For the safe and steady decommissioning of the TEPCO’s Fukushima Daiichi Nuclear Power Station, especially from the viewpoint of the workers, it is crucial to develop a task plan based on a thorough understanding of the internal working environment situation through preliminary investigation. To achieve this, we conducted research and development on a photogrammetry method that reconstructs the structure and shape of the working environment from images into a three-dimensional (3D) model. The computational load of photogrammetry generally increases with the number of images.
 To solve this problem, we developed a method that efficiently generates a comprehensive model by integrating local 3D models generated from subimage sequences, where each sequence is divided into segments with fewer images (Fig. 1). Photogrammetry generates local 3D models with varying scales, positions, and orientations. To address this issue, we introduced an algorithm that integrates these models into a comprehensive model by adjusting their scales, positions, and orientations based on estimated camera trajectories obtained during model generation. In a comparative experiment (target space: 10 m × 10 m × 2 m, total number of images: 705, size of subimage sequence: 50, 100, 150, 200 images), we demonstrated that our proposed method can produce a comprehensive model of quality comparable to that obtained by processing all images simultaneously, but with greater efficiency (in reported case, time reduction of approximately 74 %). In future work, this method could be applied to generate spatial recognition support content for remote operations.

This work was supported by JAEA Nuclear Energy Science & Technology and Human Resource Development Project Grant Number JPJA19H19210047 "Human Resource Development Related to Remote Control Technology for Monitoring Inside RPV Pedestal during Retrieval of Fuel Debris."