Nowhere in the UK is more closely associated with British sea power than Portsmouth. Home to two-thirds of the Royal Navy’s surface fleet, its historic dockyard and museums also play host to the fragile timbers of the Mary Rose and the scourge of Napoleon’s navy, HMS Victory.
Earlier this month, they were meant to be joined by another, equally illustrious vessel. White, sleek and bedecked with solar panels instead of sails, the Mayflower is the autonomous successor to the boat that bore the Pilgrim Fathers to their destiny in Massachusetts – and, say its creators, the future of shipping for the next four hundred years.
“Unfortunately, she couldn’t make it,” says Andy Stanford-Clark, a master engineer at IBM and one of the boat’s leading designers. The reason, he explained earlier to an auditorium of journalists and businesspeople, was the weather.
While the Mayflower was perfectly capable of making the short voyage from Plymouth alone, the waters proved too choppy for its manned escort vessel. Consequently, its robotic charge was confined to a much shorter patrol off Rame Head in Cornwall, its progress awkwardly depicted behind Stanford-Clark on a blocky livestream.
Despite these minor setbacks, the Mayflower is one of the most advanced examples of a series of unmanned vessels trialled in recent years. Motivated by the obvious cost and safety implications involved in removing the need for a human crew, unmanned cargo ships have already been tested in Japan and Norway. Trials of autonomous ferries and water taxis, meanwhile, have also taken place in Finland and Germany.
While these vessels have been designed to operate in coastal waters, however, the Mayflower was built to cross oceans. Constructed in collaboration with ProMare, a marine research organisation, the boat’s hull is bursting with sensors, edge computing and AI systems that help it make swift and complex decisions on route planning, how and when to avoid obstacles, and conduct climatological and ecological research.
Stanford-Clark insists, however, that the boat is not IBM’s attempt to dominate world shipping. “Our mission is cloud and AI,” he says, and in that sense, the Mayflower is a useful testbed for a whole host of technologies that have direct applicability to land-based problems.
Indeed, a lot of the systems on board are already commercially available, explains IBM’s business automation specialist, Doug Coombs. Consequently, businesses of all stripes are “looking at how we’re applying this technology and thinking, ‘Okay, I’ve got similar use cases’,” he says, in everything from product recommendation to assessing the eligibility of organ donations for transplant patients.
There are also very clear maritime applications for the lessons learned in operating Mayflower, explains Stanford-Clark. One, he says, “is the transferability of the ‘AI captain’ technology, to other types of ship.” This system is not only capable of operating the vessel in conditions where it’s cut off from communications with the outside world – important, say, if it finds itself in the middle of a storm – but also in full compliance with the Convention on the International Regulations for Preventing Collisions at Sea and the SOLAS Convention.
Crucially, explains Stanford-Clark, the AI captain also has a sense of self-preservation, which could make it a useful safety system for manned vessels. “I’ve described it as like a ‘guardian angel,’ looking over the shoulder of the human captain,” he says.
It certainly sounds like a more realistic application of AI in maritime navigation in the near term. While there have been trials of unmanned vessels around the world, the shipping industry retains a stand-offish attitude towards removing the human from the bridge.
“I think the marine industry recognises that autonomy is coming,” says Stanford-Clark, who has participated in conversations with cargo companies about the applicability of IBM’s systems to their own vessels. “It’s not if, but when. [But] we accept it’ll be a while yet before every ship is run autonomously.”
Another step toward autonomous operations may be remote control. Recent advances in satellite connectivity via LEO constellations raise the possibility that entire fleets should be controlled by a navigator sitting thousands of miles away. While companies like Inmarsat and Rolls-Royce have conducted their own research into this area, Stanford-Clark remains ambivalent. “There isn’t an AI challenge there,” he says. “Just being able to remotely pilot a ship – well, that’s like a radio control boat on a lake.”
IBM and ProMare’s aims for the Mayflower are rather more ambitious. Last year, the vessel attempted its first transatlantic voyage, following the same route charted by its namesake in the 17th century. Shortly after setting sail from Plymouth harbour, however, the ship developed a fault in a coupling that helped charge its batteries and decided to turn back. Ironically, all the AI systems worked perfectly.
Having spent the past nine months re-testing all the technology aboard, the Mayflower and its AI captain will try crossing the Atlantic again later this year. Success will mean the first civilian crossing of an ocean by an autonomous vessel and provide a window into the future of shipping as we know it – that is, “as long as it doesn’t break again,” says Stanford-Clark. “That’s always the hope.”