Rahmenwerk für die Klassifikation von autonomen Schiffen
Dr. Urs Vogler, Dr.-Ing. Pierre Sames, DNV SE; Dr. Kristian B. Karolius, Are Jørgensen, DNV AS
Maritime Autonomous Surface Ships (MASS) have evolved from early experimental initiatives into practical solutions featuring progressively higher levels of autonomy. Yet their safe and scalable deployment continues to be challenged by the lack of detailed guidance, technical requirements, and standardized verification pathways for autonomous and remotely operated ship functions. While the International Maritime Organization (IMO) continues developing its goal‑based MASS code, early implementations have relied on case‑specific Technology Qualification (TQ) processes and broad risk‑based guidelines. This has resulted in inconsistent interpretations of safety expectations and considerable engineering overhead. To help bridge these gaps, DNV have developed the Autonomous and Remotely Operated Ships (AROS) class‑notation framework, which will be outlined in this paper, including its notation structure and example applications.
The AROS framework defines four modular notation categories aligned with key functional domains: navigation, machinery and engineering systems, safety functions, and mission‑specific or operational capabilities. Each notation is paired with qualifiers for Mode of Operation and Location of Control. Together, these elements convert high‑level autonomy concepts into clear descriptions of system behaviour, human involvement, and operational design domains. This structure supports transparent communication among designers, operators, regulators, and classification societies.
Beyond functional categorization, the framework incorporates two structured qualification processes forming the foundation of its assurance methodology. System Qualification (SQ) provides technology‑specific verification for critical autonomous components, And Concept Qualification (CQ) evaluates vessel‑level designs and Concepts of Operation (ConOps). Together, these qualification processes offer a scalable, repeatable assurance pathway that reduces reliance on purely bespoke qualification approaches and enables reuse of previously qualified technologies.
This paper will illustrate the framework through examples such as for periodically unmanned bridge concepts and for shore-supported remote machinery control. These cases show how different combinations of notations and qualifiers can support reduced‑crew models, remote engineering oversight, and autonomous navigation. They also demonstrate how fallback strategies, communication redundancies and human‑machine interfaces can be designed to meet defined safety requirements.
While the DNV framework represents a significant step toward narrowing the MASS assurance challenge, by codifying functional performance expectations, human‑system interaction requirements, and verification obligations, several industry‑wide challenges remain. These include the need for more detailed and quantitative performance metrics, more robust and standardised testing and verification methods across diverse operating contexts, and clearer competency standards for remote operators. The paper concludes by outlining how the identified challenges can be addressed.
A Classification Framework for Maritime Autonomous Surface Ships
