Model-based next-best-view planning of terrestrial laser scanner for HVAC facility renovation

Recently, terrestrial laser scanners (TLSs) have been introduced to efficient as-built three-dimensional (3D) modeling in the heating, ventilating, and air-conditioning (HVAC) industry. When scanning the highly tangled installation of piping objects in HVAC systems using TLS, it is difficult to manually decide feasible scanner placement to scan the necessary regions of objects for construction with high accuracy and high point density and without any occlusion. Therefore, in this paper, we propose a model-based next-best-view planning method for TLS. The method uses a coarse 3D model generated by structure-from-motion (SfM), and it finds the near-optimum scanner placement that maximizes scan coverage while satisfying the constraints on the factors influencing scan quality such as beam incident angle, scan range, and scan overlap. The constraints are imposed based on user-specified scanning priority levels. Scanner placement is determined based on voxel-based space occupancy classification. The superiority of scanner placement by using the proposed method is verified through a comparative evaluation of the modeling accuracy with scanner placement by an experienced operator.