UAV-SRGK

In cooperation with SPEKTER GmbH and the Dresden University of Technology, we have started a project project on the "Development of a procedure for automated fusion and classification of ALS and UAV data for the generation of heavy rain hazard maps". The aim of the ZIM-Project (Central Innovation Program for SMEs) is to develop a procedure for the automated generation of heavy rain hazard maps.

Fraunhofer IOSB-INA is responsible for the development of methods for automated structural analysis based on UAV image data. For this purpose, an annotated image database is built and an AI algorithm is trained. The goal is the automated detection of structures relevant for the generation of heavy rainfall hazard maps. First, image data will be collected in a work package and an image database will be built. Then, annotation types and classes are defined and the images are annotated. In another work package, an AI algorithm is trained to perform the structural analysis automatically. After verifying the results of the neural network for structure analysis, the network will be integrated into the overall system. For this purpose, a software application for inference of the trained neural network on UAV image data has to be developed. Subsequently, a procedure for the integration of the segmented images into the 3D model has to be realized.

SPEKTER GmbH is responsible for the creation of hydrodynamic 2D simulation models and the generation of piping models for HN models. Furthermore, SPEKTER performs comparative simulations accompanying the research work of the Fraunhofer Institute and the TU Dresden in order to provide continuous feedback on the achievement of the target criteria.

The Technical University of Dresden is responsible for the development of methods for the automated fusion of ALS and UAV data and for the creation of digital terrain models. In addition, TU Dresden is to develop a method for deriving land use and roughness occupancies and provide output data for determining the runoff coefficient for the HN models.

The project is funded by the Central Innovation Program for SMEs (ZIM), a nationwide funding program that is open to all technologies and industries.