This Train TRANSFORMED Britain’s Railways | #Railnatter 302

Why It Matters

By automating track inspection, the PLPR train reduces fatal human error, enhances safety, and lowers maintenance costs, positioning Britain’s railways at the forefront of data‑driven infrastructure management.

Key Takeaways

• PLPR train automates track inspection, reducing human error
• Grayrigg derailment highlighted flaws in manual inspection processes
• NMT “yellow fleet” pioneered visual data capture for rail safety
• Omnicom’s pattern‑recognition system integrates imaging with real‑time positioning
• Reliable data capture remains critical despite AI hype in rail industry

Summary

The episode of Railnatter focuses on the Plain Line Pattern Recognition (PLPR) train, a 20‑year‑old measurement unit that has become a cornerstone of Britain’s rail‑track inspection regime. Host Gareth and guest Alex, a veteran Network Rail engineer, explain why this “yellow fleet” vehicle represents a shift from manual visual checks to systematic, data‑driven monitoring.

The discussion traces inspection from the early days of inspectors with hammers and percussive tests to the modern need for repeatable, auditable results. The 2007 Grayrigg derailment, caused by an uninspected defect that had been missed by human crews, is cited as the catalyst for adopting automated pattern‑recognition technology. The NMT (Network Measurement Train) and its French‑sourced “Ivoire” system captured images of a loose stretcher bar two days before the accident, proving the value of continuous imaging.

Alex highlights the technical challenges of capturing usable images on a moving train—mud, lighting, vibration, and precise geolocation—all of which required bespoke hardware and software from small firms like Omnicom (later Balfour Beatty). He notes that the PLPR workflow turns raw photos into actionable reports for track section managers, and that the Pendolino’s robust crash performance in the same incident underscored the importance of both rolling stock design and track integrity.

The broader implication is a railway network that can proactively identify defects, limit track‑worker exposure, and integrate machine‑learning analytics without relying on “AI hype.” As Network Rail upgrades its yellow fleet and invests in higher‑resolution sensors, the PLPR model sets a benchmark for other infrastructure sectors seeking to replace hazardous manual inspections with reliable, automated solutions.

Original Description

This week we are talking about Network Rail's PLPR train - the incredible and now pretty experienced train that put machine learning into practice in a way that makes today's AI boosters look more like hucksters than they already are. Alex Cusdin joins us to talk through the train and its tech, and tell the story of how it came into being.

Capturing an incredible 18TB of data a day, this train captures 70,000 photos a second to enable a very high resolution view of the track to automate and improve inspection, in the process making railways safer. This is an example, now well proven, of how technologies can be applied to make better use of skilled people, rather than displacing them. And it's about to be replaced.

Support #Railnatter at https://patreon.com/garethdennis. Merch at https://merch.railnatter.uk. Join in the discussion at https://discord.railnatter.uk. You can also buy my book #HowTheRailwaysWillFixTheFuture: https://bit.ly/HowTheRailways