Russia Deploys Machine Vision to Boost Welding Precision in Heavy Industry

AI-driven visual systems are transforming high-precision welding operations across aerospace, nuclear, and engine manufacturing.

Researchers at the Progress Research and Technology Institute, part of the state-owned Rostec Corporation, have developed a machine vision system for electron beam welding. The new platform uses neural networks to enhance welding accuracy, streamline operations, and remove human error from critical manufacturing processes.

At the heart of the system is a magnetic optics module that delivers high-resolution imaging with refresh rates up to 200 frames per second. An electron beam scans the weld surface while a secondary electron sensor converts the data into a live video stream, allowing real-time control and analysis of the weld seam.

The system is being rolled out across a range of industrial sectors, including nuclear and aerospace engineering, shipbuilding, and engine production—industries where weld integrity is critical.

Machine vision is also being adopted for quality assurance. For instance, Rusal’s engineering technology center recently deployed a system that uses neural networks and cameras to monitor molten metal preparation in foundries. The AI provides real-time guidance to workers via a dedicated display, helping ensure correct alloy composition and reducing the risk of defective output.