New Tech to Watch: Automated Vehicle Occupancy Detection

Print Friendly, PDF & Email

EFF.orgJour­nal­ists and trans­paren­cy activists across the coun­try have done a phe­nom­e­nal job of shin­ing light on how local law enforce­ment agen­cies use emerg­ing tech­nolo­gies to sur­veil every­day peo­ple on a mas­sive scale.  It’s often like play­ing Whac-A-Mole and Go Fish at the same time. One day, the ques­tion may be whether police are using drones. The next, auto­mat­ic license plate read­ers.  After that, facial recog­ni­tion or IMSI catch­ers (i.e. Stingrays) or Rapid DNA analyzers.

So many tech­ni­cal terms, so many acronyms. Unfor­tu­nate­ly, we need to put yet anoth­er one your radar: Auto­mat­ed Vehi­cle Occu­pan­cy Detec­tion, also known as Auto­mat­ed Vehi­cle Pas­sen­ger Detec­tion or Auto­mat­ed Vehi­cle Occu­pan­cy Verification.

For years, gov­ern­ment agen­cies have chased tech­nolo­gies that would make it eas­i­er to ensure that vehi­cles in car­pool lanes are actu­al­ly car­ry­ing mul­ti­ple pas­sen­gers. Per­haps the only rea­son these sys­tems haven’t gar­nered much atten­tion is that they haven’t been par­tic­u­lar­ly effec­tive or accu­rate, as UC Berke­ley researchers not­ed in a 2011 report.

Now, an agency in San Diego, Calif. believes it may have found the answer: the Auto­mat­ed Vehi­cle Pas­sen­ger Detec­tion sys­tem devel­oped by Xerox.

The San Diego Asso­ci­a­tion of Gov­ern­ments (SANDAG), a gov­ern­ment umbrel­la group that devel­ops trans­porta­tion and pub­lic safe­ty ini­tia­tives across the San Diego Coun­ty region, esti­mates that 15% of dri­vers in High Occu­pan­cy Vehi­cle (HOV) lanes aren’t sup­posed to be there. After com­ing up short with ear­li­er exper­i­men­tal projects, the agency is now test­ing a brand new tech­nol­o­gy to crack down on car­pool-lane scofflaws on the I‑15 freeway.

Doc­u­ments obtained by CBS 8 reporter David Got­fred­son show that Xerox­’s sys­tem uses two cam­eras to cap­ture the front and side views of a car’s inte­ri­or. Then “video ana­lyt­ics” and “geo­met­ric algo­rithms” are used to detect whether the seats are occupied.

When the detec­tion system’s com­put­er deter­mines a dri­ver is improp­er­ly trav­el­ing in the car­pool lane, the cam­eras instant­ly send pho­tos of the car’s inte­ri­or and its license plate to the Cal­i­for­nia High­way Patrol.

In short: the tech­nol­o­gy is look­ing at your image, the image of the peo­ple you’re with, your loca­tion, and your license plate. (SANDAG told CBS the sys­tems will not be stor­ing license plate data dur­ing the tri­al phase and the sys­tem will, at least for now, auto­mat­i­cal­ly redact images of dri­vers and pas­sen­gers. Xerox’s soft­ware, how­ev­er, allows police the option of using a weak­er form of redac­tion that can be reversed on request.)

Xerox­’s Auto­mat­ed Vehi­cle Pas­sen­ger Detec­tion sys­tems can be mount­ed in per­ma­nent loca­tions, such as free­way gantries, or attached to mobile trail­ers that “can be moved around, in order to keep poten­tial vio­la­tors honest.“

Xerox claims that the sys­tems have a 95–99% accu­ra­cy rate, even with vehi­cles trav­el­ing as quick­ly as 100 mph. That rep­re­sents a sig­nif­i­cant leap in capa­bil­i­ties com­pared to the 14–20% accu­ra­cy record­ed by sim­i­lar tech­nol­o­gy test­ed by SANDAG four years earlier.

So far, Xerox’s tech­nol­o­gy has only had one oth­er tri­al run: a 2013 test on the Mack­ay Bridge in Hal­i­fax, Nova Sco­tia, where the detec­tion sys­tem cap­tured 250,000 images of dri­vers in just under two weeks. Xerox claims the tech­nol­o­gy suc­cess­ful­ly deter­mined front seat pas­sen­gers 98.9% of the time and rear seat pas­sen­gers 96.2% of the time.

That suc­cess rate may seem impres­sive, but it still means that, in aggre­gate, thou­sands of peo­ple could have been affect­ed by machine error. Auto­mat­ed Vehi­cle Pas­sen­ger Detec­tion sys­tems also rais­es the usu­al ques­tions and con­cerns about pri­va­cy and mass sur­veil­lance, par­tic­u­lar­ly with­out rules yet in place defin­ing the lim­its of how this tech­nol­o­gy col­lects data and how that data will be stored, accessed, and destroyed.

If this tri­al is suc­cess­ful, it may only be a mat­ter of time before Xerox begins mar­ket­ing this new tech­nol­o­gy to oth­er juris­dic­tions. As Xerox told its investors in Novem­ber, its Auto­mat­ed Vehi­cle Pas­sen­ger Detec­tion sys­tem is “aimed at rev­o­lu­tion­iz­ing the move­ment of peo­ple and goods worldwide.”

Whether you’re a local jour­nal­ist or cit­i­zen watch­dog, add Auto­mat­ed Vehi­cle Occupancy/Passenger Detec­tion to your vocab­u­lary. In the com­ing months and years, look for it in pro­cure­ment doc­u­ments and com­mit­tee agen­das. Ask for relat­ed doc­u­ments in your rou­tine pub­lic records requests. As with all street-lev­el sur­veil­lance tech­nol­o­gy, cit­i­zens have more pow­er to influ­ence pol­i­cy when they dis­cov­er the tech in its infan­cy, rather than fight­ing to dial back its use after it has been inte­grat­ed into every­day police operations.