Autonomous ships taught to read official navigation data

UKHO is working with tech company Marine AI, which has recently secured several key deals with companies keen to exploit machine learning and data analytics.

For eight months, UKHO and Marine Ai will work on a project that aims to process ADMIRALTY Sailing Directions, or Pilots, and Radio Navigation Warnings in a way that an autonomous control system could act upon. The information is currently written in natural language for human interpretation, which cannot be read by machine. MASS therefore has to rely on human interpretation of text-heavy data that is often described as ‘non-standard nautical language’.

“The breakthrough could transform global shipping by enabling uncrewed vessels to make safe, real-time decisions based on the same authoritative data used by professional mariners,” UKHO says.

“By proving this capability on the water, we are closing one of the biggest gaps in MASS autonomy and taking a major step towards safe, fully automated operations,” said Marine AI technical director Oliver Thompson.

“This project will address the challenges posed by unstructured text, legacy broadcast formats, and the lack of machine-readability, by retraining a bespoke large language model and developing supporting AI agents to structure the data before it is fed into GuardianAI’s tactical engine and human–machine interface,” said UKHO.

“Working with Marine AI allows us to push the boundaries of how autonomous systems can use official hydrographic information,” said Mark Casey, Head of Research, Design and Innovation at the UK Hydrographic Office. “The outcomes will not only support the safety of lives at sea but also feed directly into the development of the IHO’s S-100 framework.”

A live on-water demonstration will be carried out next spring using ZeroUSV’s Oceanus12 vessel, which will be equipped with Marine AI’s GuardianAI suite.

Listen here to our podcast with Marine AI from earlier this year.