A pioneering project funded by the Defence and Security Accelerator (DASA) and led by Plymouth-based Marine AI aims to enable natural language communication between uncrewed surface vessels (USVs) and crewed ships.
Trials are underway in the Plymouth and Portsmouth areas, testing a large language model designed for autonomous ship-to-ship dialogue in real maritime conditions.
Building on earlier proof-of-concept developments, the project’s goal is to equip uncrewed vessels with the ability to hold natural, context-aware conversations with both manned and unmanned platforms, mirroring human operator interactions.
This breakthrough addresses the operational challenge of limited communication currently faced by USVs, which rely on strictly coded protocols or human oversight.
READ: Orca AI strengthens fleet safety with strategic Ionic deal
The trials involve Marine AI’s Oceanus12 and supporting craft, alongside prominent vessels such as the Royal Navy’s Patrick Blackett and the autonomous underwater vehicle XLUUV Excalibur.
Master mariners will assess the system’s responses, with results recorded in a formal evaluation report.
Oliver Thompson, Technical Director at Marine AI, said: “Uncrewed platforms can only operate safely alongside conventional vessels if they can be understood. This project is about proving that an autonomous system can use natural language in a way that makes sense to mariners in real-world conditions.”
The anticipated success of this research would mark the first demonstration of natural language interoperability between uncrewed and crewed vessels, facilitating safer integration of autonomous platforms into busy maritime environments.
By enabling trustworthy and situationally appropriate exchanges, the technology seeks to reduce misunderstandings that could compromise safety during mixed traffic operations, including military missions, search and rescue, or anti-piracy efforts.
