移動ロボットによる建造物の3次元形状計測のための視点計画

抄録

3D scanning of entire structures is performed by integrating measurement results acquired from multiple viewpoints. However, in many cases, the measurement results are not complete because viewpoints are determined empirically. In this paper, we propose a viewpoint planning method for 3D scanning of structures by a mobile robot. Our method utilizes the ray casting to detect the visibility of unobserved regions from each candidate viewpoint and then repeatedly selects the best viewpoints. In addition, we adopt a coarse-to-fine search in order to realize an efficient viewpoint planning. We quantitatively evaluated our method by simulation experiments to show the effectiveness of our method. In addition, we carried out an experiment in an outdoor environment to show that our method can operate in a real environment.