How do we tag continuous footways? This is where the sidewalk to a main road continues across a side road, so cars have to drive over the sidewalk to access the main road from the side road and vice versa.
I am wondering how to tag the crossing node.
Looking at the documented tags, the closest ones seem to be crossing=unmarked or maybe crossing:markings=surface (in addition to traffic_calming=table kerb=no tactile_paving=no etc.), but neither tag is really appropriate. In a way, a continuous footway is the inverse of an unmarked crossing. In a typical unmarked crossing, there are no markings on the road to indicate that pedestrians might be crossing it. In this case, there are no markings on the sidewalk to say that cars might be crossing it.
There has been some discussion on the tagging mailing list ([Tagging] Continuous Sidewalk or Cycleway) with some interesting comments and a Wiki page was created (Continuous Sidewalk - OpenStreetMap Wiki) but there was disagreement about the proposed tag and the discussion didn’t really reach a conclusion. (Taginfo gives 425 uses for continuous_sidewalk=yes)
Of course sidewalks very often continue uninterrupted across driveways and there has also been some discussion about whether this should be tagged with highway=crossing at all. In this case, where there is quite a bit of traffic on the side road and potential for conflict between drivers and pedestrians, I would like a tag, if only to indicate that it has been surveyed. One option that came up in the linked discussion is the tag crossing=pavement which has been used 1,042 times but is undocumented. Is this the tag I am looking for?
In cities where such crossings are common, how are they usually tagged?
Of course sidewalks very often continue uninterrupted across driveways and there has also been some discussion about whether this should be tagged with highway=crossing at all. In this case, where there is quite a bit of traffic on the side road and potential for conflict between drivers and pedestrians, I would like a tag, if only to indicate that it has been surveyed
I believe highway=crossing is a correct tag here, although technically the street crosses a sidewalk, from the pedestrian point of view this is still a crossing. I guess we should introduce a new tag, assuming the right of way is by the pedestrians, as crossing=unmarked has IMHO the opposite presumption.
I use continuous_sidewalk=yes for these, because it’s technically not a crossing (because crossing in OSM means footway crossing the carriageway). If they are not continuous, and just raised, traffic_calming=table + crossing=unmarked fits.
I used to tag them as crossing=uncontrolled + crossing_ref=continuous_sidewalk, but I deem this wrong now
No, continuous sidewalk means that the vehicles have to drive over the sidewalk, not the pedestrians are crossing the carriageway in any way. That’s why the pedestrians have the right of way on these continuous sidewalks. At least if it’s like this:
Arguing over whether the footway crosses the carriageway or vice versa is fairly academic and not useful. A crossing in OSM is place where one type of traffic crosses another type of traffic. That is the most important thing to record. Which type of traffic has the right of way would also be quite useful to record although it seems we don’t have a way to tag this yet. The physical characteristics of the crossing may also be useful to some data consumers, but I would say this is a secondary concern. One type of traffic crosses another regardless of the physical characteristics.
There is no street crossing; the pedestrians don’t cross anything, it’s the other way around.
Continuous sidewalks can have cycleways on them, so same transportation mode
Legally, that’s not the case over here, which is why the vehicles on the carraigeway don’t have the right of way. Would you say that a driveway (from a garage or house) going over a footway is also both a carriageway and a footway? Same situation legally. At least over here.
The reason people want to tag this special, is because it’s a completely different legal situation if vehicles have to go over the sidewalk compared to pedestrians crossing a street. It’s even different from zebra crossings, because on these here, everyone allowed to use the sidewalk has the right of way, which is why cycleways are increasingly built using these.
As mentioned before, I used to tag them with a special crossing type, but stopped using it. If there’s a better suggestion than continuous_sidewalk=yes, bring it forward. I’ll use anything the routers could understand, and the renderers can render accordingly. I’m not opposing any new ideas, just saying that the OSM definition of crossing isn’t met for these.
The pedestrians definitely cross the street, just as the vehicles in the street cross the footway. They cross each other. As I said above it would be good to develop a way to specify that pedestrians have the right of way and that vehicles have to wait for them. This doesn’t mean it’s not a crossing and instead some fundamentally different kind of thing.
In Poland continuous footway without pedestrian crossing markings is giving no priority rights to pedestrians, while they have right of way on really badly designed crossing with legal markings
footway and road cross each other here (maybe from legal perspective it is different but it is still crossing of road and footway)
and at following location service road and combined cycleway-footway cross each other:
I would prefer to tag what it is (a continuous sidewalk) instead of who has the right of way, because depending on the local legislation, this will vary and could get very complicated. Don’t get me wrong, of course we need to figure out the implication of this construct, but it’s easier for people to tag “this is X” instead of “this means X, Y and Z over here”.
If you really want to tag this as a highway=crossing, what about crossing=contiuous_sidewalk or crossing=sidewalk then?
Or would you prefer crossing=unmarked + priority=footway or something similar?
This is an interesting point, in some cases I’ve seen it would be much more accurate for a renderer to display one highway=* over another. This would allow for surface and width tags for example to apply more accurately, which in some cases could affect a large geographic area.
In my city, I sometimes think about this when I map alleys. Some alleys are paved with asphalt and bleed directly into intersecting residential roads. In those cases it is clear that sidewalks “cross” the alley, and I often add a (un)marked crossing way to represent that. In other cases, the alley very much “crosses” the sidewalk, as the sidewalk is continuously built across the alley. It is still an (un)marked crossing, but this time it’s the cars that would be crossing the sidewalk. I wish I could add a crossing way to the alley that intersects the sidewalk in this case.
I hope my above coincides with your premise @Nadjita. If your local laws apply differently to “continuous sidewalks,” that would be an even stronger argument for a better tagging scheme.
Arguing over whether the footway crosses the carriageway or vice versa is fairly academic and not useful.
I strongly disagree. In OSM we represent footway crossings as ways very frequently, which cannot easily be applied to this situation. I think this is a very useful discussion to have, even if the outcome is “let’s keep it how it is.” As you say, the physical characteristics of a crossing can be very useful data.
“Continuous sidewalk” and “crossing” are not mutually exclusive terms. The things were are talking about are both. I would suggest tagging as crossing:markings=surface to indicate a different surface material than the street and then something like crossing:height=raised to indicate that pedestrians don’t step down into the street, but vehicles drive up and over the crossing instead.
I would avoid continuous_sidewalk as a key or value because the same kind of crossing can happen where an independent footpath (not a sidewalk) crosses a street.
It seems to me that continuous sidewalks represent both a geometric (how ways intersect) issue as well as a potential legal one. Perhaps neither of these can be suitably tagged using a node, as OP asked.
This table in the wiki seems to be highly relevant to our discussion. It is clear that crossing nodes would be tagged similarly for either footways crossing roads or vice versa. When I first read this discussion I immediately started thinking about crossing ways, which have the potential to but do not yet represent these differences, as the table linked above illustrates with big question marks.
It may be academic, but we’re at this point partly because people started pedantically mapping “crossings” at every point where a sidewalk meets a driveway, which made the relevant tags much less useful. We already have data model support for such a broad definition of crossing: two ways can share a node.
Continuous footways are interesting because they take all the assumptions behind crossing tagging and turn them on their head. We didn’t account for that when we made crossing a value of highway=*. It’s almost as if there should be a highway=highway_crossing tag to be applied to nodes.
You know, that’s actually not the worst idea. It sounds horrible, but it does make a lot of sense, because then, things like kerb=* would be applied to the street (not way. God, OSM is so sc***ed up with steps being highways…)
I’d be interested to hear what folks think about this location in my town where a pedestrian street intersects with a vehicle street. Pedestrians have priority and vehicles must stop for them. People routinely walk straight across with very little regard to the presence of vehicles.
The vibe on the ground is certainly that the vehicle street crosses the pedestrian street because of the surface material difference, but really both streets cross each other. There is not grade difference between the streets. There is a highway=crossing node here, but no way tagged as a crossing. If there was one, footway=crossing wouldn’t be right since the street is a highway=pedestrian. I guess pedestrian=crossing has a little bit of use. I have mapped the surface change from asphalt to brick on the cross street: Way: Bank Street (1131464701) | OpenStreetMap. If there were a “vehicular street crosses pedestrian way” tagging scheme, this would be a perfect place to use it. It certainly wouldn’t be called a continuous sidewalk though.
To me, that doesn’t look like a vehicle street being the one crossing, because I can still see a (flush) kerb for the pedestrians. But the same seems to be true for the non-pedestrian road, so I’m not sure what would apply in your legislation. In Germany, I would tag this as traffic_calming=table because of the kerb for the pedestrians, but I’m sure there’s people that would argue the other way around
Well … couldn’t something like residential=crossing / pedestrian=crossing on the node itself make it clear what is crossing what? Just a thought.
