[DRAFT PROPOSAL] Should I be mapping a separate pavement or an integrated pavement? Guidance for UK mappers

I haven’t thought too hard about that. When I created crossings as nodes on highways I’d only tag them when there was identical provision on both sides, so directly opposite drop kerbs. If those are present the presence or absence of tactile paving tended to match too so I tagged the node with kerb and tactile paving tags.

If you have separate pavements, you could tag either end of the crossing with separate tags where the crossing joins the pavements, though might it cause confusion along the pavement?

If we’re not going to set a framework for reintegrating even the most frivolous examples of separated pavements then that’s annoying to me and I don’t think it’s going to fix the essential problem. But if this is the way the community wants to go then at least it’s progress.

Where another path joins the sidewalk from the side, continue this way across the sidewalk and connect it to the main road, unless there is a convenient crossing point close by that is mapped.

Thinking about it, this always needs to happen regardless, because otherwise a router ignoring footway=sidewalk tags would be stranded.

That’s a good point. But there are a couple of situations where I think you definitely wouldn’t want to do this:

  1. If the path joins literally a metre or two from an actual crossing - where it would seem silly to add an extra non-existent link path between the sidewalk and the road right next to the crossing.
  2. If there’s a barrier (e.g. a fence) stopping the path directly joining the carriageway at the point it joins the sidewalk - where you wouldn’t want to add a path that it’s not physically possible to use.

So maybe routers can’t always get away with just ignoring separately mapped sidewalks completely. Possibly they would need to be a bit cleverer, and add in some missing links. I’m not sure there is a good solution to this…

Sorry, should have been clearer. I meant the situation where a crossing connects one sidewalk to the next, following roughly the same path as the sidewalk, with little deviation. So like this footway=crossing links this footway=sidewalk to the next one.

Like you, I think they’re part of the sidewalk for the purposes of sidewalk:*=separate on the corresponding bit of roadway. People would understand them as part of a safe pedestrian corridor parallel to a road.

It would be nice for the guide to state that explicitly.

I get what you’re saying about Way: ‪Friars Entry‬ (‪1026966536‬) | OpenStreetMap. I think the natural continuation of that is as an informal crossing to the other pavement, so I’ve just sorted that out.

1 Like

Since in both instances these are ways created entirely for the router (ie they don’t exist at all) to enable ignoring sidewalk ways that seems fine to me?

I agree that it’s somewhat annoying. I consider it “clutter” when it’s not adding new information, but the bar for justifying deleting other people’s work is understandably high. I’ve tagged lots of street trees in my local area which others might consider clutter, but I’d be annoyed if someone deleted them.

2 Likes

Mapping trees isn’t negatively impacting existing walking routing, drawing unnamed sidewalks is

3 Likes

Especially, I should add, as I’ve expanded a lot of effort to improve the map on my area and consider it “my patch” but it absolutely doesn’t belong to me.

I even have an OSMCha watch on the area that notifies me when people make changes - that’s how I notice some of the annoying stuff.

I’m happy to go with your proposal (and Robert’s, I don’t think they conflict significantly, except Rob’s tolerates people who choose to map the suboptimal way as long as they do it to a reasonable standard) and ndrw6 has agreed to revisit his changes.

With most things in OSM there’s usually more than one way to do things.

And meanwhile the person who did the original drive by crap mapping has disappeared, probably to incorrectly map sidewalks along one block in lots of unlucky other parts of the world…

1 Like

If I’m honest this is something I don’t understand but I’m also fine with? If people want to add trees then fine :person_shrugging: I don’t walk on trees.

What gets me is that it worked with 1xdata - then 2xdata is added, and then the routes go up the side of trees :upside_down_face: and then you have 2-3x the cleanup to get almost back where we were :joy:

1 Like

I think one thing I really need to start doing is just referring crap pavement mapping, especially from taskers, to the DWG. In my specific instance it was done 2-3 years ago, and so the mappers were long gone by the time I had moved to Manchester and started to try learning my way around.

OSM was great at showing me around all the little nooks/crannies of previous towns/cities I moved to. Moving to Manchester really underlined how these separated sidewalks have degraded that (and hence motivated me to try and get it fixed).

3 Likes

Been doing some mapping of pavements in a residential area, roads with or without pavements and some with pavement on oneside. I normally use JOSM to edit and I have found paint styles(

Sidewalks and footways (with knobs on))

which show the edited pavement. Opened that area in ID and could not find a method to illustrate the edits. The direction of the way tags determines what is left and right - something to be advised to editors.

I found that where the pavement changes it is necessary to split the way so that the way can have the corrrect pavement tags applied. This could lead to very segmented ways - is that a difficulty for routers?

Some of the roads have verges, but I can’t find a paint style which illustrates them - any one know of on that illustrates them.

For viewing the end product I find that OpenStreetBrowser allows selecting footways and kerbs, and

Useful Maps also shows road with pavements (the thick edge to a road)

In this residential situation I think that adding attributes to roads is easier and conveys the information visually, that JOSM is my preferred tool, ID requires editor imagiination.

Previous experience in towns and cities tell me that separate footways are more useful where there are more pedestrian routes.

1 Like

Is this bridge a big enough break in permeability?

Agree with a lot of what you have said.

How could this be reasonably mapped with sidewalk tags on the road way? would the little raise section on the left have to be mapped separately a a strange loop away from the road?

I can see asphalt road… but a asphalt crossing, turned into raised paver, turned into bricks.

Mapping sidewalks as part of the road also destroys any useful data about pedestrian crossings and kerbs. I get that the original creator of this post Believes mapping sidewalks is futile because we are legally allowed to walk in the road in the UK…. But i still want a router to take me along a pavement where one is available.

I just don’t think it is possible to come up with fixed rules about when and where you cannot map physically connected pavements as separate, or part of the road way.

The above image, the current footways are mapped as the green. Both to the right of the image, and the top left the footway is separated with grass verges. But the top middle of the image, the pavement makes contact with the road (you don’t need to get grass on your boots to cross there) so should the pavement have two links into the road, and follow the orange route? You will lose data about the length of the crossings to the islands, the types of kerbs at the crossings, tactile paving information.

Why remove a 50 metre section of incorrectly mapped separate sidewalk, and glue it to the road, on an 800 metre long section of completely separate sidewalk?

Also like you say, tools for OSM contributors are…. lacking… for sidewalk information. ID has a great tool for adding bike lanes left and right side, why not adapt this to also include sidewalk tagging? And as you say with JOSM, the “with knobs” style is invaluable. It can really help debug little sections like this where some parts of an area will have separate sidewalks, others not.

In the below example a mix of roads that have a sidewalk attached, and sidewalk separate:

But when I am out walking, what route am i actually taking, the red or the orange route. A router would tell me to take the red one, I will actually walk the orange one:

1 Like

We’ve discussed two predominant methods of notating pedestrian routes. The standard OSM browser demonstrates pedestrian routing using OSRM, Valhalla and GraphHopper. In a simple test using Kipling Road and Bath Road only GraphHopper used the footways, the others directed down the road centrelines.

To me this inidcates that there is not a common specification for pedestrian routers. There needs to be a specification of what the router is to provide and what data it is to use. That last sentence is a very large piece of work for a team of product developers. When the data to be used is defined then the mapping needs can be defined as a mapping style - attached to the road or separate ways. Until a formal decision is made then both styles are valid. Each style has positive and negative impacts as you have illustrated.

It can easily be argued that finer detail can be attained with one method or another - but the router or map styler has to indicate to mappers what style their software will need or use.

2 Likes

Exactly. I think a lot of it is because routers are having to work with the current limited data, which is extremely limited for pedestrians, and assuming to route people down almost any residential or non high-way road.

ID doesn’t have a built in function for mapping sidewalks as part of a way, its something you have to know yourself, and not many renders will display this information. Even the that “useful maps” one is the only seems to check if sidewalk yes/separate exists. A road tagged with sidewalk:left=no and sidewalk:right=separate will still display as a sidewalk on both sides :frowning:

You can see why so many people prefer to map them as a separate way when everything is pointing people towards mapping like that, due to the limited creation and rendering of sideways as part of a way.

1 Like