[OSM-talk] GDF (was: prohibited manoeuvres)

[Scott]

That's interesting.  In reference to GDF then (and in particular the
picture on page 3 of chapter 7), we could either load up the nodes with
tags, effectively using the node as the manoeuvre relation (it gives us
the GDF "via" bit), or we could create seperate manoeuvre entities.

           |   |
         E ^   v F    D
           |   |      |
           |   |      |
-->--------2---3---->-4- C
           |   |
--<-----------------<---
           |   |
           |   |
           ^   v
         A |   | B

For the sake of simplicity, I've split the A and B ways at C to each
make 2 ways A followed by E, and F followed by B.

Taking the first approach we could have

<node id="3">
  <prohibited_manoeuvre from="A" to="B"/>
</node>

<node id="4">
  <prohibited_manoeuvre from="A" to="D" />
</node>

On 2/8/2007, "Frederik Ramm" <frederik at remote.org> wrote:

>Hi,
>
>> Here is a simple but common example off the top of my head.  
>> apologies for
>> the poor ascii art... this is two dual carriageways meeting at a  
>> crossroads.
>> The prohibited manoeuvre is "no u turn" from A to B. This is 3  
>> segments/2
>> nodes.
>
>It's an immensely complex topic. Prohibited maneouvres could also  
>include rules for different lanes, a concept that is still unused in  
>OSM mapping.
>
>I have had a close look at data models in use for routing, and found  
>that the two big data providers (TeleAtlas and NavTeq) use the same  
>model, called GDF ("Geographic Data Files").
>
>GDF is two things at the same time: a data (transport) format (which  
>is very ugly, fixed-column-width-ASCII-based, looks like it's been  
>spit out by a mainframe system, and luckily of little relevance for  
>us) and a data model. It is also a CEN and ISO standard (ISO14825:2004).
>
>GDF information is hard to find on the web although the GDF data  
>model is used in almost all current navigation systems. An overview  
>is here (don't be alarmed by last modification times of some time in  
>1996 - road traffic hasn't fundamentally changed since):
>
>http://www.ertico.com/en/links/links/gdf_-_geographic_data_files.htm
>
>I implore anybody thinking about how to properly model things like  
>prohibited maneouvres to read the details about the data model,  
>linked from above page - it is really worth the time:
>
>http://www.ertico.com/download/misc/GDF/TOC1-5.pdf
>http://www.ertico.com/download/misc/GDF/PR6.pdf
>http://www.ertico.com/download/misc/GDF/PR7-9.pdf
>http://www.ertico.com/download/misc/GDF/PR10-12.pdf
>
>As you will see, GDF supports features (much like our nodes and ways,  
>with a little more complexity) and attributes (much like our tags),  
>but it has two additional important concepts: realationships and  
>restrictions. Attributes can be restricted to a part of the object.  
>For example, the attribute "road width" could be 8 metres for the  
>first km of the road and 12 metres for the rest, without having to  
>split the road in two.
>
>Such restrictions can also be based on lanes, e.g. you can have a  
>road with "max speed = 50mph", where the "max speed" attribute is  
>combined with a "lane dependent validity" restriction, making it  
>applicable only for the left lane.
>
>Maneouvres are modelled as relationships between features, and they,  
>in turn, may carry restrictions. So you can have a relationship  
>between two roads meeting at a crossroads that describes the  
>possibility of a turn, but this relationship can then be restricted  
>("not 8am-7pm (except buses and cycles)").
>
>I am not saying that we should aim to implement the full  
>functionality of the GDF data model (at least not at present), but  
>almost any question that will sooner or later pop up when you think  
>about routing has already been solved - or, at least, given thought -  
>there. So even if we do it differently, it is still worthwile to know  
>how others do it, especially if those "others" supply map data to 90%  
>of the world's navigation systems.
>
>Further reads for people interested in the field may be
>
>http://www.geoit.ethz.ch/staff/PDF/Joachim/2003_iain_baldegger.pdf  
>(this is mainly about pedestrian routing)
>
>and
>
>http://www.safe.com/services/formats/GDF_Presentation.zip (a somewhat  
>superficial PowerPoint presentation).
>
>Bye
>Frederik
>
>-- 
>Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'
>
>
>
>_______________________________________________
>talk mailing list
>talk at openstreetmap.org
>http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk