Is this a routing bug?

74 views
Skip to first unread message

A Thompson

unread,
Aug 9, 2020, 9:11:46 PM8/9/20
to OsmAnd
I was driving to this car park:

Screenshot_20200810-012534_OsmAnd+.jpg


(North East England, Way 28284095: 54.78955 -1.31802)

It has a barrier=gate at the start of its service road, but nothing in OSM to say that it is inaccessible:

iMarkup_20200810_013234.jpg

 
Driving from the west we get this route:

Screenshot_20200810-012514_OsmAnd+.jpg


The service road is not taken, presumably because of the gate, even though there is nothing to say that it cannot be opened.

Instead, OsmAnd navigates to the closest point on a road and then imagines a road that is a straight line to the destination. The navigation instructions say to drive down this imaginary road as if it really exists!

This was very confusing for me when I was driving! Is it a bug, and if so do the developers already know about it?

Thanks!

(Android 10, latest Osmand+ from Google Play with latest map.)

Poutnik Fornntp

unread,
Aug 10, 2020, 12:24:53 AM8/10/20
to osm...@googlegroups.com, A Thompson
Is there anything in OSM saying it is accessible ?

Dne 10. srpna 2020 3:11:52 A Thompson <thomp...@gmail.com> napsal:

.....

It has a barrier=gate at the start of its service road, but nothing in OSM to say that it is inaccessible  .....

Greg Troxel

unread,
Aug 10, 2020, 7:27:43 AM8/10/20
to A Thompson, OsmAnd

A Thompson <thomp...@gmail.com> writes:

> It has a barrier=gate at the start of its service road, but nothing in OSM
> to say that it is inaccessible:

My long-term osm view has been that barrier=gate without any access tag
implies no access, rather that access yes/permissive.

I agree that the current wiki text implies that barrier=gate should
strictly be treated as a delay and not an access restriction. However,
I suspect there are a lot of barrier=gate tags which are not properly
accessible and it makes sense for routing to err to access no when not
tagged.

I am guessing that the appropriate action is to drive to the gate, get
out and open it, drive through, close it, and drive on. And that it is
not locked and anyone may do this, as a matter of right. Is that what
you mean?

I suggest that the proper resolution is to tag the access on this gate.

> Instead, OsmAnd navigates to the closest point on a road and then imagines
> a road that is a straight line to the destination. The navigation
> instructions say to drive down this imaginary road as if it really exists!

That is osmaand default, and if the destination is very close, e.g. a
hosue without mapped diveway, it makes sense.

> This was very confusing for me when I was driving! Is it a bug, and if so
> do the developers already know about it?

I would say it is a bug that it is not clearly "you have arrived at the
point on a road closest to your destination. The destination is 130m to
your left." But once you know it does this, it's clear.
signature.asc

A Thompson

unread,
Aug 10, 2020, 9:02:07 PM8/10/20
to OsmAnd
Thanks, folks, that's really informative!

I can see that OsmAnd's behaviour towards gates and inaccessible driving destinations is intentional and likely to be helpful with typical OSM data. The situation I described is unusual, and some announcement beyond "Turn right" (down a hypothetical road) would have been nice. But now I'm aware, it's not a big deal.

The OSM tagging for the access to this car park could be improved, but not by me. When I visited the gate *was* locked, I guess as a Covid-19 precaution. The area was well-maintained, so probably the gate would be left open otherwise. But it's not an area I know.
Reply all
Reply to author
Forward
0 new messages