|
I wonder why this routing along an esplanade (highway=pedestrian, area=yes) goes all the way around the edges of the area, while this routing along another esplanade (highway=pedestrian, area=yes) doesn't do so. What I'm really trying to achieve is to improve the first mentioned routing (in the worst case it routes a 5 m walk into a 1.82 km walk around the edges of the area), while sticking to best mapping practices (I'm aware of the difficulties associated with routing through open areas). |
|
My guess for an explanation: In the first case the pedestrian area is mapped as a polygon (highway=pedestrian + area=yes) and this is understood by the routers as a routable way. Since they cannot route over an area they take the edges instead as an approximation. In the second case the pedestrian area is mapped as a multipolygon and the routers don't accept that as a routable way and completely ignore it. Instead they route over the adjacent cycleway. It's better visible a bit to the west. You cannot do much about the problem in the first case. Routers would have to introduce routing over areas. Some mappers map virtual paths over such area but this is generally frowned upon and would require a lot of virtual paths here connecting all the entering paths. But the problem is a bit academic. A more realistic routing question would be solved quite ok in my opinion. Of course the turn-by-turn instructions don't really fit to the situation on the ground. Thanks, I think that you're probably right! Not sure that I agree that the problem is academic though. For example, my use case is to determine the which location (out of a given set of locations) that is closest (walking distance) to a user. If the user's location is on an unfortunate side of the pedestrian area polygon, this calculation becomes very wrong.
(04 Nov '19, 14:40)
Henrik
Would setting the area to pedestrian=yes instead of having it unset which i believe these both are. give the routers a chance work.
(04 Nov '19, 19:12)
andy mackey
|
