Navaid relation?

classic Classic list List threaded Threaded
33 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Navaid relation?

Florian Lohoff-2
Hi,
is there any Navaid relation to map an object to a point on
the routeable network?

Currently all Routing/Navigation application try hard to find the
nearest or best point on the routeable network for a given
destination lat/lon or object.

I have seen this fail for multiple occurences over the last
decade and i was pointed to one again from a collegue of mine.

Some examples:

- Large OSM objects like airports, campsites, parks, malls etc are not really
  identified by a node/point in OSM but an area. Currently it is mostly
  luck if this is correctly routable.
- Houses which are routeable by road a but are near road b or vice
  versa.

Adding more roads aka service/driveway does not necessary make it more
deterministic.

People try to work around by adding entrance=yes/main whatever tagging
to objects which work in application A but not B.

In my head i have a relation which has the source object, destination
node on the routeable network and the vehicle/mode of transportation
to use this for (car, bicycle and foot may have different destinations)

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

marc marc
Hello,

Le 21.05.19 à 23:46, Florian Lohoff a écrit :
> Currently all Routing/Navigation application try hard to find
> the nearest or best point on the routeable network for a given
> destination lat/lon or object.

with best, you mean : only one ? that look like wrong
a destination can have several points depending on the type
of locomotion.
the same type of locomotion can have several points, e.g. several ways
that desert a station by foot, several other for car, ...

> - Large OSM objects like airports, campsites, parks, malls etc are not really
>    identified by a node/point in OSM but an area. Currently it is mostly
>    luck if this is correctly routable.

put one or more node on the outer with entrance=yes/main :)
I use foot/vehicle/bicycle=designed to describe the type
of locomotion for this entrance

> - Houses which are routeable by road a but are near road b

add door/entrance but also add missing ways to connect to the road A

> work in application A but not B.

open an issue for B ?

Regards,
Marc
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2

Hi marc,

On Tue, May 21, 2019 at 10:02:53PM +0000, marc marc wrote:

> Hello,
>
> Le 21.05.19 à 23:46, Florian Lohoff a écrit :
> > Currently all Routing/Navigation application try hard to find
> > the nearest or best point on the routeable network for a given
> > destination lat/lon or object.
>
> with best, you mean : only one ? that look like wrong
> a destination can have several points depending on the type
> of locomotion.
> the same type of locomotion can have several points, e.g. several ways
> that desert a station by foot, several other for car, ...
What is the expectation to get navigated to when selecting a park? When
there are several for a car? Currently its random where you get
navigated to - There is no deteministic type to describe the nav point.
It depends on geometries of roads which might not describe the real
point you want to be guided to.

Geometries of the Park and Road follow strict "map what you see" or
"ground truth" - With this principle there is no way to map the nav
point without bending ground truth.

> > - Large OSM objects like airports, campsites, parks, malls etc are not really
> >    identified by a node/point in OSM but an area. Currently it is mostly
> >    luck if this is correctly routable.
>
> put one or more node on the outer with entrance=yes/main :)
> I use foot/vehicle/bicycle=designed to describe the type
> of locomotion for this entrance

This does not work for all cases and all objects.
 
> > - Houses which are routeable by road a but are near road b
>
> add door/entrance but also add missing ways to connect to the road A

Doesnt work - You are navigating by car but you only have a footway up
to hour house - Still the house is near road b - thus you get navigated
to the wrong street.

> > work in application A but not B.
>
> open an issue for B ?

So for every kind of object we open bugs for Applications a-z except
$RANDOMAPP ?

Wouldnt it make sense to have a generic description for this problem?

Currently i fake geometries to fix these kind of issues. Detaching
addresses from house geometries putting it near the road i expect
people to be navigated to.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Mateusz Konieczny-3
In reply to this post by Florian Lohoff-2



21 May 2019, 23:46 by [hidden email]:
- Houses which are routeable by road a but are near road b or vice
versa.

Adding more roads aka service/driveway does not necessary make it more
deterministic.
Can you give example of residential building with fully mapped roads, footways
and obstacles where well written router will fail?

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Graeme Fitzpatrick
In reply to this post by Florian Lohoff-2


On Wed, 22 May 2019 at 07:47, Florian Lohoff <[hidden email]> wrote:

- Houses which are routeable by road a but are near road b or vice
  versa.

That could be a "problem" due to GPS (?) system being so accurate?

For instance, where I am sitting now at the back of my house, various GPS / location / navigation systems tell me that I am actually at the address behind us, of That Road, rather than My Street, because I'm about 1m closer to that street!

If I turn OSMand on & ask it to take me from my present location to my Home, it will tell me to walk or drive around the block!

Thanks

Graeme

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Mateusz Konieczny-3
In reply to this post by Florian Lohoff-2



22 May 2019, 00:16 by [hidden email]:
Doesnt work - You are navigating by car but you only have a footway up
to hour house - Still the house is near road b - thus you get navigated
to the wrong street.
In this case router should be improved to route toward  house, not
to road segment closest to a building.

Fixing it in the router will take less effort that both
adding data over and over again (even assuming that this kind of data has a place in OSM)
AND changing router anyway to support this new relation.
> work in application A but not B.

open an issue for B ?

So for every kind of object we open bugs for Applications a-z except
$RANDOMAPP ?
Yes, there is no point opening one for $RANDOMAPP that works well.
Currently i fake geometries to fix these kind of issues. Detaching
addresses from house geometries putting it near the road i expect
people to be navigated to.
Please, stop damaging data as a workaround for low quality routers.
This is a case of https://wiki.openstreetmap.org/wiki/Tagging_for_the_renderer


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

marc marc
In reply to this post by Florian Lohoff-2
Le 22.05.19 à 00:16, Florian Lohoff a écrit :

>
> Hi marc,
>
> On Tue, May 21, 2019 at 10:02:53PM +0000, marc marc wrote:
>> Hello,
>>
>> Le 21.05.19 à 23:46, Florian Lohoff a écrit :
>>> Currently all Routing/Navigation application try hard to find
>>> the nearest or best point on the routeable network for a given
>>> destination lat/lon or object.
>>
>> with best, you mean : only one ? that look like wrong
>> a destination can have several points depending on the type
>> of locomotion.
>> the same type of locomotion can have several points, e.g. several ways
>> that desert a station by foot, several other for car, ...
>
> What is the expectation to get navigated to when selecting a park?

there is no such thing as "a single point that makes everyone agree"
take the case of a square park fenced by a fence, surrounded by 4
streets each with a pedestrian entrance to the park.
those arriving from the north will probably want to stop at the street
to the north, while those arriving from the south will probably want to
stop at the street at the south entrance.
Those who have difficulty walking will probably prefer the car park
closest to an entrance, while others will prefer parking for people with
reduced mobility or free parking, all while they have all come by car.
on the basis of which objective criteria will you decide which point of
the public network is most suitable to reach the park?

>> put one or more node on the outer with entrance=yes/main :)
>> I use foot/vehicle/bicycle=designed to describe the type
>> of locomotion for this entrance
>
> This does not work for all cases and all objects.

an exemple at least ?

>>> - Houses which are routeable by road a but are near road b
>>
>> add door/entrance but also add missing ways to connect to the road A
>
> Doesnt work - You are navigating by car but you only have a footway up
> to hour house - Still the house is near road b - thus you get navigated
> to the wrong street.

that's a bug/a feature needed in the routing, all info exist in osm
to find the path to reatch to the house, instead of leaving you at
a closer location but whose routing to the house is unknown

> Detaching addresses from house geometries putting it near the road
> i expect people to be navigated to.

so UGLY ! You decide, for the example of the park, that everyone
must use YOUR favorite entrance.
it is tagging for your-routing, instead of allowing routing to have
personalized options (shortest route, least walking, close to a parking,
disabled accessible up to the poi)

Regards,
Marc
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Mateusz Konieczny-3

22 May 2019, 00:38 by [hidden email]:
Le 22.05.19 à 00:16, Florian Lohoff a écrit :

Hi marc,

On Tue, May 21, 2019 at 10:02:53PM +0000, marc marc wrote:
Hello,

Le 21.05.19 à 23:46, Florian Lohoff a écrit :
Currently all Routing/Navigation application try hard to find
the nearest or best point on the routeable network for a given
destination lat/lon or object.

with best, you mean : only one ? that look like wrong
a destination can have several points depending on the type
of locomotion.
the same type of locomotion can have several points, e.g. several ways
that desert a station by foot, several other for car, ...

What is the expectation to get navigated to when selecting a park?

there is no such thing as "a single point that makes everyone agree"
take the case of a square park fenced by a fence, surrounded by 4
streets each with a pedestrian entrance to the park.
those arriving from the north will probably want to stop at the street
to the north, while those arriving from the south will probably want to
stop at the street at the south entrance.
Those who have difficulty walking will probably prefer the car park
closest to an entrance, while others will prefer parking for people with
reduced mobility or free parking, all while they have all come by car.
on the basis of which objective criteria will you decide which point of
the public network is most suitable to reach the park?
Also, some people may drive (bicycle), other may arrive by a public transport
others may walk or drive (by car) etc.

- a park that has multiple entrances, each may be preferred in some situation.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Jan S


Am 22. Mai 2019 00:44:51 MESZ schrieb Mateusz Konieczny <[hidden email]>:

>
>22 May 2019, 00:38 by [hidden email]:
>
>> Le 22.05.19 à 00:16, Florian Lohoff a écrit :
>>
>>>
>>> Hi marc,
>>>
>>> On Tue, May 21, 2019 at 10:02:53PM +0000, marc marc wrote:
>>>
>>>> Hello,
>>>>
>>>> Le 21.05.19 à 23:46, Florian Lohoff a écrit :
>>>>
>>>>> Currently all Routing/Navigation application try hard to find
>>>>> the nearest or best point on the routeable network for a given
>>>>> destination lat/lon or object.
>>>>>
>>>>
>>>> with best, you mean : only one ? that look like wrong
>>>> a destination can have several points depending on the type
>>>> of locomotion.
>>>> the same type of locomotion can have several points, e.g. several
>ways
>>>> that desert a station by foot, several other for car, ...
>>>>
>>>
>>> What is the expectation to get navigated to when selecting a park?
>>>
>>
>> there is no such thing as "a single point that makes everyone agree"
>> take the case of a square park fenced by a fence, surrounded by 4
>> streets each with a pedestrian entrance to the park.
>> those arriving from the north will probably want to stop at the
>street
>> to the north, while those arriving from the south will probably want
>to
>> stop at the street at the south entrance.
>> Those who have difficulty walking will probably prefer the car park
>> closest to an entrance, while others will prefer parking for people
>with
>> reduced mobility or free parking, all while they have all come by
>car.
>> on the basis of which objective criteria will you decide which point
>of
>> the public network is most suitable to reach the park?
>>
>Also, some people may drive (bicycle), other may arrive by a public
>transport
>others may walk or drive (by car) etc.
>
>See for example https://www.openstreetmap.org/#map=17/50.06288/19.91742
><https://www.openstreetmap.org/#map=17/50.06288/19.91742>
>- a park that has multiple entrances, each may be preferred in some
>situation.

I am also against mapping for navigation purposes. The data shall reflect the truth on the ground, all else is up to the programme making use of that data.

In case of a park with multiple entrances (which would obviously have to be recognisable by other programmes), it's up to the routing software to route you either to the entrance closest to you, the one that's best to be reached given the means of transport chosen or maybe offer you a choice of entrances to select from.

The same goes for big complexes like airports. Take Madrid Barajas or London Heathrow with terminal buildings at different locations, plus a cargo area and maybe a general aviation area. Here, you obviously have to chose a specific point you want to be routed to, and then the routing software should take you to an entry point of that specific place.

I imagine that like on a taxi. If you say "Take me to Heathrow", the driver will inevitably ask "Which terminal?". If you say "Take me to Central Park", the driver will, again, ask you for a more specific location or make a decision for you (the nearest/furthest entry, depending on his/her honesty ;)). These very decisions should be made, or these very questions asked by a routing programme.

Best, Jan

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by Mateusz Konieczny-3

Hola Mateusz,

On Wed, May 22, 2019 at 12:26:01AM +0200, Mateusz Konieczny wrote:

> 21 May 2019, 23:46 by [hidden email]:
>
> > - Houses which are routeable by road a but are near road b or vice
> >  versa.
> >
> > Adding more roads aka service/driveway does not necessary make it more
> > deterministic.
> >
> Can you give example of residential building with fully mapped roads, footways
> and obstacles where well written router will fail?
- Baumstraße 43a, Gütersloh, Germany
        It does not have a connection to Baumstraße but to
        Hermann-Vogelsang-Straße.

        It still will be routed through Baumstraße and the driveway to
        Baumstraße 45a

        IMHO unfixable without bending geometry

- Dalbker Straße 40a, Oerlinghausen, Germany
  Dalbker Straße 44a, Oerlinghausen, Germany
        Both fixed by moving the Address to a node, moving it further to
        Dalkbker Straße within the outline of the Building.

        Otherwise you'd be routed to Buchweizenweg without access to
        the Building. This is what multiple of my collegues complained
        about.

        Here i bend geometry until it halfway worked.

Just 2 examples for the last 10 days or so. And footways wont help
you in a car profile. You could bring a footway directly connecting
the entrance to the Street.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by marc marc

Hi Marc,

On Tue, May 21, 2019 at 10:38:23PM +0000, marc marc wrote:
> > What is the expectation to get navigated to when selecting a park?
>
> there is no such thing as "a single point that makes everyone agree"

Yes there is - there has to be an explicit location you will ne navigated
to for a certain feature.

It might be that there are multiple - then you might want to choose.
Currently you'll be guided to a random location which fits whatever
nearest point algorithm your router/navigation uses. Non deterministic
from a users perspective.

> take the case of a square park fenced by a fence, surrounded by 4
> streets each with a pedestrian entrance to the park.
> those arriving from the north will probably want to stop at the street
> to the north, while those arriving from the south will probably want to
> stop at the street at the south entrance.

Tell me ONE application that does this. None. All of them map a feature
you select (Address, POI, whatever) - map it to a single precise point
and bring you there.

> > Doesnt work - You are navigating by car but you only have a footway up
> > to hour house - Still the house is near road b - thus you get navigated
> > to the wrong street.
>
> that's a bug/a feature needed in the routing, all info exist in osm
> to find the path to reatch to the house, instead of leaving you at
> a closer location but whose routing to the house is unknown

Nope - there isnt enough information. Its all just implicit and works
for 95% of the cases. It breaks horrible in others and we fake
geometries to fix it, blame the application, invent tags to guide the
nav/routing which only fit half of the object and only half of the
apps support. In all cases the user is in trouble.

> > Detaching addresses from house geometries putting it near the road
> > i expect people to be navigated to.
>
> so UGLY ! You decide, for the example of the park, that everyone
> must use YOUR favorite entrance.

No - i expect to select a feature and when there are multiple entrances
i expect the Nav to ask me. Currently it doesnt - no application does.
They simply guide me to some arbitrarily choosen point calculated
from some geomtry which fits some nearest point matching algorithm. And
there is NO way to fix this.

> it is tagging for your-routing, instead of allowing routing to have
> personalized options (shortest route, least walking, close to a parking,
> disabled accessible up to the poi)

No - as i said - its about hinting. We have a low percentage of
nav targets which are not or suboptimally reached with current
alhorithms. Its about a navaid hint to help algorithms choose
the right and optimal destination point for a certain object and mode
of transportation.

If its foot and you have a footway to the entrance - Its the entrance.
If there is an explicit car park than its that for the car. If its
by bus you can select the bus stop (in which case you switch mode of
transportation and have your next destination from the navaid relation)

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by Graeme Fitzpatrick
On Wed, May 22, 2019 at 08:31:03AM +1000, Graeme Fitzpatrick wrote:
> On Wed, 22 May 2019 at 07:47, Florian Lohoff <[hidden email]> wrote:
> > - Houses which are routeable by road a but are near road b or vice
> >   versa.
>
> That could be a "problem" due to GPS (?) system being so accurate?

And map data beeing very accurate. With commercial data sets you'll map
addresses to street segments. Done.

We aim much higher. We want exact locations of buildings and we expect
some clever algorithms to match these addresses to roads. Works for
90% of the cases. And fails in others.

And then we aim even higher. We want all the corner cases to work.
Issues as the above - Address on road A - Reachable only via road B but
near road A. Without hinting there is no way an algorithm will be able
to determine this.

> For instance, where I am sitting now at the back of my house, various GPS /
> location / navigation systems tell me that I am actually at the address
> behind us, of That Road, rather than My Street, because I'm about 1m closer
> to that street!
>
> If I turn OSMand on & ask it to take me from my present location to my
> Home, it will tell me to walk or drive around the block!

Its not an GPS issue - Navigation works different - You tell where you
want to go and it selects a precise and explicit point on the routing
graph where to take you to. And it'll follow on the graph to that
location. The mapping of an address to a location on the routeable
graph is the Problem - not the routing/gps. This is why i called
it "navaid" not "routingaid".

Id imaging a relation which says

object A -> car -> Node B (on routable graph)

So whenever i tell my nav software to bring me to object A the node
selected on the routable graph as a destination will be Node B.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Peter Elderson
>  Id imaging a relation which says
> object A -> car -> Node B (on routable graph)
> So whenever i tell my nav software to bring me to object A the node
> selected on the routable graph as a destination will be Node B.  

One relation per mode of transport then? So a complex obejct a could have many navaid relations? Or one relation containing all nodes, with roles for transport mode?
Vr gr Peter Elderson


Op wo 22 mei 2019 om 10:02 schreef Florian Lohoff <[hidden email]>:
On Wed, May 22, 2019 at 08:31:03AM +1000, Graeme Fitzpatrick wrote:
> On Wed, 22 May 2019 at 07:47, Florian Lohoff <[hidden email]> wrote:
> > - Houses which are routeable by road a but are near road b or vice
> >   versa.
>
> That could be a "problem" due to GPS (?) system being so accurate?

And map data beeing very accurate. With commercial data sets you'll map
addresses to street segments. Done.

We aim much higher. We want exact locations of buildings and we expect
some clever algorithms to match these addresses to roads. Works for
90% of the cases. And fails in others.

And then we aim even higher. We want all the corner cases to work.
Issues as the above - Address on road A - Reachable only via road B but
near road A. Without hinting there is no way an algorithm will be able
to determine this.

> For instance, where I am sitting now at the back of my house, various GPS /
> location / navigation systems tell me that I am actually at the address
> behind us, of That Road, rather than My Street, because I'm about 1m closer
> to that street!
>
> If I turn OSMand on & ask it to take me from my present location to my
> Home, it will tell me to walk or drive around the block!

Its not an GPS issue - Navigation works different - You tell where you
want to go and it selects a precise and explicit point on the routing
graph where to take you to. And it'll follow on the graph to that
location. The mapping of an address to a location on the routeable
graph is the Problem - not the routing/gps. This is why i called
it "navaid" not "routingaid".

Id imaging a relation which says

object A -> car -> Node B (on routable graph)

So whenever i tell my nav software to bring me to object A the node
selected on the routable graph as a destination will be Node B.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by Florian Lohoff-2
On Wed, May 22, 2019 at 09:43:31AM +0200, Florian Lohoff wrote:

> > Can you give example of residential building with fully mapped roads, footways
> > and obstacles where well written router will fail?
>
> - Baumstraße 43a, Gütersloh, Germany
> It does not have a connection to Baumstraße but to
> Hermann-Vogelsang-Straße.
>
> It still will be routed through Baumstraße and the driveway to
> Baumstraße 45a
>
> IMHO unfixable without bending geometry
>
> - Dalbker Straße 40a, Oerlinghausen, Germany
>   Dalbker Straße 44a, Oerlinghausen, Germany
> Both fixed by moving the Address to a node, moving it further to
> Dalkbker Straße within the outline of the Building.
>
> Otherwise you'd be routed to Buchweizenweg without access to
> the Building. This is what multiple of my collegues complained
>         about.
>
> Here i bend geometry until it halfway worked.
>
> Just 2 examples for the last 10 days or so. And footways wont help
> you in a car profile. You could bring a footway directly connecting
> the entrance to the Street.
And most likely - Try and schools in your area. If its correctly
mapped with the school ground as an amenity=school and buildings
etc.

Try to reach it by car:

        Elly-Heuss-Knapp Realschule, Gütersloh

You'll end on the wrong street without parking and a gate although
the school has a parking lot. And you even end up in a oneway hell.

Whenever you try to reach "large" osm objects on the routable network
you are in deep trouble. camp sites, airports, golf courses, schools.
People move the attributes from the polygons to nodes to make it
explicit where to navigate. With polygons its up to the algorithm
to guess the right point.

And i had tons of stuff like this the last decade. 2 Years ago
i ended up on a road at a river (by bike) the campsite on the opposite
side of the river. 5km to next bridge to cross the river because the
road 20m on the other river side was the nearest point on the routable
network. Roads on the campsite were access=private.

I ended here:
https://www.openstreetmap.org/?mlat=53.92106&mlon=12.10538#map=17/53.92106/12.10538

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by Peter Elderson
Hi Peter,

On Wed, May 22, 2019 at 10:15:57AM +0200, Peter Elderson wrote:
> >  Id imaging a relation which says
> > object A -> car -> Node B (on routable graph)
> > So whenever i tell my nav software to bring me to object A the node
> > selected on the routable graph as a destination will be Node B.
>
> One relation per mode of transport then? So a complex obejct a could have
> many navaid relations? Or one relation containing all nodes, with roles for
> transport mode?

type navaid
name Foobar (Optional if object does not carry a name or you map
             multiple different location)
source <object to reach> (Multiple ones sharing the same transport destination)
car    <node to go to>
bicyle <node to go to>
foot   <node to go to>

Just as the rough first version. You want to keep this to a minimum for
cases where an algorithm cant choose the right solution or is making
wrong decisions based on correct geometries.

A mall with parking south and north would get 2 navaid relations

        name "Mall foo - North"
        name "Mall foo - South"

With their parking lot entries as the nodes to go to by car
and their entrance for foot. Currently you might be lucky
that your Nav Software does a full text search and somebody
named the parking lots with the complete name although
they should only carry a ref=North/South.

So a navigation preprocessor would not try to find a node on the
routable network by algorithm but by hinting in case it finds
this relation.

It currently works by accident and because people tweak
the data to get their result. Either geometries, additional tags
or additional name tags on unrelated objects.

People start mapping for the router/nav software. A relation like this
could help solve the need of hinting the software without abusing
other tags, tag move, geometry tweaking.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

marc marc
In reply to this post by Florian Lohoff-2
Le 22.05.19 à 09:43, Florian Lohoff a écrit :
>> Can you give example of residential building with fully mapped roads,
>> footways and obstacles where well written router will fail?
>
> - Baumstraße 43a, Gütersloh, Germany

you mean https://www.openstreetmap.org/way/273023376 ?
it's a good example of missing datas.
no entrance, no way between the entrance and the public network.
please, there should be an example of a target where the current schema
are not enough, and not an example where the lack of data produce
a bad routing

right now I feel that the relation type=navaids should be called
type=missingway
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

marc marc
In reply to this post by Florian Lohoff-2
Le 22.05.19 à 10:36, Florian Lohoff a écrit :
> type navaid
> source <object to reach> (Multiple ones sharing the same transport destination)
> car    <node to go to>
> bicyle <node to go to>
> foot   <node to go to>

this info, for well mapped objet, already exist
polygon  <object to reach> have several nodes
with entrance=yes and car/bicyle/foot=designated
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Florian Lohoff-2
In reply to this post by marc marc
On Wed, May 22, 2019 at 09:13:14AM +0000, marc marc wrote:

> Le 22.05.19 à 09:43, Florian Lohoff a écrit :
> >> Can you give example of residential building with fully mapped roads,
> >> footways and obstacles where well written router will fail?
> >
> > - Baumstraße 43a, Gütersloh, Germany
>
> you mean https://www.openstreetmap.org/way/273023376 ?
> it's a good example of missing datas.
> no entrance, no way between the entrance and the public network.
> please, there should be an example of a target where the current schema
> are not enough, and not an example where the lack of data produce
> a bad routing
>
> right now I feel that the relation type=navaids should be called
> type=missingway
Again a footway between the house and a road will NOT help for
car navigation because for cars a footway is NOT a routable
part of the graph.

Flo
--
Florian Lohoff                                                 [hidden email]
        UTF-8 Test: The 🐈 ran after a 🐁, but the 🐁 ran away

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Mateusz Konieczny-3



22 May 2019, 12:06 by [hidden email]:
On Wed, May 22, 2019 at 09:13:14AM +0000, marc marc wrote:
Le 22.05.19 à 09:43, Florian Lohoff a écrit :
>> Can you give example of residential building with fully mapped roads,
>> footways and obstacles where well written router will fail?
>
> - Baumstraße 43a, Gütersloh, Germany

you mean https://www.openstreetmap.org/way/273023376 ?
it's a good example of missing datas.
no entrance, no way between the entrance and the public network.
please, there should be an example of a target where the current schema
are not enough, and not an example where the lack of data produce
a bad routing

right now I feel that the relation type=navaids should be called
type=missingway

Again a footway between the house and a road will NOT help for
car navigation because for cars a footway is NOT a routable
part of the graph.
Car navigation may use footway data to select best dropoff point.

In exactly the same way as you proposed with navaid relation,
but without adding subjective data.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Navaid relation?

Mateusz Konieczny-3
In reply to this post by Florian Lohoff-2



22 May 2019, 09:53 by [hidden email]:

Hi Marc,

On Tue, May 21, 2019 at 10:38:23PM +0000, marc marc wrote:
> What is the expectation to get navigated to when selecting a park?

there is no such thing as "a single point that makes everyone agree"

Yes there is - there has to be an explicit location you will ne navigated
to for a certain feature.
This is blatantly untrue. Depending on location you will prefer to be routed
to different entrances and it is not considering different modes of transport.
> Doesnt work - You are navigating by car but you only have a footway up
> to hour house - Still the house is near road b - thus you get navigated
> to the wrong street.

that's a bug/a feature needed in the routing, all info exist in osm
to find the path to reatch to the house, instead of leaving you at
a closer location but whose routing to the house is unknown

Nope - there isnt enough information. Its all just implicit and works
for 95% of the cases. It breaks horrible in others and we fake
geometries to fix it, blame the application, invent tags to guide the
nav/routing which only fit half of the object and only half of the
apps support. In all cases the user is in trouble.
Please give a specific example.
No - i expect to select a feature and when there are multiple entrances
i expect the Nav to ask me. Currently it doesnt - no application does.
They simply guide me to some arbitrarily choosen point calculated
from some geomtry which fits some nearest point matching algorithm. And
there is NO way to fix this.
If navigation is simply doing nearest road point on matching then it requires change to both
- properly use footway data
- use your proposed relation

I see no reason for preferring second solution.



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
12