Autonomous BIM-Aware UAV Path Planning for Construction Inspection

Abstract

What are the main findings? Highlights: A Building Information Model (BIM)-aware Integer Linear Programming (ILP) framework optimizes Unmanned Aerial Vehicles (UAV) camera networks by ensuring Industry Foundation Class (IFC)-based coverage while penalizing weak stereo geometry, Ground Sampling Distance GSD, and uncertainty. Coupling ILP optimization with Traveling Salesman Problem (TSP) routing yields geometry-adaptive, battery-feasible UAV trajectories that reduce camera usage by up to 63% and mission time by up to 50%. What is the implication of the main finding? The proposed framework enables autonomous and efficient UAV inspections directly linked to BIM and digital-twin workflows for QA/QC and progress monitoring. It advances autonomous UAV applications in digital construction and infrastructure inspection, contributing to scalable, repeatable, and data-driven remote sensing missions. Accurate 3D reconstructions of architecture, engineering, and construction AEC structures using UAV photogrammetry are often hindered by occlusions, excessive image overlaps, or insufficient coverage, leading to inefficient flight paths and extended mission durations. This work presents a BIM-aware, autonomous UAV trajectory generation framework wherein a compact, geometrically valid viewpoint network is first derived as a foundation for path planning. The network is optimized via Integer Linear Programming (ILP) to ensure coverage of IFC-modeled components while penalizing poor stereo geometry, GSD, and triangulation uncertainty. The resulting minimal network is then sequenced into a global path using a TSP solver and partitioned into battery-feasible epochs for operation on active construction sites. Evaluated on two synthetic and one real-world case study, the method produces autonomous UAV trajectories that are 31–63% more compact in camera usage, 17–35% shorter in path length, and 28–50% faster in execution time, without compromising coverage or reconstruction quality. The proposed integration of BIM modeling, ILP optimization, TSP sequencing, and endurance-aware partitioning enables the framework for digital-twin updates and QA/QC monitoring, accordingly, offering a unified, geometry-adaptive solution for autonomous UAV inspection and remote sensing.