A multi-UAV cooperative route planning methodology for 3D fine-resolution building model reconstruction

Abstract In order to provide a fast multi-UAV cooperative data acquisition approach for 3D building model reconstruction in emergency management domain, a route planning methodology is proposed. A minimum image set including camera shooting positions and attitudes can be firstly obtained, with the given parameters describing the target building, UAVs, cameras, and image overlap requirements. A specific flight route network is then determined, and the optimal solution for multi-UAV data capture route planning is computed on the basis of constraint conditions such as the time frame, UAV battery endurance, and take-off and landing positions. Furthermore, field experiments with manual operating UAV mode, single UAV mode, and multi-UAV mode were conducted to compare the data collection and processing runtimes, as well as the quality of created 3D building models. According to the five defined LoDs of OGC CityGML 2.0 standard, the fine 3D building models conform to the LoD3. Comparison results demonstrate that our method is able to greatly enhance the efficiency of 3D reconstruction by improving the data collection speed while minimizing redundant image datasets, as well as to provide a normalized approach to assign the single or multi-UAV data acquisition tasks. The quality analysis of 3D models shows that the metric difference is less than 20 cm mean error with a standard deviation of 11 cm, which is fairly acceptable in emergency management study field. A 3D GIS-based software demo was also implemented to enable route planning, flight simulation, and data collection visualization.