Matching up DLR arrival data

[tfluke]

Has anyone had any success combing DLR arrival data to work out when a particular train will reach a future stop? For other modes, this is easy to do using the vehicleId field. But for DLR arrivals, vehicleId is empty. Any clever work arounds out there or alternate sources of data?

[Tim]

Hi Luke, unfortunately the backend system “DLR Daisy” doesn’t provide us with a VehicleId that we can use. Here’s a sample of all the information we receive:

24/03/2017 12:25:161 Lewisham 3 mins2 LEWISHAM 9 MINS3 LEWISHAM 13 MINS11 Bank 4 mins2 BANK 9 MINS3 BANK 14 MINS2CROCrossharbour24/03/2017 12:25:121 Woolwich Arsn2 LEWISHAM 2 MINS3 BECKTON 5 MINS11 Bank 1 min2 TOWER GATEWAY 3 MINS3 BANK 6 MINS2SHAShadwell24/03/2017 12:25:191 Woolwich Arsn 3 mins2 WOOLWICH ARSN 13 MINS3 WOOLWICH ARSN 23 MINS31 Stratford Int 8 mins2 

…etc.

[tfluke]

Ha! I remember before the unified API, trying to screenscrape the DLR website. This sample output looks very familiar.
For some reason I’d assumed that TFL would have access to a system further upstream than this. I guess there’s no obligation on Serco to provide anything more detailed…

[Tim]

Yes that’s probably about right

[grahamwell]

This is not a new issue, but I’d like to know if it’s simply an oversight. The arrivals feed for the DLR (which is humongous) doesn’t contain any data that identifies a vehicle. The vehicle ID field returns a blank. The ‘id’ field identifies the station, not the vehicle. The closest is the destination field, but typically there will be four or five quite different sources of arrivals data with the same destination field.

Previously mentioned here:

How do I uniquely identify a vehicle?

I’m looking for a way to distinguish different vehicles from one another. On the face of it vehcileId looks like a good candidate. Buses’ IDs are based on registration plates, so guaranteed to be unique. River buses are simple as they’re quite lower integers which aren’t zero padded. DLR trains don’t have vehicleIds, so we can safely ignore them. But I’m having trouble handling tube trains. I made the silly assumption that they’d also be unique, but then I was displaying weird routes (like…

It’s an issue because it stands alone, all other services have an identifier of some kind, although for the tubes there is often duplication. For the DLR there’s nothing at all. Is this a fundamental difficulty with the data source, or is it possibly just an oversight which could be fixed?

[LeonByford]

Hi, unfortunately the source system (DAISY) still does not provide vehicle IDs and therefore we cannot make this available in the Unified API.

This causes problems for us, too, so it’s definitely something we would like to address, and I know my colleague @jwithers is exploring possible ways we could potentially provide this in future.

[grahamwell]

Thanks and good luck with getting the data.