IIMEO is mainly focusing on railway failures caused by extreme weather situations (e.g. landslide, flood, fallen trees etc.). In this post, we discuss the pilot use case of Serbian Railway Infrastructure Monitoring with Serbian Railway as the end user. Reliable transport infrastructure is crucial for economic prosperity, with road and rail being the main land transport modes, with railways being generally safer. However, many accidents still occur each year due to derailments and collisions with obstacles on or near the railway tracks.

Common obstacles includes people trespassing the railway tracks, maintenance equipment left on the tracks. Furthermore hazardous railway obstacles often result from geological and natural hazards like landslides, flooding, and extreme weather. These can damage or collapse railway bridges and block tracks with fallen trees.​ Besides collisions with obstacles, derailments can also occur due to track deformations from earthquakes or extreme heat. Hot weather can cause rails to expand, bend, or break, known as „buckling rails.“

To handle extreme weather events, the infrastructure monitoring service needs to be developed and validated to offer obstacle detection by using state-of-the-art anomaly and change detection algorithms to identify obstacles, flag anomalies, and compare new measurements with previous healthy ones. The targeted static objects are all static objects found on or near the tracks that are not part of the railway infrastructure. We have elaborated the following key requirements, which are of high significance for developing a service suited to the needs of our Serbian Railway use:

• The intended service shall be able to provide information on-demand, continuously or autonomously based on user defined triggers. Rationale: This flexibility ensures, that the operator has access to critical information when needed (on-demand), consistently (continuous monitoring), or automatically when specific events occur (autonomous). This supports various operational scenarios such as scheduled maintenance, real-time monitoring, and emergency response.
• Service users should be able to define conditions or events which will start an autonomous service.
  Rationale: Autonomous monitoring based on predefined triggers (e.g., weather alerts, seismic activity) enhances early detection and minimizes the risk of accidents by reacting to environmental and infrastructure changes in real time without manual intervention.
• The intended service shall provide information about changes in railway environment with a delay less than 60 min. Rationale: Rapid response is critical in railway operations to prevent accidents or delays. A delay of less than 60 minutes ensures timely alerts, enabling the operator to take preventive or corrective action swiftly.
• Sub-meter accuracy for the change detection in railway environment shall be provided. Rationale: High-precision monitoring is crucial for detecting small but significant changes, such as track deformation or minor landslides, which could pose serious safety hazards if left unchecked.
• It shall be possible to define change detection along a linear feature with the predefined width of 12 m, which corresponds to the width of railway track with sidewalk. Rationale: The defined area (12 meters) ensures the system focuses specifically on the railway tracks and nearby areas critical to railway safety, optimizing resources and minimizing false alerts from irrelevant changes outside the critical zone.
• The user shall have the possibility to define a fixed area of interest for change detection in the user interface.
  Rationale: This feature allows the system to monitor specific areas of concern, such as high-risk regions (e.g., tunnels, bridges) or high-priority railway lines, ensuring focused and efficient monitoring of the most vulnerable parts of the infrastructure.

In our next blog post, we will explore in more detail the current development activities of this service suited to our pilot use case. Stay tuned!