Visualisation of cyber vulnerabilities in maritime human-autonomy teaming technology

Abstract Marine operations are increasingly leveraging AI technologies to improve performance and efficiency. However, there are many factors that affect safety, from the remote operating centre (ROC) to autonomous vessels. This includes both hardware and software that augments or replaces direct human control. Further, the integration of AI for autonomy also introduces new cyber security vulnerabilities arising from adversarial threats and complex interactions between conventional and AI-driven systems. In response, we propose a comprehensive assessment approach that assesses the security of marine technologies by addressing risks to both traditional systems and emerging AI components. In this wider system-of-systems view, the authors detail the key elements of a thorough security assessment of a Maritime Autonomous Surface Ship (MASS), ROC and vessel ecosystem and present the corresponding cyber security mitigations for systems running AI in MASS. These penetration tests are carried out on real instances of AI, ROCs, and autonomous vessels to demonstrate feasibility and impact. These individual tests and evaluations are then compiled into a single case study that highlights the potentially devastating consequences of deploying inadequately secured technologies in MASS. This case study is then used to discuss possible mitigations that can be used to better secure and protect the physical and digital assets of MASS.