NS Launches AI Train Inspection Technology
Written by Carolina Worrell, Senior Editor
Screenshot Courtesy of NS
Norfolk Southern (NS) on Oct. 26 announced that it is deploying Digital Train Inspection Portals to enhance rail safety across the Class I railroad’s 22-state network.
According to NS, the portals feature cutting-edge Machine Vision Inspection technology developed in partnership with the Georgia Tech Research Institution (GTRI), which engineered the hardware, and NS’s Data Science/Artificial Intelligence (AI) and Mechanical teams, who “built the brains behind the program.”
The project, the Class I railroad says, “aims to supercharge NS’s safety infrastructure and inspection processes” with more than a dozen portals to be deployed by the end of 2024. NS leveraged GTRI’s expertise in advanced technology solutions, which the railroad says, “has already helped further national security and economic development.”
“We are a safe railroad, and we’re going above and beyond to become even safer,” said NS President and CEO Alan H. Shaw. “These new portals combine advanced technology with human expertise, giving our people and the public further confidence in Norfolk Southern’s safe operations. It’s all part of our promise to become the gold standard of safety in the rail industry.”
This end-to-end process, NS says, includes hardware, software, and people. First, the Digital Train Inspection Portals are equipped with an array of 24-megapixel trackside cameras and stadium lighting. Together, this Machine Vision Inspection technology captures ultra-high-resolution, 360-degree images of passing railcars. The cameras are synced to the microsecond, taking 1,000 images per rail car on average as they pass through the tunnel at speeds up to 70 miles per hour. The high-speed cameras are strategically placed at angles to capture things that are difficult to detect with the human eye during stationary inspections. In addition, capturing images while the train is in a dynamic state provides an inspection for various defects that cannot be done while the train is stationary, according to the Class I railroad.
AI analyzes these images for potential defects. According to NS, the railroad’s in-house Data Science/AI team has developed 38 advanced Deep Learning algorithms and has already deployed them across heavily trafficked lanes. “These best-in-class, field-proven algorithms have demonstrated very high accuracy levels, while having very low false-positives,” NS said. The AI transmits the information to NS’s Network Operations Center where the data is reviewed by subject-matter experts to “identify and address issues to proactively ensure the safety of rail operations.” Critical defects found are flagged for immediate handling.
The first portal was deployed in Leetonia, Ohio, where trains pass through approximately every hour.
