MarineMAS: A multi-agent framework to aid design, modelling, and evaluation of autonomous shipping systems

ABSTRACT Autonomous shipping brings a great opportunity to improve the maritime traffic safety, reliability, and to reduce the cost, whose prototypes are expected to be industrialized in coming years. Current studies and projects focus on proof development of autonomous vessels, evaluation of autonomous vessel safety, and formulation of guidance for autonomous shipping paradigm. In system level, autonomous shipping is anticipated to realize advanced intelligence and autonomy in terms of planning, situation awareness and shipping safety, decision-making, etc. In this study, being aligned with the aforementioned aspects, a multi-agent system (MAS) framework, called “MarineMAS,” is proposed to aid the autonomous shipping modelling and evaluation. We emphasize the design perspective that especially focuses on critical knowledge transfer to make transition towards autonomous shipping, the core components to achieve system intelligence, and the potential technological set and experience gained to support efficient MAS modelling. A prototype system has been implemented from scratch to demonstrate the key features of the proposed MarineMAS framework. Through characterizing the multi-agent definitions and incorporating the corresponding evaluation logics, MarineMAS can be further developed and applied for various applications in autonomous shipping evaluation and management scenarios