AI-based autonomous track inspection

With annual revenues of nearly 15 billion US dollars in 2022 (approximately 14 billion euros) and a track network of 35,750 miles (approximately 57,534 kilometres), CSX is one of the top three US Class I railways both financially and in terms of network under management.

To ensure the safety and reliability of such a network, precise and reproducible inspections are required. They must provide the necessary details for informed infrastructure decisions.

To complement its existing inspection programme, CSX recently deployed Railmetrics' LRAIL autonomous inspection system on a fleet of enclosed freight cars, which allows inspection systems to be easily integrated into scheduled operations and provide around-the-clock monitoring of key routes.

Insights for all departments

The LRAIL systems are capable of continuously inspecting the track day and night while travelling at permitted speed and capturing 2D images, 3D scans and track geometry. The scan data is automatically processed on board according to CSX's business rules to identify track locations which require further investigation.The systems' inspection functions include: centreline mapping, tie/sleeper inventory and grading, rail joint connector and rail bolt inventory and counting, elastic rail fastener inventory and inspection, spiking inventory and inspection, tie plate inventory and inspection, anchor inventory and inspection, ballast cross section and surface fouling detection and more.

The system outputs both linear and georeferenced data and uses open data formats to enable organisation-wide use, including Geographic Information Systems (GIS), engineering and maintenance departments.