OPTIMAL SETUP OF A TERRESTRIAL LASER SCANNER FOR CAPTURING PIPING OBJECTS IN PLANTS

Satoshi Kanai1, Shin-ya Yamanishi2 and Hiroaki Date

1) Dr. Eng., Prof., Division of Systems Science and Informatics, Hokkaido University, Sapporo, Japan.
2) MSc, Candidate, Division of Systems Science and Informatics, Hokkaido University, Sapporo, Japan.
3) Dr. Eng., Assoc. Prof., Division of Systems Science and Informatics, Hokkaido University, Sapporo, Japan.

Abstract: Recently, terrestrial laser scanning has been introduced to efficient 3D as-built modeling of piping systems of industrial plants. When capturing piping objects such as straight pipes or elbows using a laser scanner, due to their highly tangled placements, occlusions of back-side objects by the front-side ones are inevitable. Therefore, scans must be repeated from different setup positions. However, the quality of scanner placement still depends on the judgment of experienced operators, and there is no guarantee that these setups will enable efficient and sufficient coverage of the object surfaces for as-built modeling of piping objects. To solve the issue, we propose a computer-aided method to find a sequence of next-best view (NBV) scanner setup positions specialized for as-built modeling of piping objects with a terrestrial laser scanner. Our method differs from the previous NBV studies in that it seeks the optimum scanner position based on maximizing the observability of local unseen spaces in which a piping object is likely to be found. The process is based on the voxelization of the measured space and the evaluation of the observation probability between two voxels. The simulation shows that the method outperforms a conventional approach in recognition accuracy and efficiency.

Keywords: Laser scanning, Next-best view, View planning, Occlusion, As-built modeling.

Bibliographical Reference:
Satoshi Kanai, Shin-ya Yamanishi, and Hiroaki Date. “OPTIMAL SETUP OF A TERRESTRIAL LASER SCANNER FOR CAPTURING PIPING OBJECTS IN PLANTS.” In Proceedings of International Conference on Civil and Building Engineering Informatics (ICCBEI 2015), 109. Tokyo, Japan, 2015.

BACK