RE: Shapes.txt - Roger Slevin - com.googlegroups.googletransit

10 messages in com.googlegroups.googletransitRE: Shapes.txt

From	Sent On	Attachments
Tranplanner	19 Mar 2008 08:59
Dale Noll	19 Mar 2008 09:38
Tranplanner	19 Mar 2008 10:58
Accidental Guru	19 Mar 2008 15:49
Dale Noll	19 Mar 2008 18:58
Dale Noll	19 Mar 2008 19:40
Roger Slevin	20 Mar 2008 01:02
Bob Heitzman	20 Mar 2008 10:07
Accidental Guru	21 Mar 2008 02:58
Accidental Guru	21 Mar 2008 03:01

Subject:	RE: Shapes.txt
From:	Roger Slevin (ro....@slevin.plus.com)
Date:	03/20/2008 01:02:31 AM
List:	com.googlegroups.googletransit

I can see why this can be a significant problem in American towns laid out primarily with grids of roads ... but in Europe, with its fundamentally different road layout, my experience is that you CAN plot routes following the shortest road route between adjacent stops. There can be a few problems - but for the majority of circumstances this works well. Look at the route maps for any service at travelinesoutheast.org.uk - these are all created automatically on the fly, using only the stop points data, and the underlying GIS road-routing data ... no shape files or anything else extra to make them work.

The problems arise typically when there are special traffic management measures for public transport that do not reflect those which apply to ordinary traffic (turns that can be made by bus, roads that are closed to all traffic except buses, etc) - and these need some work to enhance the road routing engine. And there may be shorter routes using roads which are unsuitable for buses - again requiring attributes to be attached to those unsuitable road links. The third "problem" can arise with services which stop infrequently - but most of those follow the quickest road route.

One final comment, though, is "does accuracy matter" in this context ... after all, a passenger can only get on or off at the defined stopping points, A credible route line that follows the underlying roads might be better than the current straight line projections - even if occasionally that line follows roads which the bus doesn't follow.

Roger

-----Original Message----- From: goog...@googlegroups.com [mailto:goog...@googlegroups.com] On Behalf Of Dale Noll Sent: 20 March 2008 01:59 To: Google Transit Trip Planner Subject: Re: Shapes.txt

Daniel,

Speaking from a bus operation standpoint, the 'cross over land where no road exists' issue is simply a matter of drawing a straight line between to consecutive bus stops. If the stops of sufficiently close together it would be possible for Google to guess the routing, but if the stops are even a couple of blocks away, there can be ambiguity in the routing. This is not a problem when driving in a car, but for a fixed route bus, it does not work. That is the importance of the shapes.txt file. For instance, we have a specialty route that serves an industrial area in the suburbs, but the west end stops are miles apart while the east end stops are closer together. On the east end, the route trace stays on the streets pretty well, but on the west end there is adherence to streets. Google _could_ calculate a route between stops but that does not mean that is where the bus actually goes.

Bus routing does not always follow the shortest path, fastest path or necessarily an apparently logical path. Especially when politics may be involved. ;-)

Dale

Accidental Guru <dani...@gmail.com> 3/19/2008 5:49 PM >>>

Hi Heather,

HASTUS export to Google Transit does not include the shape file, yet.

It is one of the things that needs to be done to make Google Transit easier for agencies to publish. Right now the data for trips gets to Google, but sometimes the routes will cross over land where no road exists.

It is the shortest point between two lines so it makes sense, but Google driving directions do not do this.

If any Google folks are reading this thread, can you explain why transit has this unique problem?

Is it because of trains/ferries/etc?

Thanks,

Daniel

On 19 Mrz., 10:59, Tranplanner <Hmba...@gcrta.org> wrote:

Hi,

I have no knowledge when it come to feeds, so your explanation makes sense....

Questions: What does GCRTA you use to build your schedules?

HASTUS

Does that package have the routes build graphically in it?

Yes

Do you have a lot of branching or alternate routing on a trip by trip

basis?

Yes

Is your GIS data for a route including all patterns or do you have each

possible pattern mapped separately?

GIS data is set up to show the entire route, not individual patterns.

I dont have patterns in GIS. I did realize when the Web guy asked the question, that it might be time consuming to try and compile the data this way?

Is there something that can be done with HASTUS, it seems to export some of the required files, but not the shape file needed?

Thanks,

Heather

On Mar 19, 12:38 pm, "Dale Noll" <dn....@mcts.org> wrote:> Heather,

First off, let me say this is coming from someone at a transit system

(Milwaukee, WI) not a Google Transit person.

Now, I do not know how technical you are when in comes to data feeds, so

please forgive me if I get to basic or too geeky.

The Google Transit Feed Spec is located at:

http://code.google.com/transit/spec/transit_feed_specification.html

Visit that site and it will describe what the files need. The shapes

file is basically a series of lat/lon values. But the key to this is the shape_id. That shape ID is also part of the Trip data. There will be at least 2 trip patterns for a route, one for the first direction and one for the return direction. Take for example the Rotue 8. A trip that starts at A and travels westbound to C. The shapes.txt will have a list of all lat/lon points along that trip, in the order the vehicle travels the route. That collection of points is assigned a shape_id, for instance 8_0_1, for route 8, direction 0, pattern 1. Any trip that travels the exact same sequence from A to C will have the shape_id of 8_0_1 assigned to it in the GTFS data. Now the return trip from C to A would be the sequence of lat/lon points going the other way and have a unique shape_id as well, say 8_1_2 and those trips would need that shape_id included on the records as well. Now if there is a branch or alternate route, the whole thing is repeated again, for both directions. As you could guess, this could become extremely time consuming if you do not already have the individual trip patterns in your GIS system.

Questions: What does GCRTA you use to build your schedules? Does that package have the routes build graphically in it? Do you have a lot of branching or alternate routing on a trip by trip basis? Is your GIS data for a route including all patterns or do you have each

possible pattern mapped separately?

Dale

Tranplanner <Hmba...@gcrta.org> 3/19/2008 11:00 AM >>>

Hi.

I have been asked by our webmaster to create the shapes.txt file. Being completely new and inexperienced with Google Transit feed, I need help. I work with GIS files and have a map of all the routes. I have no idea how to create the shapes.txt file, how to get the data, or anything.

i would greatly appreiciate any guidance you might have for beginners like me.

Thanks

Heather Bates