This is a remote visual inspection system under development that uses robust imaging* technology and AI to reduce inspection time and improve work efficiency during inspections inside the reactor pressure vessel. The system enables inspections in complex structures and confined spaces that were previously difficult to access, reducing inspector workload and improving inspection accuracy.
Images resistant to defocusing are generated using a phase plate and restored through image processing, expanding the camera’s depth of field. This relaxes constraints on camera position and angle, making it easier to capture images in complex shapes and confined areas.
Lighting that optimizes the balance between reflected and diffused light is used to clearly highlight cracks and other anomalies.
AI learns the characteristics of cracks using historical visual inspection images. During inspections, trained models automatically identify potential crack locations. This supports inspectors’ judgment and enables shorter inspection times compared with conventional visual inspections. We are advancing the development of systems incorporating these technologies with the aim of applying them to actual nuclear power plants.
Robust Imaging and AI-based Remote Visual Inspection System for Reactor Internals
