Ph.d.-forsvar Eigil Yuichi Hyldgaard Lippert

Principal supervisor

Senior researcher Henning Heiselberg

Co-supervisor

Professor Ole Baltazar Andersen

Examiners

Professor John Peter Merryman Boncori, DTU Space (chair)
Dr. Ernst Krogager, Danish Ministry of Defense Acquisition & Logistic Organisation, Denmark
Professor Victor Lobo, Portuguese Naval Academy, Lisbon, Portugal

Chairperson at defence

Head of GEO, Michael Schultz Rasmussen

Summary

The maritime domain plays an important role in global trade, security, and environmental sustainability. However, approximately 20% of large Ships operate without transmitting their locations, making them dark ships. Although not all dark ships are involved in illegal activities, most ships engaged in activities such as illegal fishing, smuggling, and threats to national security are dark ships. Improving technologies to monitor these dark ships is important, as effective maritime surveillance is essential for addressing such issues, ensuring safe navigation, and protecting marine resources and borders.

This PhD thesis researches and develops methods to enhance maritime surveillance by integrating advanced satellite data and artificial intelligence (AI). The main goal is to research and implement techniques for detecting and tracking dark ships. The research focuses on different areas including ship detection, ship feature extraction, trajectory prediction, and data fusion.

The study begins with ship detection, where advanced AI models are developed to identify ships in satellite data and differentiate them from other objects, like icebergs. This is followed by research on ship feature extraction, which provides detailed information about each ship, such as its size and behaviour. This can in turn, help to identify potentially illegal dark ships. The study then focuses on trajectory prediction, implemented through an additional AI model, which uses historical data to predict future ship trajectories, enabling satellites to be tasked with monitoring specific areas of interest.
Lastly, the research combines data from multiple satellite sources, including Synthetic Aperture Radar, Radio Frequency, and Automatic Identification Systems, to enhance the overall accuracy of maritime surveillance. By fusing these data sources, this PhD thesis provides a more comprehensive view of ship activity and improves the ability to track dark ships.

This PhD thesis contributes to enhancing maritime domain awareness by offering practical methods that can potentially support and improve global maritime security, contribute to detecting illegal activity, and protecting national borders. While the thesis focuses on developing AI models with different objectives, the dissertation also places the research in the larger context of maritime domain awareness. Additionally, these techniques may have broader applications in areas such as environmental monitoring and disaster response.