Closed ssotoen closed 2 years ago
Any highway with either
lanes:forward>=3
,lanes:backward>=3
, orlanes>=3 AND oneway=yes
.
And ending with split into more than one road. To avoid asking for example on https://www.openstreetmap.org/way/128209964 or https://www.openstreetmap.org/way/24821501
It would not allow to fill info fully, as road may be split after turn lanes start, not sure whatever it would confusing or noticeably lowering quality of collected data.
Good point, I didn't think of that.
Is it possible to filter elements according to whether they border on a highway where turn lanes are already deposited?
Depends on your definition of "border", but generally, yes.
Read through this quest suggestion. There are the following issues:
The biggest concern I have is with the splitting of ways. Will users really follow the instructions to always first split up the ways (and then return later days after to supply the actual information)?
I think, the splitting problem is more about the "How many lanes"-quest. The "Turn Lanes"-quest would be only asked if the number of lanes is already deposited. Besides, the question would refer to a way and not to a note. Therefore I think it should not come so easily to confusion.
Okay, another problem:
I think, there could be an ability to add a lane, like this. But you will still often have to split ways.
Seems like it's a good idea to tag turn lanes where they're implicit as well. Computers are really bad at sorting out implicit intent and need it spelled out explicitly.
@BalooUriza I don't understand what you mean
@BalooUriza I don't understand what you mean
Basically that this, with the arrows explictly marked,
and this, with no arrows,
both should have the same left|through|through;right
value.
both should have the same left|through|through;right value.
Of course, only in cases where left lane is for left turn only, center for forward only and right for right turn and forward (sometimes it is different).
both should have the same left|through|through;right value.
Of course, only in cases where left lane is for left turn only, center for forward only and right for right turn and forward (sometimes it is different).
Sure! My point is, however, specifying that it must be marked on the ground is unnecessarily stringent in this case. Just having a working knowledge of how the lanes work should be sufficient.
Any highway with either
lanes:forward>=3
,lanes:backward>=3
, orlanes>=3 AND oneway=yes
.
This would be overly restrictive in the UK, I'd hope it's possible to use some heuristics to deal with >= 2 lanes (e.g. coming onto a roundabout or a junction with a major road).
And ending with split into more than one road.
Depending on how it's been mapped, isn't @westnordost 's orange/green T junction a split into just one road, but could potentially still be eligible? Or do you mean excluding a splice of == and == @matkoniecz ?
For the orange way, it is not clear for which intersection the user should answer: The one with blue or the one with green. So, ways like these would need to be excluded (extra effort)
In the UK at least, the turn lanes are only marked on the incoming side of the junction, so you'd be tagging for green on the W to E direction of orange and for blue/red on the E to W direction of orange.
both should have the same left|through|through;right value.
Of course, only in cases where left lane is for left turn only, center for forward only and right for right turn and forward (sometimes it is different).
Sure! My point is, however, specifying that it must be marked on the ground is unnecessarily stringent in this case. Just having a working knowledge of how the lanes work should be sufficient.
Yeah although it depends what's on the other side of the junction and where most traffic goes. If the major road is actually turning to the right or left in your photo, with a minor road continuing straight on, then there may technically be no turn lanes, or it might be through|through|through;right for example.
Hmm, I guess all has been said what should be said about it. Generally, recording this information fulfills all the criteria, with strings attached, see the concerns uttered in my earlier comment.
I'll probably not implement this anytime soon as it is both quite the effort and has some problems but others are free to give it a try.
On Thu, Apr 8, 2021 at 6:05 AM Daniel Krause @.***> wrote:
I think this quest can only be asked safely for roads with cycleway=no or with cycleway=track only. Otherwise, situations like this https://wiki.openstreetmap.org/wiki/Lanes#Crossing_with_a_designated_lane_for_bicycles will cause problems.
https://camo.githubusercontent.com/8e735651e12b6c53bcde855fc96bcb92a140e2cd0d195d946cef78bd54eccb38/68747470733a2f2f77696b692e6f70656e7374726565746d61702e6f72672f772f696d616765732f632f63632f4b7265757a756e675f6d69745f526164737075722e6a7067 The road is tagged lanes=3, but there are four fields for turn:lanes.
Some more example photos with matching tags can be found here https://wiki.openstreetmap.org/wiki/Berlin/Verkehrswende/Radwege#Radfahrstreifen_in_Mittellage_.2F_Fahrradschleuse_.2F_Fahrradweichen (de-DE).
I keep pointing out to proponents of the status quo that this is a serious problem that would be most expediently fixed by simply not excluding bicycle lanes in lanes:= values, especially since like any other reserved lane, there's no safe assumption as to where it's placed on the road or that there's only one each way. I recommend pushing this on the various lists as well.
that would be most expediently fixed by simply not excluding bicycle lanes in lanes:= values
Redefining widely used tags is a bad idea and something that will be definitely not done by StreetComplete on its own.
On Thu, Apr 8, 2021 at 9:33 AM Mateusz Konieczny @.***> wrote:
that would be most expediently fixed by simply not excluding bicycle lanes in lanes:= values
Redefining widely used tags is a bad idea and something that will be definitely not done by StreetComplete on its own.
Was not suggesting SC do that. However, it probably would be a good idea to not prompt for lanes where cycleway=lane is true.
So, ways like these would need to be excluded (extra effort)
Can't SC automatically split ways at intersections? @westnordost
This auto-splitting could also be considered for ways that are too long for quests to display on them (don't know where the max value is set).
No. Not automatically.
On June 19, 2021 12:29:08 PM GMT+02:00, Atrate @.***> wrote:
So, ways like these would need to be excluded (extra effort)
Can't SC automatically split ways at intersections? @.***
This auto-splitting could also be considered for ways that are too long for quests to display on them (don't know where the max value is set).
-- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: https://github.com/streetcomplete/StreetComplete/issues/2574#issuecomment-864387669
-- Diese Nachricht wurde von meinem Android-GerΓ€t mit K-9 Mail gesendet.
It is one of quests that I had on TODO list for potential implementation but sadly I have no idea how to handle necessary splitting in way that would be understandable for mappers.
Yeah, difficult. Maybe it would be possible like this:
Though I am not sure if that would be easy and straightforward enough for SC users.
Another approach would be to only ask for street sections that lead to an intersection with an odd number of lanes, e.g. 3 forward and 2 backward lanes.
In general, it may be more effective if there was a dedicated editor that also displays road lanes as WYSIWYG or if such feature would be integrated into an editor like iD where you see the satellite picture.
Two great talks about lanes, their interpretation etc.: https://files.fossgis.de/Konferenz/OSM-Event/highway_1020_osm2lanes.intro.mp4 https://files.fossgis.de/Konferenz/OSM-Event/highway_1120_Micromapping_OSM_data_model.intro.mp4
Another approach would be to only ask for street sections that lead to an intersection with an odd number of lanes, e.g. 3 forward and 2 backward lanes.
Even in such cases splitting would be sometimes needed
...and limits the number of allowed splits to 1.
Limiting is not correct. The street could be original only one osm way (orange) accrossing the street with only the yellow part with the turns:
So it seems the overall complexity - both in implementation and for the user - of this is not worth it even it may in theory be done.
Also, another issue that came to my mind is that a quest always only refers to one element. So even if the quest would select only those roads that end in one intersection (see previous post), there may be the situation where it starts and ends in an intersection. The quest that doesn't know for which end to ask the question.
It is a tiny bit more feasible as an overlay.
Still, my enthusiasm is limited.
On Sun, Jul 17, 2022 at 12:03 PM Mateusz Konieczny @.***> wrote:
It is a tiny bit more feasible as an overlay.
Still, my enthusiasm is limited.
For what it's worth, I'm kind of of the opinion that lanes doesn't need to be a streetcomplete quest at all, mostly because no survey on the ground is typically required except where the pavement markings aren't obvious from aerial imagery, which isn't common.
General
Affected tag(s) to be modified/added: turn:lanes Question asked: Are there any turn markings on this piece of road?
Checklist
Checklist for quest suggestions (see guidelines):
Ideas for implementation
Element selection: Any highway that ends in more than one road and has either
lanes:forward>=3
,lanes:backward>=3
, orlanes>=3 AND oneway=yes
. To prevent spam this quest should only be shown for ways with at least 3 (or even 4) lanes in one direction. The quest should only ask for one direction at a time, even if both directions meet the lane criteria.Proposed GUI: Similar to the cycle lane quest:
Possible Values:
These are pretty much just copied from the wiki. I excluded sharp_left and sharp_right because they're barely used.