Track 1: Safety and workability

Track 1 focuses on increasing safety and workability in offshore operations by developing machine learning models. Specifically, track lead Daan van der Made wants to develop models to forecast workability and motion sickness. The track is split into two parts: one that focuses on near real-time predictions (minutes ahead) and one that focuses on short-term forecasts (hours ahead). Currently, both sides of the track are finalizing their project plans and beginning work on the first iterations of the machine learning models.

The near real-time development focuses on assessing the current vessel operation, environmental conditions and resulting performance metrics from readily available measurements. For example, the relation between ship motions and sea sickness occurrence can be evaluated from the combination of existing sea sickness models and reported sea sickness. Also, there is existing work on finding the wave conditions from ship motions with relatively short measurements, giving a near real-time view of the current sea state. The resulting assessments can be used to explore adjacent operation modes that would improve on the current operation. For longer term advice based on measurements in the current conditions, the extraction of configuration parameters, like gm or RAOs, can be used with known upcoming conditions or specific operational needs to find the best route, mooring heading, or actuator settings.

The short-term forecast side of the track is developing a machine learning model to predict MSI (Motion Sickness Incidence). We chose to focus on CTVs (Crew Transfer Vessels) because there is a strong use case for these vessels. CTVs make daily trips to the wind farm and back. If technicians get seasick on the way there, they may be forced to return to port without completing any work. With an MSI forecast, an informed decision can be made in the morning about whether it is worthwhile for the vessel to sail. Unnecessary sail-outs can be prevented, saving fuel and costs, reducing emissions, and preventing seasickness for technicians and crew.

Track 2: Emission reduction

Track lead Prof. dr. ing. Coraddu Andrea addresses one of the most urgent challenges facing the maritime sector today: reducing emissions through improved fuel efficiency under real operational conditions. By leveraging onboard data streams, we are developing robust, data-driven models capable of predicting and optimizing vessel fuel consumption while accounting for variable weather, sea state, and the evolving impact of hull fouling. These tools will support real-time operator guidance and advanced routing strategies, supporting safer and greener maritime operations.

Consortium and Participants
Track 2 brings together a strong consortium representing industrial, research, and technological expertise:

• Industry Partners: Spliethoff, Damen, RH Marine, HMC, Ocean AI
• Research Partners: TU Delft, MARIN

Each partner contributes essential datasets, domain knowledge, and validation environments, ensuring that the solutions we build are realistic, scalable, and aligned with industrial needs.

Milestones and Activities
We are currently advancing several core developments:

• Predictive models for vessel fuel consumption, trained on real operational data and enriched with AI-based performance monitoring techniques.
• A modular toolkit for emission reduction and hull-fouling diagnostics, suitable for deployment both onboard vessels and in shore-based control centers.
• Collaborative validation and multi-platform demonstrators, drawing on partners’ operational case studies and technical experience.

Recent Progress and Next Steps
After establishing the foundational architecture of our technical approach, recent efforts have focused on harmonizing datasets across the consortium, establishing consistent data protocols, and preparing the environment for coordinated model development and testing. The next phase will involve iterative model integration and validation across representative vessel classes.

Recruitment Update
We are also pleased to share that the Postdoctoral Researcher position supporting this track is about to be open. We kindly ask partners to circulate this opportunity within their networks.

Legal and ethical aspects of AI

Servicing all tracks within the lab Lenneke Sprik focuses in her Professional Doctorate (PD) research on the legal and ethical aspects of AI use and data sharing in the maritime sector. In her research, she explores how the maritime sector can design AI systems that are in compliance with relatively new European regulations, such as the AI Act and the Data Act. The development of these systems also comes with questions regarding data governance and the ethical values that should be taken into account in order to be supported by their users and the public. The security of generated data is another pressing issue in relation to AI that will be addressed in this PD project.

Students play an important role in this research. Two graduating students from the Law programme (HBO Rechten) are currently contributing through their final theses: one examines data ownership and data sharing practices in light of the EU Data Act, while the other focuses on certification and regulatory requirements for AI systems used on board of vessels.

Although not directly part of the Maritime AI Innovation Lab, a related student project involving four law students investigates liability in collision cases involving autonomous ships. These outcomes will will be of use for future tracks within the Maritime AI Innovation Lab, dealing among others with autonomous shipping.

AI is everywhere: in healthcare, the classroom, the factory, traffic, the courtroom, and the military. It is rapidly changing how we work, learn, and live together. The question is not if AI will change our society, but how. And what you, as a professional, can do to help shape these developments.

On Tuesday, April 14, 2026, at De Fabrique in Utrecht, the Dutch AI Congress will bring together government, knowledge institutions, businesses, and start-ups to answer that question.