The safety of landing aircraft is a critical concern, and the wake vortex generated by preceding aircraft during landing can pose a significant threat to the succeeding air-crafts. The safety and efficiency of landing aircraft heavily rely on the judgments of air traffic controllers. In order to improve safety, airport capacity, and environmental sustainability, more accurate, efficient, and transparent tools are needed.
In previous studies, many deep neural networks have been applied in order to automatically detect the wake vortex , and they have achieved intriguing results. However, deep neural networks suffer from the problem of lacking transparency, as their decision-making mechanism is not human-interpretable. Therefore, they are not trustworthy enough in safety-critical tasks. For the current thesis, we decide to apply state-of-the-art algorithms in Explainable AI to improve the transparency of deep neural networks and to develop trustworthy assisting tools for air traffic controllers.
Ausschreibung: Download (pdf)