I used to be opposed to sidewalk mapping, and I still think it is often
done poorly. I've changed my mind in the last year or two though. When I
first moved into my current neighborhood and started mapping the area, I
hated at all the poorly drawn sidewalks. They weren't well aligned, they
didn't do anything to indicate crossings, and they were far from
complete. For a while I was temped to delete the lot of them, but
instead worked to gradually fix them up, noted marked or signalized
crossings, added in traffic islands, pedestrian barriers etc.
Once you have a high-quality, relatively complete mapping of sidewalks,
I really think they add a lot of value. You can see where sidewalks end,
where crossings are absent, how long crossings are, whether there is
separation from other traffic by e.g. fence or bollards.
It's not just about routing. Sidewalks (and crossings) are
infrastructure in their own right and deserve to be mapped as such, at
least in many dense urban areas, and especially where they vary
significantly from street to street. I'm not saying it should be done
everywhere, but it definitely does have value in some places.
Best,
Nate Wessel, PhD
Planner, Cartographer, Transport Nerd
NateWessel.com <https://www.natewessel.com>
On 2020-04-03 2:49 p.m., Frederik Ramm wrote:
Hi,
On 4/3/20 19:45, Martin Chalifoux via Talk-ca wrote:
This morning I checked some large cities namely New-York, Paris, Amsterdam,
London, Berlin. Since OSM is best developed in Europe these capitals make
sense. I just checked Tokyo, Shangai, Seoul, Sydney to sample Asia. None of
them have this sidewalk mapping as separate ways.
There are pockets here and there in Europe as well. Mostly what happens
is this:
1. Someone wants to make a cool pedestrian/wheelchair/schoolkid routing
project
2. The person or team has limited programming capability or budget, and
hence must attack the problem with a standard routing engine
3. Standard routing engines do not have the capability to infer a
sidewalk network from appropriately tagged streets (i.e. even if the
street has a tag that indicates there's sidewalks left and right, the
routing engine will not generate individual edges and hence cannot do
something like "follow left side of X road here, then cross there, then
follow right side" or so
4. Hence, tons of sidewalks (and often also pseudo-ways across plazas)
are entered into OSM, to "make the routing work".
(5. often people will then find that the routing engine generates
instructions like "follow unnamed footway for 1 mile" which leads them
to copy the road's name onto the sidewalk geometry... to "make the
routing work").
(6. In some countries a pedestrian is allowed to cross a street
anywhere. Happily I haven't yet encountered people cris-crossing the
streets with footway connections to "make the routing work" in these
countries. If you're in a country where you are only allowed to cross at
marked crossings then that is easier.)
All this is a sad state of affairs; if we had routing engines that could
work well with simple "sidewalk" tags (and also make standard
assumptions about which road types in which countries would usually have
sidewalks even if not explicitly tagged), then we could save ourselves a
*lot* of separately mapped sidewalks that really do not add valuable
information, and just serve as crutches for routing engines.
Personally I am very much opposed to the separate mapping of sidewalks,
though I recognize that unless we have routing engines that work without
these crutches, I will have a hard time convincing people to stop doing
that.
Bye
Frederik
_______________________________________________
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca
