Perm Polytechnic presented the first system for automatic diagnostics of railway bridge defects

Russian engineers have created a system that automatically detects damage to railway bridges during drone overflight and immediately displays them on a digital drawing. The technology allows to abandon lengthy manual analysis of video recordings and significantly speeds up the search for dangerous defects. The project was authored by specialists from Perm Polytechnic.

The new development combines an unmanned aerial vehicle, a computer vision system, and a software complex for automatic linking of detected damage to specific structural elements. Previously, even after using drones, specialists had to re-search for the location of a detected crack or corrosion on engineering documentation, which significantly increased inspection time.

The complex is based on a neural network trained to recognize cracks, traces of corrosion, and various deformations from thousands of images. During autonomous flight, the drone simultaneously analyzes the video stream and records the coordinates of each detected defect. Then, the software automatically correlates this data with the bridge's geometry and marks the digital drawing with the exact section of the structure.

The inspection route is set by the operator in advance via a ground control station. After that, the drone performs the flight independently, and artificial intelligence analyzes the incoming image in real time. Upon detection of damage, the operator confirms the result, after which the system instantly saves the coordinates and displays the defect location on the engineering diagram.

The developers expect that in the future, the technology will be used not only for monitoring the condition of railway bridges. It is also planned to adapt it for diagnosing tunnels, overpasses, elevated roads, and other transport infrastructure facilities.