landuse=residential and routing problems

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

landuse=residential and routing problems

mbranco
Doing remote mapping in Nigeria (project #2768), I've a very basic question (sorry, but I searched a lot in wiki pages and mailing lists, I didn't find anything about) :

all highways (usually 'unclassified') connecting a village need to be connected between them, for routing alghoritms?
If so, and the village has no residential highways, have we to draw "virtual" roads (with highway=residential) inside the village to get the connection?

Please look at this picture [1] to have an example.

Cheers,
Marco

[1] https://drive.google.com/open?id=0B65acVCG5NRQZ3VHTkFlaXR0RHM
Reply | Threaded
Open this post in threaded view
|

Re: landuse=residential and routing problems

john whelan-2
I do a lot of validation and in my opinion each settlement should have at least one highway that connects to the outside world.  That way its routable.  I might not draw in every path or track and small tracks leading only to fields or the river don't really add a lot of value.

When traveling to a settlement beyond the larger settlement you've drawn in people will pass through the village and the path they follow ie between huts should be mapped.  Most traffic will be foot, donkey or motorcycle so a highway=path is acceptable.

Cheerio John

On 31 July 2017 at 06:20, mbranco <[hidden email]> wrote:
Doing remote mapping in Nigeria (project #2768), I've a very basic question
(sorry, but I searched a lot in wiki pages and mailing lists, I didn't find
anything about) :

all highways (usually 'unclassified') connecting a village need to be
connected between them, for routing alghoritms?
If so, and the village has no residential highways, have we to draw
"virtual" roads (with highway=residential) inside the village to get the
connection?

Please look at this picture [1] to have an example.

Cheers,
Marco

[1]  https://drive.google.com/open?id=0B65acVCG5NRQZ3VHTkFlaXR0RHM
<https://drive.google.com/open?id=0B65acVCG5NRQZ3VHTkFlaXR0RHM>



--
View this message in context: http://gis.19327.n8.nabble.com/landuse-residential-and-routing-problems-tp5900081.html
Sent from the Humanitarian OpenStreetMap (HOT) mailing list archive at Nabble.com.

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot


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

Re: landuse=residential and routing problems

Arne Kimmig
In reply to this post by mbranco
Hi Marco,

Looking at your pictures I’d say that in both the highways are mapped too detailed.

My suggestions for mapping these:

-The main path is the one east of the res area. It is clear that it near to the res area and you don’t need to map every small access path to the village.
-also the two small villages to the west are nearby and IMHO don’t need extra access paths mapped, May be they could be included to the existing res area
-assuming that the path in the east leading out of the village leads to another village then, I think, it should be connected with the main path I mentioned above even we cannot discern it in the imagery. In such a case I assume a „virtual“ path meandering between the houses and connect it to the main path.

I’m mapping also in #2768. Feel free to have a look at the squares in the NE edge of the project. Suggestions are always welcome.

Best,

Arne






> Am 31.07.2017 um 12:20 schrieb mbranco <[hidden email]>:
>
> Doing remote mapping in Nigeria (project #2768), I've a very basic question
> (sorry, but I searched a lot in wiki pages and mailing lists, I didn't find
>
> all highways (usually 'unclassified') connecting a village need to be
> connected between them, for routing alghoritms?
> If so, and the village has no residential highways, have we to draw
> "virtual" roads (with highway=residential) inside the village to get the
> connection?
>
> Please look at this picture [1] to have an example.
>
> Cheers,
> Marco
>
> [1]  https://drive.google.com/open?id=0B65acVCG5NRQZ3VHTkFlaXR0RHM
> <https://drive.google.com/open?id=0B65acVCG5NRQZ3VHTkFlaXR0RHM>  
>
>
>
> --
> View this message in context: http://gis.19327.n8.nabble.com/landuse-residential-and-routing-problems-tp5900081.html
> Sent from the Humanitarian OpenStreetMap (HOT) mailing list archive at Nabble.com.
>
> _______________________________________________
> HOT mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/hot


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

Re: landuse=residential and routing problems

pierzen
Hi Marco,

Like Arne, I believe that too many roads are mapped and it such semi-desertic open space, we should not map all the traces that we observe on the ground.  The network of roads connecting between villages should be simplified. We should not see two, three higway=unclassified between two small villages.. Plus, it is essential that all these roads are connected one to the other (ie. they share a node) to let the routing softwares calculate properly the shortest distance.

On your image, the three residential areas are all part of Mula.  People and cattle can walk in all directions. We dont need to had highway=path between these three residential areas. If we want to add paths or track going to fields, it is important to connect these ways to the rest of the network.

If we zoom-out and observe the area, it is very difficult to understand the highway network. Too may roads classified highway=unclassified going from one village to the other. Also, the way coming from Madizo is a a dead end in Mula. It should be  connected to other roads.


Routing results Albineba to Mula show the detour the software are taking since they dont use the way from Madizo.

regard
 
Pierre



De : Arne Kimmig <[hidden email]>
À : mbranco <[hidden email]>
Cc : [hidden email]
Envoyé le : lundi 31 juillet 2017 10h50
Objet : Re: [HOT] landuse=residential and routing problems

Hi Marco,

Looking at your pictures I’d say that in both the highways are mapped too detailed.

My suggestions for mapping these:

-The main path is the one east of the res area. It is clear that it near to the res area and you don’t need to map every small access path to the village.
-also the two small villages to the west are nearby and IMHO don’t need extra access paths mapped, May be they could be included to the existing res area
-assuming that the path in the east leading out of the village leads to another village then, I think, it should be connected with the main path I mentioned above even we cannot discern it in the imagery. In such a case I assume a „virtual“ path meandering between the houses and connect it to the main path.

I’m mapping also in #2768. Feel free to have a look at the squares in the NE edge of the project. Suggestions are always welcome.

Best,

Arne


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

Re: landuse=residential and routing problems

mbranco
Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco




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

Re: landuse=residential and routing problems

Arne Kimmig
The only note I’m using when uncertain is fixme=confirm hoping that some time a local expert can fix it. I saw it being used in other projects and don’t think that another tag is necessary.

Arne





Am 31.07.2017 um 19:22 schrieb mbranco2 <[hidden email]>:

Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco





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

Re: landuse=residential and routing problems

pierzen
In reply to this post by mbranco
We have to understand that this is a sandy area in semi-desertic territory. Sometimes, we have too many infos (ie. all the paths of people going in various directions, taking shortcuts are printed at the rainy season). We have to interpret this info in the context of this semi-desertic area. And sometimes the traces + images in an area with no intense car circulation and no road infrastructure do not let us confirm from the image that there is a link passing through the village. Since there is no obstacles and it is logic that the two roads be connected, yes we draw the road link to connect the two roads where this looks the more logic.

 
Pierre



De : mbranco2 <[hidden email]>
À : Pierre Béland <[hidden email]>
Cc : "[hidden email]" <[hidden email]>; Arne Kimmig <[hidden email]>
Envoyé le : lundi 31 juillet 2017 13h22
Objet : Re: [HOT] landuse=residential and routing problems

Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco






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

Re: landuse=residential and routing problems

Palolo
In reply to this post by mbranco
Interesting situation Marco, one I've often seen in Ethiopia.
I would suggest that creating "virtual" roads is not a good idea.

The purpose of routing software is to help someone navigate the real world. If someone were to try to drive an imaginary route it would cause problems. Additionally I find it hard to imagine a scenario where someone would use a routing app to go 100 meters through a village.

The paths you have mapped are fine, but the key thing is to map the route to the next village or highway=unclassified+.

Emmor

On Mon, Jul 31, 2017 at 10:22 AM, mbranco2 <[hidden email]> wrote:
Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco




_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



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

Re: landuse=residential and routing problems

john whelan-2
Echoing Pierre's comments should we come up with better instructions?  I think the requirements for the NGOs are where do people live and how do we get there?

Which suggests we want the settlements and a highway to reach them.

This is probably different to normal OSM which sometimes seems to be map everything in sight.

Personally I think there are two styles of mapping a settlement.  One tightly maps round the buildings the other is looser in nature and maps groups of buildings more loosely.  Both have advantages.  Tight mapping means you can get a rough approximation of population without mapping every building.

Mapping fewer highways would let us cover a larger area for the same amount of effort but how to describe the optimum mix is difficult.

Cheerio John

On 31 Jul 2017 3:40 pm, "Vao Matua" <[hidden email]> wrote:
Interesting situation Marco, one I've often seen in Ethiopia.
I would suggest that creating "virtual" roads is not a good idea.

The purpose of routing software is to help someone navigate the real world. If someone were to try to drive an imaginary route it would cause problems. Additionally I find it hard to imagine a scenario where someone would use a routing app to go 100 meters through a village.

The paths you have mapped are fine, but the key thing is to map the route to the next village or highway=unclassified+.

Emmor

On Mon, Jul 31, 2017 at 10:22 AM, mbranco2 <[hidden email]> wrote:
Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco




_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot


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

Re: landuse=residential and routing problems

pierzen
I did not say to create "virtual roads" but to interpret the imagery in a specific context, here rural area in a semi-artic region with no road infrastructures like paved roads or gravel roads going to the small villages.  We have to interpret the traces on the sand. At the rainy season, cattle, people walking and cars leave traces in all directions.

It is yes difficult to interpret in such areas and this is a challenge to provide significant infos about the road network.

We often see that  too many highway=unclassified are traced in parallel going from village A to village B.

Your answer is too short saying do not connect A and B.

In the case of this specific village, Mulla, which ways would you keep as highway=unclassified?
If you keep one from the north, one from the south, would you say that this is impossible that a car goes through the village to move from the north way to the south way ?

What is virtual, what is the reality in such a context?  :-)
 
 
Pierre



De : john whelan <[hidden email]>
À : Vao Matua <[hidden email]>
Cc : [hidden email]
Envoyé le : lundi 31 juillet 2017 16h07
Objet : Re: [HOT] landuse=residential and routing problems

Echoing Pierre's comments should we come up with better instructions?  I think the requirements for the NGOs are where do people live and how do we get there?

Which suggests we want the settlements and a highway to reach them.

This is probably different to normal OSM which sometimes seems to be map everything in sight.

Personally I think there are two styles of mapping a settlement.  One tightly maps round the buildings the other is looser in nature and maps groups of buildings more loosely.  Both have advantages.  Tight mapping means you can get a rough approximation of population without mapping every building.

Mapping fewer highways would let us cover a larger area for the same amount of effort but how to describe the optimum mix is difficult.

Cheerio John

On 31 Jul 2017 3:40 pm, "Vao Matua" <[hidden email]> wrote:
Interesting situation Marco, one I've often seen in Ethiopia.
I would suggest that creating "virtual" roads is not a good idea.

The purpose of routing software is to help someone navigate the real world. If someone were to try to drive an imaginary route it would cause problems. Additionally I find it hard to imagine a scenario where someone would use a routing app to go 100 meters through a village.

The paths you have mapped are fine, but the key thing is to map the route to the next village or highway=unclassified+.

Emmor

On Mon, Jul 31, 2017 at 10:22 AM, mbranco2 <[hidden email]> wrote:
Thank you all for your answers, and for your hints not to map too much detailed.

For my question, the key seems to be: "roads must share a node to let the routing software calculate...".

A lot of times I find situations like this one [1]: there are no roads connecting the highway from NW and the highway from SE, but I think routing sw must know the two highways are connectable...

So, I'm undestanding that sometimes it could be necessary to draw "virtual" roads around or inside a village, to connect between them several external roads (of course, if it seems appropriate, looking at the aerial images).
If so, maybe it could be suitable to put a note (if not a specific tag), indicating that segment as "virtual"?

Thank you again,
Marco




______________________________ _________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.or g/listinfo/hot




______________________________ _________________
HOT mailing list
[hidden email]
https://lists.openstreetmap. org/listinfo/hot

_______________________________________________
HOT mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/hot



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

Re: landuse=residential and routing problems

mbranco
In reply to this post by mbranco
@Vao Matua (or Ebbor?) :

Sure you're right, and I also don't like to draw "virtual" roads (in the OSM world it should be unthinkable, but I got that HOT has specific issues to solve, and certain "golden rules" can have less priority).
So I think it's right also Perre Béland, when he says "Since there is no obstacles and it is logic that the two roads be connected, yes we draw the road link to connect the two roads where this looks the more logic".
Maybe not doing a "slalom" within the huts, but with an highway (maybe highway=road?) outside the residential landuse (with a tag fixme=confirm, as Arne Kimmig wrote).

Finally, John Whelan says the most important thing: "...should we come up with better instructions? ".  
I strongly think so, because I got that routing is a main need in many HOT projects, but the two most quoted guides (Highway Tag Africa wiki and West African HOT Mapping Tips by Bgirardot) don't mention routing issues, and the same in the mapping instructions of the HOT project.

Ciao,
Marco