26w%3D203%26h%3D100%26yaw%3D219.34166%26pitch%3D0!7i13312!8i6656
2. Those where pedestrian access is an important part of their use,
e.g. at the central median on O'Connell Bridge in Dublin. Mapped as
pedestrian area linked to pedestrian ways at the south end, but not
the north end, effectively making it a cul de sac. I'm not sure how
route planners treat areas.
https://www.google.ie/maps/@53.3470972,-6.2591266,3a,75y,90t/data=!3m7!1e1!3m5!1s7ed2ZerYJnbZnrW-0cHENQ!2e0!6s%2F%2Fgeo2.ggpht.com%2Fcbk%3Fpanoid%3D7ed2ZerYJnbZnrW-0cHENQ%26output%3Dthumbnail%26cb_client%3Dmaps_sv.tactile.gps%26thumb%3D2%26w%3D203%26h%3D100%26yaw%3D94.366425%26pitch%3D0!7i13312!8i6656
3. Traffic islands at signaled junctions generally, e.g. at the
junction of O'Connell Bridge / D'Olier Street / Westmoreland Street.
Mapped as pedestrian areas and pedestrian ways.
Thank you
Colm

=========

Hi Colm,

Perhaps you're the same Colm that I was in touch with on a similar
subject a few days ago? If so, we had discussed raising this on the
mailing list so either way this would be as good a time to add some
further thoughts to the subject if that's ok. Apologies for the long
read...

The aspects you drew attention here to are really interesting as they
highlight how differently all contributors to OSM can see the value,
use and appeal of the map. I myself am more biased towards visual
reading of maps (yes, that old fashioned way) and sometimes see people
question the value of mapping something where I see no question at
all. All valid questions though, of course. In my attempts at
recommending OSM to various friends and work colleagues for actual map
usage most try and use it for visual reading of a map (they generally
find it doesn't work as well as Google for directions), whereas any
friends who would be more technically inclined often find the routing,
tools and contributing more appealing.

I've come across a potential clash of solutions for drawing and
tagging for routing applications vs drawing and tagging for visual
applications on OSM. As mentioned above, people contributing to the
map can be attracted to it for varying reasons and I caught a pattern
occuring in lots of areas I was contributing to. Seeing that all
applications and uses should be catered for as best as possible,
myself and VictorIE (the same Colm?) got talking about our difference
of opinions on the matter.

The issue I raised revolves around the use of short, sporadic footways
that appear in locations where there isn't a designated footway. In
terms of visual mapping I am not a fan. I've seen them appear as
multiple desire lines across fields or as someone's preference for
where they may cross a traffic island on a busy road where there is no
markings or crossing in real life. In this instance, an almost
infinite combination of footways could be drawn across the traffic
islands/fields and, although there are certainly exceptions where the
footway aids in suggesting where to walk, it seems to benefit routing
applications to the detrement of visual applications.

Footways drawn in urban areas can be potentially confusing at
communicating visually. I've seen understandable additions where
they've been added in locations where the footway area is wide to give
some visual sense to the void of space on the map between roads (e.g.
some may read a large void between roads as potentially containing
buildings/fields). In my opinion, the presence of a short footway on
the map can suggest that there's something preferential to it over all
the adjoining ones I assume are there in real life, or that there's a
lack of others. Where the footway is wide, the limitation of an OSM
way being uniformly thin can be hard to decipher when trying to be
confident you're in the right location from a visual map reading.
Pedestrian way areas solve most of the visual problem but don't help
routing where it's needed with most routing applications, so a
combination can be useful.

Conversely, I've seen urban projects on OSM where a large collection
of footways have been added to aid pedestrians. The sheer abundance
removed some of the doubt I just mentioned though this is a
controversial approach in that it can look incredibly messy if care is
not given to the amount of ways running parallel. It also a rather
luxurious addition that can take low priority given that most urban
highways are assumed to have some form of footway neighbouring them.

VictorIE correctly pointed out that OSM is a connection of joined up
roads. A dominant use of OSM is for driving and the streets are the
primary feature (the clue's in the name). It's essential that this use
is adhered to for applications that provide directions and similar
services. His own reason for the sporadic footways I'd seen in areas I
was working on were to ensure roads were joined correctly for these
services. The technicalities of this are not my forté but I would have
thought there were other means to link roads for routing purposes. I
also assumed that a routing application would not realise that a
highway joining another highway was traversable only because a footway
linked the two. (Amendment: Although OSM is a connection of joined up
roads, I'd like to question that principle as it grows. For example,
in rural parts of the country ther are many former donkey tracks that
are invaluable for hikers. Many of these have overgrown in parts to
the point that they do not exist in any form in sections. The sections
that do remain are increidbly useful on a map but would be misleading
and potentially dangerous if they are connected on OSM.)

I originally discussed the issue with VictorIE after a footway added
to a traffic island struck me as a particularly dangerous spot to
cross at. There were no markings or designated footway in real life so
I removed it out of worry. After he reinstated it I got in touch to
figure this where he explained his reasons relating to routing. There
was also a suggestion that doing so helped unclutter diagnostic tools.
Again, in my opinion, I would give preference to solving problems with
day to day usage of the map rather than solving problems on diagnostic
tools. The latter should (and mostly does) aid in solving the first
but should not be the target themselves if not confident it aids all
day to day usage.

Examples: http://www.openstreetmap.org/?mlat=53.38877&mlon=-6.07409#map=19/53.38877/-6.07409
There are now two footways extending northwards off the looping
driveway in front of the train station, to connect it to the footway
running parellel against the station entrance. For routing, I see few
issues as there's no wall blocking where they're drawn. Visually, they
suggest to me that there is a designated way to walk here where there
isn't. The left-most one is typically blocked by cars as it's a
parking space. The right-most one is a dangerous spot to cross as it's
a busy, wide road. There's a designated crossing-space further west
from it that wasn't marked on the map.

http://www.openstreetmap.org/?mlat=53.37103&mlon=-6.20664#map=19/53.37103/-6.20664
As mentioned earlier. There is no designated pedestrian area where the
footway's drawn. The traffic island is occasionally used by
pedestrians but the way drawn is (in my opinion, I live nearby) a
lethal part to cross as cars swing around the blind corner quite fast
sometimes. Most people cross further up Collins Avenue (where there's
still no designated crossing) when they've experienced a few near
misses and often avoid the traffic island.

I cannot see the sense in OSM working in such a way that this solves a
routing issue at the detriment of visually reading the map (in any
renderer).

By the way, the fact that these little footways keep popping up just
after I add new roads clearly indicates I'm doing something that is
annoying diagnostic tools. I get no validation errors and the nodes
are joined correctly. What should I be doing in terms of routing
validation to stop annoying those who take the time to check for these
errors on diagnostic tools (amendment: or are this particular error
redundant as it's encouraging visual mapping difficulties?)


Thanks,

Conor



More information about the Talk-ie mailing list