UAVs represent one of the most relevant emerging technologies in the remote sensing domain of the last two decades, becoming a valid alternative to traditional acquisition techniques in a wide range of applications. However, the generation of image-derived 3D point clouds and the semantic understanding of the scene are usually generated after the acquisition flight, strongly limiting the applicability of UAVs in dynamic and complex contexts. Examples of UAVs able to generate 3D real-time semantic maps from UAV data are still in their very early stage. However, these solutions would promise to revolutionize the use of UAVs in many applications, such as mapping in emergency scenarios. 

The candidate will work at the crossroad of photogrammetric computer vision and deep learning to develop the next generation of real-time algorithms for the reliable generation of 3D semantic maps from UAV data. Both onboard processing and remote/cloud solutions will be considered for this task to deliver flexible solutions usable on both high-end and low-cost UAV platforms. Different operational scenarios will be considered during the PhD, with specific regard to times of emergency after disaster events. The candidate should have an MSc in geomatics, computer science or robotics, with expertise in deep learning and computer vision. Specific experience in C++ programming will be needed to develop the algorithms and embed them in the already existing DLR solutions.

The PhD will be developed at the German Aerospace Centre, Institute of Optical Sensor Systems – in Berlin (Germany). At the same time, you will be enrolled as PhD student at the University of Twente.

The PhD will be co-supervised by DLR researchers and academic staff of the University of Twente. The PhD degree will be awarded by the University of Twente (The Netherlands) within a collaboration framework between the UAV Centre of the University of Twente and the Department Security Research and Applications of DLR.