The MTA is taking one other swipe at tapping into know-how that may detect when an individual, object or animal is about to finish up on the subway tracks.
The transportation authority is in search of a supplier to design, construct and take a look at a synthetic intelligence-supported observe intrusion detection system, in keeping with a contract discover it posted in April. The prototype would “consider efficiency below real-world situations” at an underground station, in addition to at one elevated cease. The stations haven’t but been decided.
About 6% of all subway delays final yr had been pinned on the presence of an individual or particles on the tracks.
Jamie Torres-Springer, president of MTA Development & Improvement, informed The Metropolis Reporter that the undertaking is akin to different high-tech advances throughout the subway system, together with the rollout of tap-and-go fare fee and the testing of latest fare gates.
“One of many large pushes we’re in the course of on this new capital program is the modernization of the system, acknowledging that we’ve acquired a 100-year-old system and we’re dutifully attempting to tack all these nice new applied sciences onto it to make it work higher,” he mentioned. “We’ve acquired a number of successes, nevertheless it’s very difficult to implement them.”
The most recent effort to check observe intrusion detection know-how comes after there have been 1,297 unauthorized entries on the tracks final yr — a 22% enhance from the 1,062 such incidents in 2019, MTA numbers reveal. An “unauthorized entry” is outlined by the MTA as an unauthorized individual coming into off-limits areas akin to tunnels.
That may embody individuals attempting to retrieve a dropped merchandise, being pushed down throughout a battle or assault, ending up on the tracks whereas below the affect of medicine or alcohol or trying suicide.
The figures additionally present there have been 491 observe intrusions over the primary 4 months of 2026, down barely from 505 in the identical interval final yr — and from a excessive of 537 from January by April of 2022.
The six-page solicitation outlines the enduring hope of growing subway-specific detectors that may spot individuals or objects above a sure measurement coming into the tracks deliberately or by accident, after a number of different methods had been examined at a pair of Manhattan stations from 2014 to 2019.
It’s a tall order. The MTA desires a system that’s “able to detecting pre-intrusion behaviors below high and low passenger density on the station platforms.”
In a transit system identified for the idiosyncrasies of some passengers, the methods would even be required to flag “erratic actions” and other people leaning into the tracks whereas producing a “well timed and applicable response” by alerts for prepare operators, station employees and management middle personnel.
The solicitation notes that the two-year take a look at at two stations would price between $10 million and $50 million.
Torres-Springer mentioned the MTA is aiming to award a contract for the undertaking by the top of this yr.
MTA officers mentioned that earlier exams yielded key takeaways in regards to the strengths and limitations of detection methods, whereas conceding that quickly advancing AI capabilities could possibly be an answer in a subway setting.
“We decided that the know-how labored to establish a observe intrusion, nevertheless it didn’t work to do it in a precise-enough method that we may handle how we reply to it,” Torres-Springer mentioned. “And now it’s obvious that AI know-how has advanced such that we must always be capable of do significantly better than that.”
Nonetheless, nonprofit the Surveillance Expertise Oversight Undertaking warned that the enlargement of cameras within the metropolis and using AI-driven tech could also be trigger for concern.
“As with all software that could possibly be used for surveillance and can broaden cameras throughout the town, transparency and reporting is vital to accountability,” mentioned William Owen, communications director for the native tech watchdog group. “MTA continues to check unproven AI on New York Metropolis Transit as an alternative of actual infrastructure and security enhancements to the subway.”
In an effort to discourage individuals from going onto the tracks, the MTA put in metal platform-edge limitations at greater than 100 stations by the top of 2025, and has introduced plans to have them in near 200 of its 472 stations by the top of this yr. The company had beforehand dedicated to testing platform doorways at three stations after straphanger Michelle Go was pushed to her dying in entrance of an R prepare in January 2022 on the Instances Sq.-Forty second Avenue station.
The observe intrusion take a look at would mark the most recent step for the MTA on a winding path to growing methods designed to chop down on security dangers to commuters, transit staff and anybody who results in the trail of a prepare.
“We’re engaged on one thing referred to as the station of the longer term, which is type of what a station within the New York Metropolis subway system goes to appear to be and the way it’s going to operate 20, 30, 40, 50 years from now,” Torres-Springer mentioned. “That is additionally a part of that, it turns into a part of the bundle of applied sciences and methods that we start to implement in future stations.”ever
“Something that may assist make the system safer for riders and can hold the trains shifting whereas lowering danger to staff is useful,” mentioned Lisa Daglian, government director of the Everlasting Residents Advisory Committee to the MTA. “The rising emphasis on newer know-how is an strategy that we’re glad could possibly be examined in two stations with two totally different environments.”
A 2022 request for data detailed how the earlier pilot tasks checked out closed circuit tv cameras mixed with lasers and video analytics; laser scanners with visible and infrared picture verification; thermal digital camera detection know-how and microwave scanners that would file entries onto the tracks.
Potential obstacles to deploying observe intrusion detection methods embody discovering methods to make sure that subway on-time efficiency shouldn’t be compromised by the amount of intrusion alarms, staffing to keep up the methods — and funding.
Celeste Kirkland, security director for Transport Employees Union Native 100, mentioned the know-how must match with the quirks of a subway system that opened in 1904.
“There are a number of challenges — we’ve acquired water intrusions, smoke, poor lighting, curved stations with restricted sightlines, crowded platforms,” Kirkland mentioned. “It sounds good, however how are they going to deal with the false positives, how are they going to tie it to the prevailing methods we’ve got?”
The detectors on Vancouver’s automated and driverless SkyTrain are typically triggered by the presence of birds or particles, a TransLink spokesperson informed The Metropolis Reporter in 2022, with staff then having to examine the realm earlier than service can resume safely.
Kirkland mentioned the potential tripping of alarms by birds or different animals would even be a priority.
“We now have rats all by the system,” she mentioned. “Would they need a prepare to cease mistakenly as a result of a rat jumped off a benchwall onto the tracks and is seen by a sensor? So, I believe they’ve quite a bit to do.”
The detectors are uncommon in U.S. subway stations, with the Chicago Transit Authority final yr saying plans to check camera-based capabilities at two rail stations. The Los Angeles County Metropolitan Transportation Authority beforehand examined the know-how at a platform at its Civic Middle cease.
Daglian, of PCAC, mentioned it’s “very encouraging” that the MTA is shifting forward with its plans.
“It’s why you pilot issues,” she mentioned. “AI and machine studying are nonetheless issues we have to study to belief.”
