Railway track health monitoring is one of the major tasks in railway inspection and monitoring system which is performed in order to maintain safety and security. The track-monitoring task involves inspection of various railroad components such as loose rail fasteners, defect in clips and switches, broken and misplaced crossties, cracks in various components of track and gauge measurement between the rails.
Due to the course of time, rail track component come across various defects like: loose rail fasteners, rail cracks, rail burns, misplaced crossties, broken crossties, a problem with the joints, and defect at switches as well as less visually evident defects like shifting from the mathematical model of track geometry over time. In particular, a common problem in the railroad industry is the tendency of rails to deviate from their proper gauge.
The existing system is expensive, time-consuming, involves human inspection, and automated vehicle-based system that needs proper track engagement for inspection. Researchers from IIT Roorkee is using computer vision for railroad component analysis to improve efficiency, objectivity and accuracy in the inspection system. This system helps to achieve cost-effective solutions with a higher level of performance, which is often unattainable through human inspection.
The inspection of rail track is done by applying Image processing, Computer Vision techniques on the images sent by drone. Images and generated data obtained from the drone is analysed which gives useful information about the health of the rail tracks.
Speaking about the technology, Dr. Dharmendra Singh, Coordinator, RailTel – IIT Roorkee Center of Excellence in Telecommunication, Professor in Department of Electronics & Communication Engineering, Microwave Imaging & Space Technology Application Lab, IIT Roorkee, said, “_Computer Vision Approach for the Drone data is a good alternative to monitor the railway track health in less time and it is also a very cost effective system. Some modules has already been developed which are giving quite satisfactory results and in some other modules like crack detection and all work is in progress and hopefully it will be completed soon”.
Objectives of the technology are:
- pre-processing of the data collected using a drone
- creation of reasonable, simple, and fast computer vision algorithm that is capable of processing the experimental field data and finding railroad defects reliably
- comparative evaluation of the performance of different algorithms and design schematics uncovering their better and worse features
- automated gauge Inspection through data provides by drone to see whether the gauge is constant throughout, and tracks are aligned or not
- find the localization of defect in a particular area using latitude and longitude
- to provide the GUI based approach for method and results
Automated computer vision mechanism for railway inspection system provides a fast, accurate and cost-effective way of detecting various anomalies present in the railway track.
Drone data has proven much more effective as it provides high-quality images that contain large information for monitoring and analysis. Inspection through drone does not require dedicated track for inspection, hence, it does not affect the smooth running of trains. The calculation of gauge gives the highly accurate results.