[OSM-talk] A new highway tagging scheme - thinking about tagging highways for cycling and walking as well

Martijn van Oosterhout kleptog at gmail.com
Mon Aug 27 22:24:21 BST 2007

Interesting post, got me to think :) Comments inline.

On 8/27/07, Peter Miller <peter.miller at itoworld.com> wrote:
> I am unclear how the cycleway tag helps with a link that is also a road. How
> is a router or a renderer going to be able to identify those roads that are
> better and less good for cycling? Nearly all roads 'can' be cycled on but a
> cycle map for a town and a cycle routing engine need more information than
> that. What I was saying was that we need a scale of useability for cycling
> and also for many other modes besides cars.

In my experience here in NL is that paths for cyclists really only
come in four levels.
1. No provision for cyclists (highway=*)
2. Lanes for cyclists (cycleway=lane)
3. Dedicated cyclepaths (highway=cycleway)
4. Pedestrians areas where bikes are permitted (highway=pedestrian,cycle=yes)

Within each category I can't think of any meaningful differences,
other than ofcourse the usability for cars. So the better the road for
cars, the less usable it is for bikes. Lanes improve the score and
ofcourse dedicated paths have no cars. Pedestrian areas are obviously
speed limited.

> To be clear I am saying that it is 'likely' that a national cycle route will
> be a good cycle route, but may not be. I cycled about 30 miles of a national
> cycle route today and covered primary, secondary, tertiary, unclassified
> roads and a track. The cycle-ability tagging is separate from the signing
> and also from administrative classification. In the absence of a
> cycle-ability tag on a link then the fact that it is a signed cycle routes
> (national, regional, local) may be used to by a router to give the link
> preference.

Heh, possibly, though I find cycle routes tend to be fairly arbitrary
in their routing, not being significantly better than than other paths
in the area. They do tend to go through interesting areas you might
not eitherwise come.

(I did a quick search and it looks like cycling routes in UK tend to
focus more on quiet roads than routes in NL, so it might indeed be a
good metric for there).

> I would be interested in your thoughts on how to provide information for
> cyclists and other users on the main road network.
> Two examples; Firstly there is a sandy track some woods near here that 'can'
> be cycled on, walked on and ridden on and even, I believe by driven on (it
> is a by-way).

Ok, for me it's clearly a highway=track, because it's primary feature
is that it is unsealed/unmaintained. A kind of use-at-your-own-risk

> As a horserider I guess it would be the best. Wide, soft with good
> visibility
> As a walker it is fine (and is on a regional walking route), but not too
> good with a buggy.
> As a cyclist it is really hard going but possible.
> For a vehicle it would need to be 4x4
> As a wheelchair user it would be impossible.

Hmm, all that is for me implied by the highway=track tag, though I can
imagine other people might view that differently. Unsealed roads are
fairly similar on the whole.

Contrast with a highway=footway, which would be hardened in some way,
thus suitable for the buggy and the bike, and no cars. A hardened road
where cars are permitted would make it unclassified or service for me,
even if it is unsealed. Buggies/bikes/cars/wheelchairs all have wheels
in common. People and horses have legs: that's the real difference.

> Secondly; consider two normal paved roads with pavements; one of which is
> busy, noisy and the pavements are narrow and often obstructed with bins etc.
> The other is quieter and the pavements are more useable. What tagging
> information can be used to allow a rendered and router to distinguish these
> from each other for a walker?

I don't have a good answer to this except possibly something like the
cycleway tag, ie, a footway tag. footway=none/narrow/wide or something
like that.

Have a nice day,
Martijn van Oosterhout <kleptog at gmail.com> http://svana.org/kleptog/

More information about the talk mailing list