Although uptake of artificial intelligence (AI) in the marina sector is rising, industry experts agree that its adoption is measured. In order to better understand the wide-ranging benefits AI can offer, Charlotte Niemiec looks outside the marina space to the broader marine industry. In talking to Tom Rooney, General Manager at UK-based MarineAI, she discovers how AI is set to revolutionise safety at sea.
Firstly, can you tell me a bit about MarineAI and what you do?
Essentially, we build smart systems that help crewed, lean crewed and uncrewed vessels operate more safely and efficiently. Our core technology product is a suite of software called GuardianAI – the latest generation of our flagship autonomy and navigation platform which, for a crewed or lean crewed vessel, acts like an extra watch keeper in the wheel house or bridge. It’s designed to assist crews with their dynamic route planning decision making and, if permitted, can even take full control of a vessel when needed using a smart network of sensors like radar, sonar, cameras and GPS. We’ve deployed it on several vessels, including uncrewed surface vehicles and the Royal Navy’s experimental ship, XV Patrick Blackett.
We’re now expanding into more defence and deep-sea missions, especially where it’s risky or impractical to send people. The system is modular, too, so we can tailor it to different vessel types and missions. And we’re working closely with organisations like the International Maritime Organization, the Centre for Assuring Autonomy at the University of York, and Lloyds Register to help shape future certification standards. We recognise that if AI is going to be trusted at sea, it needs to be developed, trained and tested with transparency, accountability and strong governance from the start.
That sounds exciting! More broadly, what is the potential of AI in the marine sector?
Technology, like mobile phones, virtual assistants and GPS, has transformed our world. AI is at the heart of this change and impacts every industry. In the marine sector, AI can plot routes by considering weather, traffic or currents, which can result in less fuel used, shorter travel times and fewer emissions. Predictive maintenance systems use it to prevent breakdowns, improve performance and reduce downtime. All this is just the beginning of AI’s potential because it can quickly process huge amounts of real-time data and spot patterns a human might miss, so it can help make smart decisions faster than us.
Is it mostly about efficiency, then?
That’s one part of it, but another important part is safety. Crews deal with constant streams of information, such as mechanical checks, radar data and weather updates. This can lead to cognitive overload, where reaction times slow and mistakes happen. AI can help manage that load by automating more routine tasks, processing data and alerting crew members when something might be wrong. For example, at MarineAI we have developed and trained our own AI models that analyse and fuse real-time hazard data from a variety of vessel sensors, to make rule-compliant decisions for safe navigation. It’s like adding a supercharged, tireless assistant to the bridge – one that never sleeps or gets distracted!
What about real world events? Could AI have prevented incidents like the Suez Canal blockage?
Absolutely. Take the Ever Given incident in 2021, where one of the world’s biggest container vessels became lodged sideways in the Suez Canal after strong winds and other factors pushed it off course. It completely blocked the canal for six days, holding up more than 400 ships and causing huge delays in global trade. The economic impact was massive, with billions of dollars in losses each day. An AI system with environmental awareness and real-time wind analysis could have warned of dangerous drift and recommended a route adjustment.
Another incident was the Norwegian frigate collision in 2018, where the frigate HNoMS Helge Ingstad collided with an oil tanker near the Sture Terminal in Norway. The crew mistakenly thought the well-lit tanker was a stationary object and, despite multiple warnings, they didn’t change course in time. The impact caused a massive gash in the frigate’s hull and, although the crew was safely evacuated, the ship eventually sank. Investigations pointed to human error, poor communication and training gaps.
AI-powered situational awareness systems would have detected nearby vessels, calculated potential collisions and issued evasive instructions in time for the human crews to react, and an AI-supported intelligent autopilot would have manoeuvred the vessel in such a way to avoid the incident altogether. These technologies aren’t theoretical – they’re ready now.
In what other ways can AI help?
AI can help in so many maritime sectors, like the offshore energy market, science projects and defence. For example, our autonomous surface vessels can help deliver cost-effective, low impact hydrographic and scientific surveying, because they can operate for months at a time without human intervention. They also have applications in border security, where autonomous vessels can patrol around the clock and help spot suspicious activity in real time. This can also be applied to commercial shipping where, in addition to providing safe navigation support, the real-time situational awareness picture and contact behaviour modelling created using AI tools can provide vessel operators an early warning of potential piracy or malicious attacks.
Thanks to AI it’s possible to remove crew completely, in some cases. For example, in June 2022, the Mayflower Autonomous Ship completed the world’s first autonomous Atlantic crossing, a landmark achievement powered by our technology and made possible through collaboration with MSubs, IBM and ProMare.
AI is also bringing a whole new level of luxury to leisure boating, especially in the superyacht space. GuardianAI can handle things like fuel and power management, assisting with anchoring and even helping to run ship-to-shore transfers using autonomous tenders, making guest transport seamless and stress-free. It’s also great for personalised experiences like adjusting lighting, climate and even itineraries based on real-time sea conditions. AI tools, in partnership with low-cost sensors, can also ensure the vessel speed is optimised to provide best comfort for the passengers and crew. GuardianAI is compatible with existing sensors and systems, which means retrofitting a yacht with this technology takes just a few days.
We also work with partners like Navtech Radar to integrate all-weather radar technology that helps detect even the harder-to-spot things – like buoys, swimmers or kayaks – in rough conditions.
With all these benefits, then, what’s holding back AI adoption in the marine sector?
The short answer? Trust. People worry about handing critical decisions over to machines. The public see high profile AI failures in other sectors, like self-driving car accidents or biased algorithms, and assume there will be similar risks at sea. There’s also the “black box effect” where, if you don’t understand how a decision is being made, how can you trust it?
What is important is how AI models are trained and bound. MarineAI has put years of effort, experience and resources into constraining our in-house models to focus on specific tasks and training them in the maritime environment. GuardianAI is not one big AI tool, it’s a suite of solidly designed and tested software modules, each supported by a very focused, bounded and highly trained AI model to ensure the output is elegant, predictable and repeatable. Each module provides one element of the system’s overall capability, allowing vessel operators to tailor the functions they need, consistent with their level of trust.
Are there ethical questions, too? Can AI handle morally complex situations?
This is one of the more important areas we focus on. Take a situation where an AI system has to choose between avoiding a collision with another ship or protecting a coral reef – it’s not a simple choice. That’s why our GuardianAI is built using an ethical framework that puts human life first, while also respecting the environment and following all regulations. With GuardianAI onboard, a vessel is even more unlikely to get to the point where such an extreme decision is required. Plus, every decision it makes is logged and can be traced, replayed, reviewed and improved over time.
Cybersecurity is another key concern. The more connected a vessel is, the more exposed it can be. To address that, GuardianAI has been developed to be cyber secure and we offer a range of security measures to suit a customer’s requirements. We also use real-time anomaly detection and back-up systems so that, if one part is ever compromised, another can take over seamlessly.
This sounds promising, but what does the future look like?
The future of AI at sea is all about taking things step by step. We’re not expecting crews to hand over full control right away. Instead, we start with lower-risk, high-reward tasks, like dynamic navigation route planning and collision avoidance or engine monitoring. As the technology proves itself and people get more comfortable, we can move on to more advanced features like conditional control of autopilot for evasive manoeuvres, full intelligent autopilot, remote control or even autonomous docking.
Just as important is training and shifting the culture. Mariners need to understand how AI works, where it can really help and what its limits are. This isn’t about replacing people, it’s about giving them smarter tools to do their jobs even better.
