Artificial intelligence for the detection and characterization of wake vortices in LiDAR scans
Goal
The aim of the project is to use artificial intelligence (AI) to automatically detect and characterize wake vortices in LiDAR measurements in near real time.
Wake vortices are turbulent air structures generated by aircraft that can be dangerous for following aircraft. Minimum distances between landing aircraft are necessary for safe flight operations, but limit runway capacity. Real-time monitoring of wake vortices on final approach helps to reduce minimum separation distances and increase runway capacity. At present, there is a lack of methods to quickly detect and characterize wake vortices automatically and with a high degree of accuracy.
The aim of the project is to use artificial intelligence (AI) to automatically detect and characterize wake vortices in LiDAR measurements in near real time, to enable reliable error estimation and to use simulation data as a training data set. A reliable method for evaluating LiDAR measurements would be a major step towards dynamic separations, making aircraft landings safer, more efficient and ultimately more environmentally and climate-friendly.
The central activity is the application of suitable artificial neural networks (ANNs) to LiDAR scans. The characterizations of the ANNs should be comprehensible for air traffic controllers, which is achieved by integrating explainability components, correlations between the results of the ANN and wake vortex physics. In addition, the CNN should take into account the effect of weather on the transport and decay of wake vortices. Work is also being carried out on how simulation data can be used in conjunction with real data to improve the CNNs. In joint workshops with air traffic control companies, user requirements for the ANN algorithms are being collected.
Contact
Dr. Norman Wildmann
Head of Department
Institute of Atmospheric Physics
Applied Meteorology
Münchener Straße 20, 82234 Oberpfaffenhofen-Wessling
