Intelligent aerial surveillance for safer railways using machine learning

Abstract

The integrity and usability of the rail systems are seriously compromised by problems including broken welds, unseen blockages, and non-functioning rails. Since they are primarily manual and offer no real-time information about what is happening, the present inspection methods are extremely labor-intensive, especially when it comes to remote and inaccessible places. The suggested system is a drone-based railway track surveillance system that can identify anomalies like fractures, welding flaws, and obstacles in real time. High-resolution camera drones and artificial intelligence (AI) models such as YOLO gather and evaluate data in a variety of environmental settings, and the problems they identify are geotagged. Resilient data transmission is ensured via a hybrid 4G/5G and LoRa network. Actionable insights and abnormalities are visualized and shown on a real-time dashboard. By accurately, scalably, and robustly observing the railway, the system increases maintenance efficiency.