Rail Trespassing Prevention: Cameras Provide Unblinking Eye

Rail trespassing fatalities and injuries come with a unique pall of tragedy; one mixed with the heartbreak of loss and the frustration that these incidents are preventable.

The Transportation Safety Board of Canada reports there have been 34 trespassing incidents on Canadian railways between January and July 2022. The rate is pacing slower than the 64 incidents that occurred in 2021, but ahead of the incidents in 2020. In the United States, the Federal Railroad Administration (FRA) reports that trespassing on freight, intercity passenger and commuter railroad rights-of-way is the leading cause of all rail-related fatalities. FRA reports there were a total of 765 trespassing casualties on railroads in 2011; that number increased to 1,140 in 2021. Between January and May 2022, FRA statistics show commuter railroads have experienced 75 trespassing casualties versus 37 during the same months in 2021.

Time and resources have been invested to develop educational initiatives and engineering solutions while partnerships have enhanced enforcement, all with the goal of bringing these statistics down.

One tool with increasing significance in the fight against trespassing incidents is advancing camera technology. With the increased use of video in and around railroad property, cameras and camera vision systems are becoming important tools in amplifying safety messages and collecting data to improve safety programs.

‘Silent Witness’

Earlier this spring, Metrolinx released a video, slightly more than a minute long, showing a heart stopping near miss incident that took place on May 20, 2022, between a GO Train traveling on the Milton Line across the Humber River overpass and three young people who came too close to tragedy.

The video’s view is facing outward from the locomotive, showing the double track bridge with the train operating along the track on the left side of the screen and a non-occupied track to the right. The video, which Metrolinx has titled the “silent witness,” first shows a young person wearing a dark t-shirt and shorts standing on the right side of the screen holding the railing of the bridge. As the locomotive passes, the only reaction from the person at the railing is a slight turn of the head – the person is in and out of frame in a little less than three seconds. The video then captures a more unnerving sight: two additional young people, one in a white shirt behind and to the left of a companion wearing a bright blue t-shirt, running between the two sets of tracks. As the train – now well into emergency braking and sounding its horn – approaches the two running figures, the young person in the white shirt inexplicably crosses directly into the path of the train, glances over his shoulder and jumps over the rail and to the left railing barley clearing the front of the locomotive. The envelope of clearance was estimated to be 12 inches by Metrolinx. The third person in the blue shirt veers right and out of frame as the train passes.

In a blog post detailing the “silent witness” video, Metrolinx says the video serves as an appeal to “anyone working with young people to help the agency reinforce the dangers of walking on train tracks.”

Metrolinx has invested in rubber anti-trespassing mats at several locations on the Lakeshore East, Stouffville and Lakeshore West lines. Additionally, in known areas where public compliance with rail safety rules can be a challenge, such as the Port Credit bridge where people jump off the bridge into the river, Metrolinx has installed upgraded fencing and boosted engagement and enforcement efforts.

But the impact of the “silent witness” video is significant. Metrolinx Chief Communications Officer David Jang provided the board with a quarterly report in June 2022 in which he said there was an “overwhelming response from media worldwide” to the video. The communications report says the video attracted more than 23,000 views and a strong engagement rate on various social media platforms making it one of the top performing posts.

While the video’s engagement statistics are impressive, Metrolinx explains amplifying a message of safe conduct on or near train tracks is the goal.

“Safety campaigns, including the release of this ‘silent witness’ video, is one of the many ways Metrolinx works with the public to educate of the dangers of being on or near train tracks and to empower the public to make safe decisions,” explained James Wattie, Metrolinx senior advisor, media and issues, communications. “Every year, Metrolinx works with Operation Lifesaver Canada to promote Rail Safety Week (this year, Sept. 19-25) to stop track tragedies. Our Transit Safety Officers also conduct community rail safety outreach and partner with local police and fire services and other community safety groups to increase rail safety awareness. We continue to work in collaboration with municipalities to improve safety at our grade crossings.”

AI-Aided Crossing Trespassing Detection

As the commuter rail and rail transit sectors work to improve safety compliance around rail property, there is a growing understanding of where and how useful data can be collected. A group of researchers from Rutgers University is looking beyond injury and fatality data and has developed a way to account for all trespassing occurrences, including near miss incidents and events where there are no injuries. The Rutgers team is looking to improve understanding of trespassing trends by utilizing the vast amount of video railroads have and glean insight through artificial intelligence (AI).

The FRA-funded research uses an AI driven computer vision system that analyzes video and collects several key pieces of information on all trespassing events occurring at a given location.

“[We had a hypothesis] that there were many more of these types of trespassing events or violation events occurring, but they just weren’t recorded because no one is looking at this big video data and we also thought that AI could tackle this challenge,” Asim F. Zaman, project engineer at Rutgers, said during his presentation at the American Railway Engineering and Maintenance of Way Association annual meeting held in Denver, Colo., in August.

The initial research involved two grade-crossings in urban New Jersey and less urban Ashland, Va. Researchers set up a camera aimed at the grade crossing, determined a region of interest, which appears as a green bounding box in a video stream, selected signals to monitor and began AI monitoring.

The crossings were observed during a 13-month timeframe, which produced more than 38,000 hours of live AI analysis with information collected on time, date, the type of infraction, trajectory and several other points on more than 25,000 violation events.

From the data set produced, researchers were able to identify trends, which Zaman explains all tie into the “three E” approach for safety enhancement: engineering, education and enforcement.

For example, Zaman noted in his presentation that the New Jersey grade crossing saw an increase of trespassing violations on Thursday and Friday nights between 5:00 p.m. and 8:00 p.m. This information can help inform a railroad owner’s partners in law enforcement of when to patrol a certain crossing. The data also helped identify violation rates for different categories, such as cars, trucks, bicycles and pedestrians. In another example, Zaman says pedestrians at one crossing had a significantly higher rate of non-compliance with safety rules than other categories, which can help inform actions taken on an educational or engineering front. A third example Zaman pointed to was the information collected on where in the crossing activation cycle violations took place. At one of the observed crossings, most violations occurred as the gates were rising. However, Zaman pointed to a group of violations that took place when the crossing gates were horizontal. He noted this was a group to target with educational efforts as they were most at risk.

“You manage what you measure and if you’re only measuring fatalities then you’re only managing that,” explained Zaman. “[With the AI-aided system] you can manage to a much deeper level.”

In addition to the FRA funds, the team was awarded a grant through the Federal Transit Administration to develop the hardware elements of the system and received a Consolidated Rail Infrastructure and Safety Improvements (CRISI) in June 2022. The CRISI grant will see the Rutgers team, researchers from the State University of New Jersey, with support from Amtrak, the Louisiana Department of Transportation and Dover and Rockaway River Railroad, demonstrate the AI-aided monitoring system on five grade crossings in Connecticut, Massachusetts, New Jersey and Louisiana. The CRISI grant will see an enhanced version of the system deployed to review trespassing trends, understand the effectiveness of applied solutions and bolster future trespass prevention initiatives.

Zaman noted the research team is looking to work with new partners on future deployments.

----------------

Asim Zaman, P.E., can be reached at [email protected].