SeXTANt (Secure X Trustworthy AI Navigation)

The International Maritime Organisation have highlighted an increasing need for AI in the maritime sector from as early as 2017 and have run regulatory scoping exercises focused on Maritime Autonomous Surface Ships (MASS). To work effectively and efficiently they would need to have a high level of autonomy, however this comes with its own risks including:

• Security of AI needs to be incorporated throughout the lifecycle of AI development
• Potential for AI developed for a specific task is re-purposed for a negative application
• Robustness of the AI
• Legal, explainability, ethical and bias issues potentially involved

It is currently illegal to sail a fully autonomous ship on UK waters, however, semi-autonomous are allowed. In order, for fully autonomous ships to be legalised, the above issues need to be addressed and AI development/tools across the sector to be regulated and standardised internationally.

This project focuses on the network and system security aspects of Maritime Autonomous Surface Ships (MASS).

Rather than concentrating on the security of the AI itself, the project examines the security of the hosting environments where the AI operates within autonomous systems.

Additionally, the project explores data visualisation, specifically how cyber and cyber-physical security data impact marine technology and engineering design and implementation.

Cyber vulnerabilities in human-autonomy teaming technology

To demonstrate the cross-discipline research undertaken by the team, they developed an animation that outlines the key considerations of this project. The video explains the projects focus on secure and trustworthy AI for autonomous maritime systems, delving into the possible cyber vulnerabilities of AI systems on uncrewed surface vessels and in remote operating centres. The video then also goes on to demonstrate how AI can be developed to both be operational and have cyber resilience using data fusion.