[OSM-talk] Is anyone making public transport routing maps based onOpenStreetMap data?

Hugh Barnes list.osm at hughbris.com
Wed Dec 17 09:12:35 GMT 2008

On Wed, 17 Dec 2008 02:29:56 +0000
Peter Miller <peter.miller at itoworld.com> wrote:

> On 16 Dec 2008, at 23:30, Ævar Arnfjörð Bjarmason wrote:
> > I'm interested in completely mapping my city bus network, it would
> > be great if there was some online routing application that I could
> > go to that could plan my routes. Of course I'd have to provide it
> > with sufficient survey information to do this, which would be part
> > of mapping it obviously.
> >
> > Routing applications based on OSM data also have the opportunity to
> > do inter-network routing. You could step onto a bus in one city,
> > take a rail to another one, inter-city bus to yet another city,
> > then a bus and walk on a footway to your destination. All based on
> > OSM data.
> >
> I am very interested in such an application and have taken some time  
> to see what is happening around the world.

It's the main application I had in mind when I got into online mapping a
few years back. http://busmonster.com was particularly inspirational for

> Can I suggest that one takes a layering approach to this (as the  
> professional transport sector does) and some layers belong in OSM
> and some not...

Yes, timetabling probably doesn't belong.

> Firstly the bus stops (or more generally 'stop 'points) which is
> where one physically accesses the transport system which should be
> point features within OSM.

Could you elaborate on this? Reuse existing nodes on ways? Point
features as opposed to what? Are you stressing "within OSM"?

> Secondly the routes the vehicles take which traveling on the network  
> to get from stop point to stop point. In most cases this is obvious,  
> but in a limited number of cases one will need to include route
> points that are not stop points. These might use the route relation
> and detail every way that is involved for every route, but this is
> more detail than a route planner needs that can work out most stop to
> stop routing without guidance.

I have been (mostly) following the guidelines on
http://wiki.openstreetmap.org/wiki/Relation:route, where we use a
relation, which principally includes ways and the stops (which sit
adjacent as nodes as I've done them), as members. Then your problem
goes away.

(It's admittedly a little sad to have to break up ways for routing. I'm
also conscious that other mappers further dissecting them will probably
break the relation, but I think that might be an issue for the API to
address :~) )

> Including detailed routing in OSM
> means that it has to be updated every time the schedules change.

The schedules? or the route? Whichever, remember there are currently
mappers embedding business phone numbers as shop metadata.

> All of the rest of the data can then be in Google Transit Feed  
> Specification (an open source data standard controlled by Google)
> and can feed GraphServer or equivalent for route planning. GT is not  
> perfect and can't represent complex rail journeys but it is open  
> source and there is data available already in it that can be used
> and it is a good starting point:
> http://code.google.com/p/googletransitdatafeed/wiki/PublicFeeds

Ewww, CSV serialisations requiring their own purpose-built validator …
Like you say, we can build from it. Let's look through the
fields/elements, but make something proper and scalable that leverages
XML as it should.

> I am not sure how one would explicitly refer to the schedules file  
> from OSM. Possibly all the stop points in a area would be part of a  
> 'network' relation that that network relation would refer to the  
> external schedules file using a 'schedules' URL.
> http://wiki.openstreetmap.org/wiki/Relations/Proposed/Network

Works for me.

> Would it be useful to create a list for discussion of public
> transport applications within OSM. Could I suggest a title of
> 'talk-transit' or should this conversation be part of the
> 'talk-routing' list? Including PT routing in the talk-routing list
> might make some sense because there is always a walking element to
> the routing and people interested in routing may also be interested
> in PT routing. I certainly think this conversation needs a 'home'
> that is off the main talk list which is too busy already.

+1, preferably for a separate list. I'm not sure how welcome transit
discussion would be on the routing list. (I'm curently subscribed to it
only on the off-chance some transit material will surface.)

I have been under the impression there isn't too much interest for this
within OSM, judging by the paucity of activity on the lists and content
on the wiki. I think it's another potential killer app for OSM.


More information about the talk mailing list