Proposed addition of Direction and Position to stops.txt
Sutherland, Van
Like TriMet, we have been providing our data to the Sendero Group for some time. But in our system, stop direction is related to the route, and is not specific to the stop. As an example, one of our stops in downtown Austin serves 15 separate routes. The direction of travel for 13 of the routes is northbound, and is eastbound for the other 2. Having a single direction assigned to the stop is not feasible. It would be possible, although more complicated, for Sendero or any other user of the GTFS to derive direction by linking the information in stop_times.txt to trips.txt using the trip_id field.
Van Sutherland
GIS Coordinator
Capital Metro
2910 East 5th Street
Austin, TX 78702
phone: (512) 369-6296
Roger Slevin
Your situation in downtown Austin points to the very reason why the UK NaPTAN database of public transport stops holds the variable “bearing” – which is the direction in which a bus is pointing when it is at the stop … and it is not necessarily the direction in which you might fly to the ultimate destination of the vehicle (which might not even be very local to the stop concerned). I don’t think the concept of direction (of the ultimate destination) is helpful in the European context – not even in the few locations where we have Grid-layout roads.
Roger Slevin
Traveline south east (UK)
Tom Hixson
an overall sense of a route's direction and is a good choice for
trip_headsign in trips.txt.
stops.txt is for stops only and is independent of service at the
stops. Stop direction is where *any* bus that might stop there is
heading, and merely helps you find the right stop.
In Van's example at Brazos/8th he agrees all buses are NB, which is
the stop direction. Knowing *which* NB bus to take is where the route
direction becomes useful. It's the context of the question (stop vs
service) that determines which kind of direction (stop vs route) to
consider.
With some spatial analysis one can get the direction of the trace
segment nearest the stop. (The distance from stop to trace is a handy
edit for the schedulers.) This could serve as a basis for stop
direction.
Tom Hixson
Sacramento
Joe Hughes
determine headings at a given stop based on the stop sequence (and
shapes if present).
I'm still unclear about whether that would meed the needs of the
Sendero application, though. Van (and/or TriMet folks), could you
give us some context on how Sendero uses this direction annotation in
their apps/devices? I want to make sure that we end up with something
that's as useful to riders as possible.
Joe
Devin Braun
MTS also uses the annotation Nearside, Farside, and Opposite, in addition to Mid-block. These annotations describe the stop location, not the direction of the bus at that stop. This notation allows us to tell passengers on which corner of an intersection a stop is. Since it is a property of the stop and relative the street direction, it doesn't matter which direction the routes travel from that stop.
Devin Braun
San Diego MTS
Mike Gilligan
Roger points out the NaPTAN example and "bearing" seems a better fit than my original proposal of direction.
As Devin pointed out, TriMet and MTS use the same notation for describing the corner of the intersection to our passengers. Many of our stops have the same name, usually the intersection name, and this data helps orient the customer to the intersection. However, it is only useful in combination with direction/bearing.
At this point, if Sendero is able to use the GTFS without direction & relative position, I do not have any compelling reason to add them.
Thanks,
-Mike
Sutherland, Van
I put the question of how direction is used in Sendero to Mike May, CEO of the Sendero Group. His response is below:
“Our users can query the actual street direction by using the virtual exploration mode. As I recall, we wanted to know the designated direction of the bus so a blind passenger could tell another sighted person they were waiting for the xxx Southbound. Perhaps that bus even announces itself as the xxx Southbound.”
It seems for this particular application, direction of travel for the route is more meaningful than facing direction of the stop.
Joe Hughes
about their use case--is the headsign information useful for this
application, or is it more important for them to have information
about cardinal direction that matches up with what the device gives
for the direction of the street?
Also, it's good that Roger clarified the distinction between the
bearing of the vehicle at a stop, and the qualifiers (like "near
side", "far side") that some agencies use to tell stops at the same
intersection apart.
Joe
T Sobota
provided to the Sendero group), is that the direction meant is related
to the trip - not the bus stop. Our data export needed multiple
instances of single bus stop coordinates, to the extent multiple
routes (or directions of a route) served that stop coordinate. For
example, one row of the data reflected trips in the northbound
direction of Route 2 serving a stop at coordinate x1,y1 - as did a
subsequent row of data for trips in the eastbound direction of Route 4
- as did yet another row for trips in the westbound direction of Route
4 (typical of a one-way loop travelled in both directions of a
route). All while the compass bearing of the bus on any of these
trips stopped at this location could well be 180 degrees south(bound).
It would seem the GTFS headsign data would (better) respond to the
question of what particular bus one is waiting for. Otherwise, as
noted earlier, the compass bearing of a vehicle when serving the stop
can often be the only consistently true value for a stop coordinate.
>
> >> Van refers to the nominal route direction, which is useful for getting
> >> an overall sense of a route's direction and is a good choice for
> >> trip_headsign in trips.txt.
>
> >> stops.txt is for stops only and is independent of service at the
> >> stops. Stop direction is where *any* bus that might stop there is
> >> heading, and merely helps you find the right stop.
>
> >> In Van's example at Brazos/8th he agrees all buses are NB, which is
> >> the stop direction. Knowing *which* NB bus to take is where the route
> >> direction becomes useful. It's the context of the question (stop vs
> >> service) that determines which kind of direction (stop vs route) to
> >> consider.
>
> >> With some spatial analysis one can get the direction of the trace
> >> segment nearest the stop. (The distance from stop to trace is a handy
> >> edit for the schedulers.) This could serve as a basis for stop
> >> direction.
>
> >> Tom Hixson
>
> - Show quoted text -