Commuter Rail onboard electronic passenger counting

[StefanW]

Today for the first time I was on the consist that has all the coaches equipped with the electronic passenger counters. It was 2008-838-825-745D-722D-1818 for train 123.

(There have been a couple of mentions of the camera / video counting previously http://www.railroad.net/forums/viewtopic.php?f=65&t=137379&p=1429658#p1429658 but I think this deserves its own topic.)

I first saw the pilot coach of the project 838 back on Sept. 6 2016. On that trip I met one of the project engineers and he gave me all the details.

The company behind the project is mTHINX Integrated Technologies http://www.mthinx.com/ of Dallas. They are also the company behind the electronic video-analysis passenger counting trial that happened at South Station on the Track 5 platform back in 2015.

http://www.bostonmagazine.com/news/blog/2015/05/06/mbta-count-commuter-rail-passengers/

https://www.bostonglobe.com/metro/2015/05/05/commuter-rail-tests-out-new-technology-count-passengers/MAl66uifH9UPRLRIVky41L/story.html

Keolis is conducting a 30-day trial with new motion sensor technology on Platform 5 at South Station, said Leslie Aun, a spokeswoman for the company. The technology comes from MThinx Integrated Technologies, a Dallas-based company, Aun said. Michael Verseckes, a state transportation department spokesman, said the T and Keolis are testing the technology for free.

The mTHINX engineer told me that the pilot on coach 838 was also being done at no cost to the MBTA or Keolis. The idea was to prove (and refine) their technology with real-world passenger counting prior to any MBTA contract.

Their video cameras and hardware/software can actually identify a particular body, and count when & how many times & in what direction that body crosses a defined threshold - like the door of a train coach. (Note I'm saying "body" and not "person" because it doesn't make any personal identification. The software simply sees a blob of shape and color that looks like a human, but is distinct from another human.) There's a demo example: http://www.mthinx.com/?page_id=10286

According to the company engineer, the reason for having all the cameras in each coach is so that the movement of a passenger through the consist can be factored into the boarding/exiting counts. If the cameras "follow" a passenger between two or more coaches and the passenger remains inside the coach it's not going to register as an exit. Only if a passenger is tracked from inside the coach through the vestibule door and then does NOT show up in the adjacent coach would that be counted as an exit. Even more interesting, the mTHINX hardware includes their own GPS receivers. That allows them to factor in the geographic location and train speed when counting. For example if someone leaves the interior of one coach, does _not_ appear in the next coach, but the train is moving at speed and not near a station then the person didn't exit the train - they are simply hanging out in the vestibule.

What about train crew moving between the coaches, and on/off the train at stations? The software is even able to identify a standard crew uniform so that the crew boarding/exiting doesn't get counted as a passenger statistic!

In case you are concerned about all those video cameras being used for surveillance... the mTHINX engineer and several Keolis managers have told me that there is NO long-term video recording done by that equipment. The video from each camera is processed for passenger counts only and not saved for any long time. There is some on-demand short-term saving of video that mTHINX can do if they need to refine their counting / detection algorithms but I was told that the hardware on the coaches simply doesn't have the storage capacity for surveillance-type video recording. (That's subject to change of course.)

One more interesting thing about the equipment... each has a cellular data feed connection. The mTHINX engineer showed me how he was monitoring the passenger counts in real-time on his laptop. There could be a future use where overcrowding / underutilization could be addressed by dispatching equipment on-the-fly in direct response to demand. (I'm not holding my breath.)

I have lots of pictures of the cameras and other gear which I'll post soon.

(Edited to fix broken link)

[StefanW]

By the way, for reference the most recent comprehensive passenger count I could find was from the first half of calendar year 2012.

http://www.ctps.org/data/pdf/studies/transit/2012_MBTA_Commuter_Rail_Passenger_Counts/MBTA_Commuter_Rail_Passenger_Count_Results.pdf

Is there a more recent one with as much detail?

[StefanW]

It just occurred to me that even though mTHINX and Keolis people have said that the passenger-counting equipment isn't used for security / surveillance video recording, it's still possible that other equipment could be recording the video from those same cameras. It would be a bit odd if there were not video recording done as a security enhancement since (almost) every bus has cameras, and all the new Red/Orange/Green cars will too.