Rail transport is the second fastest and the least cheapest land transit. What is even better is that it has the least carbon footprint. But these advantages significantly rely on the maintenance of thousands of kilometers of tracks. Rail track defects can lead to revenue loss and sometimes, even the loss of human lives.
In this paper, we present different types of rail track defects and the current inspection techniques that are in use. The paper then proposes a hybrid solution which can be deployed on in-service trains to scan the tracks on-the-go and keep the path safe for the following rolling stock.
The hybrid approach uses machine vision cameras and high-speed line scanners. Compute Vision is used to detect various surface and visible defects and classify them in severity categories as per industry standards. The analytics part of the application can predict defects based on the periodic scans and send area-wise alerts to maintenance departments.
Compute Vision can be extended to include Track Geometry and Maximum Moving Dimension Envelope scans.