[Tagging] POI data and Addresses on areas - Was: addresses on buildings

Florian Lohoff f at zz.de
Fri Jan 10 11:29:42 UTC 2020


On Fri, Jan 10, 2020 at 12:11:10PM +0100, Marc Gemis wrote:
> > Footways are not part of the by car routeable network. And
> > access=no/access=private or highway=track and neither.

> This depends on the router, AFAIK Magic Earth navigates over private
> roads. It even combines it with access=destination roads that are
> connected to them.

I havent seen one in common use which does this. And i dont think any
navigatio solution right now does inter-modal selection of the nearest
spot on a road network, which you would need to do to include footways.

> > Commercial dataset map addresses back to nodes on the road - On specific
> > points. We as OSM rely on algorithms to do this automatically - And
> > fail in an astonishing large amount of cases.
> >
> > If you go to the center of a park by car there is no way the algorithm
> > can decide on the best spot to go there by _car_. So it uses the centroid
> > of the park, tries to find the nearest highway usable by your mode of
> > transportation. If that happens to be by car you might end up multiple
> > kilometers away from the optimal place to be. And there is nothing
> > we can do about it with this level of data abstraction.
> 
> There is no single 1 spot that is the best, even not for cars (or
> vehicles). Satnavs should allow you to refine the search for
> centroids. After telling the system that you want to drive to a park,
> it should (or could) offer you a few choices. Whether this is based on
> park entrances being mapped or parking places nearby is up to the
> software. I think this is not always solvable by mappers.

It is not solvable by algorithms. Mappers could select a node, or
multiple like on an Airport selecting multiple terminals and their
spots to go to.

> As for car navigation not taking paths and driveways into account.

> Again, that is a problem of the navigation software.
> I know OsmAnd uses driveways, as it always took my neighbour's
> driveway since in OSM it comes close to my house than my own driveway.
> OsmAnd does ignore any fence you draw. BTW, the private footpath to
> the frontdoor is not mapped.
> I do not think mappers should try to work around software limitations
> of their favourite satnav program. Those programs should be improved.

OSMAnd uses driveways, this is something i use extensively to fix
a lot of problems. But there are cases your CANT solve with this.
But OSMAnd (As all other Nav solutions - are not using inter-modal
road network nearest selection).

Think - Gated community with only footways. There is always a random
road near the building. But you never will get to the official parking
spot.

> We are so used to give an address and assume the software will bring
> us to the right spot, perhaps that flow has to be changed and not the
> data driving that flow.

Its not only we expect to reach the building for an address. We expect
this for ANY POI which is in our Database.

And as long as the mapper does not specific a very specific node
it is pure luck that we and up in the right spot. POIs on areas,
or addresses on areas and using a centroid of that area is just a
very broken approximation.

And the expectation is that a Nav Software offering a destination
to go to should REALLY know how to get there - Or show some
error bar e.g. "I can get you near the spot but its just an
approximation and my the error range is 1.5km"

Flo
-- 
Florian Lohoff                                                 f at zz.de
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20200110/b281858d/attachment.sig>


More information about the Tagging mailing list