Named after the company’s founder, Dr Elmer Sperry, the system minimises the number of on-track verifications required, enhancing safety and lowering track exposure.
Elmer uses neural networks - a set of algorithms designed to recognise patterns - to identify the signature of cracks in scans of rails. The system currently processes more than 1600km of track per day and is built on the scalable Amazon Web Services (AWS) platform.
The system can direct analysts to the highest priority defects, which reduces the risk of error while also limiting the amount of time staff must spend lineside inspecting track.
Elmer is currently used to process all of Sperry’s non-stop fleet data in North America.
