Instructions (voice, visual) mismatch with own route on map

46 views
Skip to first unread message

Michel van der Kleij

unread,
Apr 8, 2021, 1:14:44 PM4/8/21
to OsmAnd

Hi All!

I'm a relatively new user, but I noticed something odd during navigation. When I create my own route by adding and dragging waypoints onto the map, there seems to be a mismatch between the voice and visual instructions to the point where the instructions seem totally unrelated (or at the very least, not synchronised) to the actual route shown on the map. I have "car" selected and during the track creation process, the route is properly following the roads after I add a waypoint, resulting in a route map that is (largely, see below) correct.

  • However, while driving I often get the instruction to go right whereas the map (correctly) shows a left or I'm asked to turn into some non-existing road. Not surprisingly, the instructions often use the wrong street name, whereas the map is correct.
  • The time needed to complete the route is way off: for something like 80k (again, for a car) it will start off with something like 40 hours but getting a little closer to something more realistic as we progress along the route
  • Often, there seem to be points the navigation wants to go to that are maybe 100m or so off the route, in which case even the map shows these spurious "detours". Most often these occur in intersections like T-junctions. So navigating on the map, they're easy enough to avoid as you can usually see where the correct route goes.

Strangely enough, routes that I download (I've tried primarily those from the web site of the Dutch tourist organisation ANWB) seem to do much better at these things than the routes I create myself.

Am I doing something wrong? Are there any settings I need to change?

Thanks!

Mich.

Tablet: Samsung Tab A (SM-P550), Android v7.1.1, kernel 3.10.49-13808168
OsmAnd: 3.9.10 (free)

Xavier

unread,
Apr 8, 2021, 3:45:37 PM4/8/21
to OsmAnd
On Thu, Apr 08, 2021 at 10:14:44AM -0700, Michel van der Kleij wrote:
>I'm a relatively new user, but I noticed something odd during navigation.
>When I create my own route by adding and dragging waypoints onto the map,
>there seems to be a mismatch between the voice and visual instructions to
>the point where the instructions seem totally unrelated (or at the very
>least, not synchronised) to the actual route shown on the map.

One question: When you add the waypoints, do you make sure that the
point of the "pin" of the waypoint is on an actual road? I.e., if the
waypoint is off to the side of a road perhaps the miss-match is OsmAnd
trying to take you "through" the actual waypoint spot?

I.e., I've accidentally dropped waypoints on the wrong side of a
divided highway and had odd routes created that have me doubling back
to cross the waypoint. I've since been careful to make sure the
waypoints are actually "on" the road segments I want to traverse at
that location in the route (this sometimes requires magnifing the view
to place the waypoint accurately).

> - However, while driving I often get the instruction to go right whereas
> the map (correctly) shows a left or I'm asked to turn into some
> non-existing road. Not surprisingly, the instructions often use the wrong
> street name, whereas the map is correct.

This might be caused by the waypoint being a short distance to the
right of the road at this location.

> - The time needed to complete the route is way off: for something like
> 80k (again, for a car) it will start off with something like 40 hours but
> getting a little closer to something more realistic as we progress along
> the route

Unsure on this one, other than the trip/arrival time is an estimate
based upon the speeds that OsmAnd believes you may travel. If most of
the road segments do not have a speed limit tag in the OpenStreetMap
data, then OsmAnd simply has to use some fixed value. If that fixed
value and the actual speed limit differ substantially, you can get
wildly incorrect estimates. So this might mean someone (or you) might
need to add the appropriate speed limit tags to OpenStreetMap for these
roads.

In any case, the predictions will better fit reality after you begin a
trip, as then OsmAnd can use your actual speed to better estimate a
rate of travel for the remainder, and produce a better estimate of time
left/arrival time.

> - Often, there seem to be points the navigation wants to go to that are
> maybe 100m or so off the route, in which case even the map shows these
> spurious "detours". Most often these occur in intersections like
> T-junctions. So navigating on the map, they're easy enough to avoid as you
> can usually see where the correct route goes.

This could also be a case where the waypoint pins are 100m off from the
road you are traveling.

>Strangely enough, routes that I download (I've tried primarily those from
>the web site of the Dutch tourist organisation ANWB) seem to do much better
>at these things than the routes I create myself.

Is it possible their routes better match the underlying streets?

>Am I doing something wrong? Are there any settings I need to change?

I am unsure. If for no other reason than it is easy, for your next
route, try to make sure any waypoints you drop are directly on top of a
road segment you want to traverse, and see if that improves the
results.

If it does, then this would imply that OsmAnd may not try to "snap"
waypoints to nearby roads when in the "car" routing mode.

Michel van der Kleij

unread,
Apr 9, 2021, 2:48:06 AM4/9/21
to OsmAnd
Thanks for that reply Xavier!

I think the "precision"of placing the points on the map could explain those spurious detours. However, those voice and visual instructions are WAY off and they seem to have almost nothing to do with the actual map. I mean, very often there's simply no road where it wants me to turn or it sends me in the opposite direction as to what the map shows on screen, it's as if it thinks I'm in a (very) different location for the instructions while the map is just spot on.

But for the next route, I will test higher precision with placing the waypoints, see if that helps ...

Op donderdag 8 april 2021 om 21:45:37 UTC+2 schreef Xavier:
Reply all
Reply to author
Forward
0 new messages