OTC-DaTA2Model – Development of an AI Foundation Model and Digital Twin for Maritime Applications

A collaborative project to strengthen digital innovation in the maritime sector

The OTC-DaTA2Model project aims to develop a generic AI foundation model and a modular digital twin for the maritime sector. Led by the Fraunhofer IGD in Rostock and in close collaboration with leading partners from academia and industry, the project addresses key challenges in the efficient use and analysis of heterogeneous underwater sensor data. The results will form the basis for innovative applications, ranging from object detection to the sustainable monitoring of maritime infrastructure, and will support the digital transformation of the Ocean Technology Campus Rostock.

Project Description

Objectives and Technical Focus

OTC-DaTA2Model aims to provide an AI foundation model capable of processing multimodal data from numerous object classes in maritime environments and adapting flexibly to specific tasks. In parallel, a modular, probabilistic digital twin is being developed that links sensor data with background knowledge, quantifies uncertainties, and derives the current and future states of maritime objects. Key technical aspects include the development of a scalable software architecture, the integration of representation learning methods, and the implementation of a robust error model for sensor data.

Challenges and Solution Approaches

The maritime environment places high demands on data quality, model adaptability, and real-time performance. OTC-DaTA2Model addresses these challenges through cluster-wide data harmonization, the use of pretrained models for object detection and segmentation, and the development of methods for fusing and evaluating heterogeneous sensor data. The integration of edge-device inference and the use of generative models for data augmentation enable efficient deployment even under limited resource conditions.

Importance for the Future of Maritime Technologies

OTC-DaTA2Model will create innovative digital tools that pave the way for data-driven business models and sustainable technologies in the maritime sector. The developed foundation models and digital twins promote the integration of AI and simulation in maritime research and industry, strengthening the Ocean Technology Campus Rostock as a leading innovation hub.

© Fraunhofer IGD
Overview of the content components of the maritime digital twin

Application Examples and Benefits

  • The project results target port and wind farm operators, logistics companies, and providers of maritime surveillance solutions. Potential application areas include the assessment and prediction of underwater infrastructure conditions, optimization of measurement campaigns, automation of inspection processes, and long-term habitat monitoring. The AI-based methods reduce the effort required for data annotation and analysis while increasing the efficiency and reliability of maritime operations.

Project Partners

  • Fraunhofer IGD Rostock
  • PLANET AI GmbH
  • EvoLogics GmbH
  • GEO-DV GmbH
  • Subsea Europe Services GmbH
  • Leibniz Institute for Baltic Sea Research Warnemünde (IOW)
  • University of Rostock
  • Associated partners: Rostock Cargo and Fishing Port GmbH, UHH Umschlags- und Handelsgesellschaft Haldensleben mbH

The strong networking within the cluster enables the development and validation of practical use cases and ensures sustainable exploitation of the project results.

Industry

  • Maritime Industry