SEPTA Platform-Track Incursion Detection

SEPTA Platform-Track Incursion Detection

Philadelphia, Pennsylvania
United States
SEPTA Platform-Track Incursion Detection

TYLin assessed technological alternatives for detecting unplanned people or objects on active SEPTA tracks near stations and alerting dispatchers and police.

The Southeastern Pennsylvania Transportation Authority (SEPTA) is vigilant about keeping people safe from oncoming trains entering train stations. Methods include installing over 4,000 closed-circuit television (CCTV) cameras throughout its properties to monitor, police, and record incidents, such as people unintentionally falling (or purposely jumping) off platforms onto active tracks.  

SEPTA contracted TYLin to assess possible cost-effective technological alternatives to installing very costly barriers or double-door systems. Technology solutions had to aid in rapid detection and alerts so that trains can be stopped safely before entering a station.  

The TYLin team researched a wide variety of possible technological approaches, such as treadle-pressure mats, laser beam-break, and radio frequency (RF) echolocation. Ultimately, we decided on a video analytics solution based on an infrared (IR) sensor immune to lighting and weather variations. This was the most economical, maintainable, and suitable solution for use when installed across SEPTA’s 290 stations. 

The IR sensor uses video footage and analyzes it using algorithms to detect and identify people or objects on train tracks. It offers several advantages over the other options studied. It provides a more complete and accurate detection capability by analyzing the visual information captured by the cameras. It eliminates the need for physical installations like mats or laser beams, which can be prone to wear and tear or environmental interference. Finally, the IR sensor's immunity to lighting and weather variations ensures consistent performance regardless of changing environmental conditions, enhancing reliability and minimizing false alarms.


Related Projects