For what it’s worth, I don’t think it’s necessary to record every detail about signal phasing. But without a way to record protected turns and a consensus on turn-on-red restrictions, this feels like the tail wagging the dog. The result is that shared_green=yes and shared_green=no end up being required in so many cases based solely on intersection geometry, yet a data consumer won’t be able to make strong assumptions based on either value.
As far as I know, protected turns must be verifiable: either the vehicular signal has an arrow lens instead of a ball lens, or the intersection has a separate signal with a sign saying it’s a dedicated left turn signal. Otherwise, the motorist has no way to know that oncoming traffic could stop for them.
Conversely, some jurisdictions use a flashing yellow phase or put up signs to remind motorists to yield to oncoming traffic, which affirms that the turn is unprotected, though this isn’t universal.
Yeah, I also have difficulty to find crossings with shared_green=no. Even the examples I know are mostly trivial (i.e. following the rule that through traffic doesn’t share green with crossing pedestrians and for one reason or another, no right turn into it is possible).
Only if pedestrians are the tail and vehicles are the dog . This proposal is aiming only on pedestrian crossing safety / potential pedestrian crossing hazards, while the specifics you are interested in, turn on red and protected turns, are generally referring to vehicle traffic.
If vehicles can turn on red at the same time as pedestrians can cross, the value for shared_green would be yes (or some other kind of more specific yes), although literally it would be a combined “red” and “go” phase for those turning right on red.
turn on red is almost non-existent where I map (Italy and Germany), so it is not something I ever missed, while protected turns are frequent, but not relevant for pedestrian safety (or are they)?
Sure, from the above poll with 9 respondents, it seems I am the only one to support the tag, so it doesn’t look very probable that I will continue pursuing it
Protected turns wouldn’t be relevant to pedestrian routing, except that this aspect affects the central assumption behind the proposed key. If there is a protected turn, then you’re right, the pedestrian is safe from that direction at least. But my point was that we currently don’t say whether or not there’s a protected turn. For all the data consumer knows, there’s an unprotected turn that makes the pedestrian quite unsafe, even if there’s no turn on red. (Turning on red is the norm across the U.S. and Canada. Restrictions are common in many central business districts, but there are always signs to that effect.)
In the UK. A typical four-way junction (at least in the areas I’m familiar with) will have three phases, one for north-south traffic, one for east-west traffic, one for pedestrians.
. This proposal is aiming only on pedestrian crossing safety / potential pedestrian crossing hazards, while the specifics you are interested in, turn on red and protected turns, are generally referring to vehicle traffic.