Speaker
Description
This presentation will explore the intersection of neural networks and differential programming in addressing critical challenges within the maritime domain. The presentation will begin with an overview of key issues facing the sector, followed by an overview of research conducted at the DLR Institute for the Protection of Maritime Infrastructures where research using differentiable methods plays a central role. Highlighted topics will include ongoing work in cosmic-ray tomography and optical reconstruction of dynamic scenes, with additional insights into applications in underwater acoustics and maritime robotics. The talk will also propose directions for future research, emphasizing the potential for unified, deep-learning-based frameworks capable of integrating and analyzing diverse maritime sensor data. Finally, the application of reinforcement learning within such a unified framework will be discussed as a promising avenue for advancing maritime situational awareness.
