Antarctic waters are an icy world of extremes where danger often appears without a warning.
Hidden icebergs and dangerous growlers – fragments of floating ice that are often transparent, truck-sized, and less than one meter high, can remain nearly invisible in rough seas, and creeping fogs that arrive unannounced!
Vessel crews rely on every available tool to navigate safely through the Antarctic – one of the most challenging maritime environments on Earth.
But now, Artificial Intelligence (AI) has made navigating these extremes less challenging. It should be noted that maritime environments are exceptionally difficult because no two conditions are ever identical.
Icebergs appear differently depending on light, sea state, weather, water temperature, viewing angle, and time of day. Thermal signatures shift constantly.
Marine mammals surface unpredictably. Fog, snow, spray, and darkness all complicate detection.
Australian maritime technology company SEA.AI stated that teaching AI to distinguish waves, debris, marine mammals, vessels, and icebergs in constantly changing ocean conditions requires enormous amounts of real maritime imagery.
AI-generated imagery can help supplement training data, but it cannot fully replicate the complexity of real operational environments. That is why the data collected by Malizia Explorer is so important. It has captured thousands of hours of visual and thermal imagery from Antarctic waters, including real-world iceberg detections, thermal contrast data, marine mammal observations, and low-visibility navigation scenarios.
Maritime AI
Unlike automotive AI or industrial computer vision, almost no labeled maritime data existed. SEA.AI and Team Malizia trained maritime AI for this.
IMOCA race yacht became an early testing platform, exposing the system to some of the harshest conditions in offshore sailing, including Southern Ocean crossings and Vendée Globe campaigns. Every voyage helped refine the technology, improve detections, and build the proprietary maritime dataset that now forms the foundation of SEA.AI’s machine vision systems.
Boris Herrmann, Skipper of Team Malizia, says they have been using SEA.AI technology for years aboard our IMOCA race yacht. “As the systems evolved, we saw the value they could bring not only for offshore racing, but for broader maritime operations and exploration. Equipping Malizia Explorer with SEA.AI was a natural continuation of that work.”
Solenn Gouerou, Head of Marketing at SEA.AI, highlighted that the partnership worked because the technology was being tested in real operational conditions. “You immediately learn what works offshore and what still needs improvement. And when the crew tells you directly what they experience on the water, that’s the kind of feedback that’s invaluable when developing maritime systems.”
Also Read: Integer and University of New Hemisphere to Advance US Navy Situational Awareness
Challenges
Captain Jonathan Morice has seen the value firsthand during Antarctic operations. He shared that the main challenges are icebergs.
“We have radar, which is quite reliable for big icebergs, but then you have those medium-sized icebergs, which are very heavy and can do a lot of damage. The radar doesn’t always pick them up because they don’t go out of the water so much.
“The camera has been very useful during the boat’s trips to Antarctica,” he continues. “I was also quite surprised by the thermal version. Even though you would think an iceberg is cold, with the temperature difference between the ocean and the iceberg, you can really spot anything, even at night.”
The data contributes to the development of machine vision systems intended for broader maritime use from commercial operations and research vessels to cruising sailors and professional mariners navigating increasingly complex waterways.
SEA.AI is now experimenting with AI-generated images to supplement our training datasets.