|
Hey, I got a question regarding this turn restriction. Did I map it correctly? Which kind of turn restriction do I have to choose in a case like that? Is the turn restriction type based on the from way or on the via way? In this case it is kind of a u-Turn, but it is mapped as separate ways. So you are allowed to turn left, but not again left. //edit: I try to explain it more in detail: these to routes are allowed:
this one is not allowed:
in reality there is a sign that forbids u-turn. so logically it is a u-turn restriction, but does a routing software counts turning left twice (like the ways are orientated on OSM) as u-turn or turnleft restriction type? Thanks |
|
I would tag it as no_u_turn but otherwise it is correctly mapped. Turn restrictions type should be based on the from member. there a no_u_turn AND a no_left_turn relation. OSRM is still routing about this way, see http://osrm.at/aE1 so something seems not to be correct.
(15 Jan '15, 11:10)
wiedergaenger
|
|
Why not no_u_turn? If you are simply not allowed to turn left from what is now the via way, then a no_left_turn, but simply for current via to current to would be enough. Can any software detect that u-turn means turning left twice? You are allowed to go left, but not in case you're going to turn left again right afterwards.
(14 Jan '15, 19:12)
wiedergaenger
I guess that doesn't really matter. The software sees that exactly this route is not allowed. The classification (no_u_turn, no_left_turn etc.) instead is mostly helpful for displaying icons for the driver.
(14 Jan '15, 19:16)
scai ♦
okay so technically it doesn't matter which either positive or negative classification im giving, the routing should be correct anyways?
(14 Jan '15, 19:21)
wiedergaenger
It does of course matter whether the classification is no_... or only_.... And this is just my personal point of view, in the end every application will have its own implementation. So always use the correct turn restriction category.
(14 Jan '15, 20:24)
scai ♦
2
I asked this question and the using OSRM may be a way to test the no uturn https://help.openstreetmap.org/questions/32420/turn-restrictions-and-routing-engines
(14 Jan '15, 21:16)
andy mackey
Oh I see there a no_u_turn AND a no_left_turn relation. OSRM is still routing about this way, see http://osrm.at/aE1 The no_u_turn is about a year old so out of date OSRM data might not be the reason
(15 Jan '15, 10:43)
wiedergaenger
showing 5 of 6
show 1 more comments
|
