Project ArchEyeAutomatic - Model-based Optimization of Multirotor UAV Trajectories for Automatic and Complete Three-Dimensional Photogrammetric Documentation of Buildings in Art History and Archaeology (original) (raw)
Due to demands of precision of 3D documentation and the analysis of such data in historic building research and archaeology, we present the recently started project "ArchEyeAutomatic" for an automated documentation of historical buildings using Unmanned Aerial Vehicles (UAVs). This is realized by an optimized calculation of a flight trajectory. Financial support is gratefully granted by the German Excellence Initiative within the institutional strategy of the University of Heidelberg and its special funds for Twinning projects. The previous work consists of two parts. First, the UAV-project "ArchEye", which is used to document excavation areas using a hexa- and octocopter. The software from this project can calculate 2D trajectories for a given ground-resolution by camera-resolution, focal-length and the flight's altitude. The result is a high-resolution orthogonal image which can easily be rectified for further use in a Geographic Information System (GIS). The second part of the previous work is the photogrammetric Stucture-From-Motion approach. By using this method it is possible to create detailed 3D models of objects just by taking photos in a particular way. The resulting 3D models are useful for an extended documentation in Historic Building Research as well as in Archaeology. With additional software for visualization and analysis like GigaMesh, new insights can be achieved. To implement an automated acquisition of a building in 3D by an UAV, research in different disciplines is necessary, in this case the specific documentation-methods, robotics, optimization, computer vision and scientific computing. To acquire the UAV's environment as well as the object to be modeled for the calculation of the trajectory, optical methods like photogrammetry or stereo vision will be applied. The UAV will be represented by a mathematic model. Using these models, the trajectory can be calculated and optimized, solving an Optimal Control Problem in real-time to control the UAV with constraints given by the environment, like obstacles or wind, as well as the UAV but also the 3D model's quality. This is challenging, because the aspect of the model's quality, which is clearly visible for the expert, has to be expressed by a mathematic term. This term is used for optimization of the in real-time calculated next movements for the UAV. Preliminary results from the project "ArchEye" and some photogrammetric models as well as the plans for the project are shown.