Open ulnn opened 4 years ago
@mapbox/navigation-api @danpat @danpaz Could you take a quick look at this when you have a chance 🙏?
I'm going ahead and transfer this into https://github.com/mapbox/mapbox-java as it's not directly an issue coming from the Navigation SDK.
@mapbox/navigation-api @danpat @danpaz If necessary, feel free to transfer it to somewhere else so you can keep track of it.
maybe it is also because a connecting footway is not explicitly set to "yes"? But the routing algo should also deal with it if there is a "paved way" element, or?
We got some advice from Mapbox support to add the tag foot=yes.
But....we have applied the suggested changes to the OSM dataset and it got moderated.
The comment for the moderation was: "For paths in OSM, there are some default values if there are no explicit other rights (https://wiki.osm.org/wiki/OSM_tags_for_routing/Access_restrictions#Germany). Therefore in this situation, it is unnecessary to add foot=yes."
So as far as we see this, the Mapbox DirectsionAPI routing algorithm requires a different configuration to run "smoothly" in Germany and the algorithm should - although this is not explicitly foot=yes tagged - consider this as a valid path.
What do you think? Many thanks in advance.
Hi all-
I am using turn-by-turn navigation with the Mapbox DirectionsAPI on Android and explore strange routing behavior when selecting the "walking" profile. Some routes are 3x longer than a comparable cycling route, especially highway=path and =footway seem to be avoided at all cost. I also pass walkwaybias=1 and alleybias=1 but it does not change the behavior.
An example here:
Any idea?