Bridging Photogrammetry and Building Energy Modeling: A UAV-Based Framework

Abstract

Scalable workflows for acquiring as-built building data suitable for energy modeling remain scarce, despite buildings accounting for approximately 40% of global energy consumption. This paper presents a UAV-based multi-sensor workflow that combines RGB photogrammetry and LiDAR scanning to document existing building geometry. The workflow begins with aerial mission planning and the acquisition of complementary RGB imagery and LiDAR measurements. The RGB images are processed using Structure-from-Motion and MultiView Stereo, while the LiDAR measurements are independently processed into a georeferenced point cloud. The resulting threedimensional datasets are prepared for subsequent Building Information Modeling and energy simulation. The workflow was demonstrated on a mid-rise academic building at the University of Waterloo. During the acquisition mission, the UAV captured 522 high-resolution RGB images and recorded 2.62 million LiDAR points. Flight altitude remained stable within $\pm 0.4 ~\mathrm{m}$, while the RGB- and LiDAR-derived geometries achieved mean absolute errors of 9.34 cm and 7.20 cm, respectively. The BIM contained 115 spaces across eight functional categories, and lighting and equipment represented 51% of simulated annual energy use.