REVISED WITH DATA: King County Metro: Metro trakker shows very different data than onebus api

47 views
Skip to first unread message

Rafael Zimberoff

unread,
Feb 13, 2012, 12:54:56 AM2/13/12
to onebusa...@googlegroups.com
Sorry for the multiple emails.

Each time, I have more data!

Please see attached. This shows:

- king county metro trakker site
- onebus api response
- raw onebus response xml



Let me know how I can help!
onbus-itis-apir-2-12-2012 9-51-20 PM.png
onbus-itis-raw-response.xml
onbus-itis-trakker-2-12-2012 9-50-46 PM.jpg

Brian Ferris

unread,
Feb 14, 2012, 2:15:48 AM2/14/12
to onebusa...@googlegroups.com
Here's the deal:

Currently, King County Metro buses are tracked by one of two systems:

1) An older odometry+radio beacon system that's been around for over a
decade (I'll call this "legacy").
2) A newer GPS-based system that Metro has been installing for the
last year or so (I'll call this "gps").

KCM is slowly upgrading vehicles from the legacy system to the GPS
system. Currently, they've upgraded close to half of their vehicles.
Estimates on when the upgrade will be complete range from this summer
to this fall (but note that these deadlines have slipped in the past).
How can you tell the difference in OBA data? Vehicles that are using
the GPS-based system will have a "lastKnownLocation" element in their
vehicleStatus element that shows the last reported GPS coordinates of
the vehicle.

Both the old and the new systems are experiencing tracking errors.
Regarding the old system, this seems to be the most pervasive when it
comes to errors. The problem here is that the legacy prediction
system is very much a black-box in that there is little that
OneBusAway can do on its end to fix bad prediction data. What's more,
the staff at KCM haven't expressed much interest in trying to fix the
legacy system either and are instead putting most of their efforts
into upgrading to the new system.

Of course, the new system is not without bugs either. I know Metro is
working a lot harder on fixing those bugs, but they do appear from
time to time. However, they don't seem to be quite as pervasive.

Why were you seeing different data from the KCM Tracker pages and
OneBusAway? Without going into too much detail, it's important to
first note that you were looking at a vehicle tracked with GPS because
I can see a lastKnownLocation in your response data. For GPS-based
vehicles, we get data directly from an internal feed at KCM. However,
KCM wanted data from the new system to also go into their tracker
pages, which expects data in the old legacy format. To make that
work, someone at KCM wrote an adapter that converts GPS data into a
format that looks like the old legacy data. I can say that this
converter doesn't always work, as I've seen a lot of strange
prediction data for GPS-based buses showing up in the KCM tracker
pages. My short answer is: don't trust the Tracker pages, especially
for new buses.

Unfortunately, the larger take-away from this is: 1) tracking data
from KCM is messed up and 2) there isn't much that can be done to fix
it.

Thanks,
Brian

> --
> You received this message because you are subscribed to the Google Groups
> "OneBusAway API" group.
> To post to this group, send email to onebusa...@googlegroups.com.
> To unsubscribe from this group, send email to
> onebusaway-ap...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/onebusaway-api?hl=en.

Rafael Zimberoff

unread,
Feb 14, 2012, 9:52:17 AM2/14/12
to onebusa...@googlegroups.com
Brian,

Thank you for the nice write up!

So now I "get it" ....
 
Unfortunately, the larger take-away from this is: 1) tracking data
from KCM is messed up and 2) there isn't much that can be done to fix
it.

Now that is a serious mess for me, given that my entire app is built to consume KCM tracking data!

I will try woofing at a couple of the guys there, because this is a mess.

Raf

Reply all
Reply to author
Forward
0 new messages