Mayflower autonomous ship makes unmanned Atlantic crossingMayflower autonomous ship makes unmanned Atlantic crossing
With “AI Captain” at the helm, research vessel completes transatlantic voyage despite mechanical issues
June 14, 2022
With 'AI Captain' at the helm, research vessel completes transatlantic voyage despite mechanical issues
The Mayflower Autonomous Ship (MAS) on June 5 completed a 40-day, transatlantic voyage from Plymouth, U.K., to Halifax, Nova Scotia, without any human captain or crew on board, almost a year to the day that its first attempt was thwarted by a mechanical issue.
While this trip didn’t go according to plan either — more mechanical problems diverted the ship to the Azores Islands for repairs and then to Halifax instead of Virginia as planned — it was still a significant accomplishment and demonstrates the possibilities of autonomous AI, says IBM, the lead technology and science partner behind the ship, which was built by marine-research nonprofit ProMare.
Mayflower was piloted by an “AI Captain,” which was powered by IBM automation, AI and edge computing technology. The cognitive pilot used six AI-powered cameras, more than 30 sensors and 15 edge devices to provide actionable recommendations, which it then interpreted and analyzed.
The AI Captain’s decisions were based on if/then rules and machine learning models for pattern recognition and were often “crucial, split-second” ones that involved rerouting around hazards and marine animals. During the voyage, the AI Captain learned from the outcomes of its decisions, made predictions, managed risks and refined its knowledge through experience, all while integrating more inputs in real time than a human can do alone, IBM says.
The ship itself is a trimaran with a main hull and two outriggers on either side. The configuration was selected for its low, highly stable profile in the water. Made from aluminum and composite materials, the ship weighs about five tons and is 49 feet long and 20 feet wide — half the length and less than 3% of the weight of the original Mayflower, according to ProMare.
Today’s Mayflower is propelled by twin 20 kW permanent-magnet electric motors instead of sails, although the solar panels that power them use lithium ion-phosphate batteries and exterior solar panels to power its computers, which produce less carbon than traditional diesel engines. Its cargo bay is capable of holding 2,200 pounds of scientific equipment, a key feature of the project.
While the feat of an unmanned ocean crossing is itself a scientific experiment, the ship’s larger mission is oceanographic research. ProMare believes autonomous research vessels like the Mayflower will be able to collect the vast amounts of data needed to save the planet’s failing oceans much more economically than manned craft.
During the 3,500-mile voyage, interested observers were able to track the Mayflower’s progress through an online dashboard that provided a wealth of real-time information, including speed, weather, energy usage and live video.
While the Mayflower’s first crossing was focused on shaking out any bugs in the AI Captain — the problems that diverted the craft from its planned route were mechanical; the autonomous piloting system worked perfectly, IBM says — the trip did allow researchers to conduct several experiments that delved into three main topics: Ocean health, marine mammals and open-ocean tides and waves.
One of the experiments tested a new IBM technology, HyperTaste, which was developed for AI-assisted rapid-chemical testing of liquids. Inspired by the sense of taste, the technology uses electrochemical sensors that react to molecules and ions in liquids to find the digital chemical fingerprints of the test matter. This particular test studied the chemistry of the ocean over the course of the voyage in an effort to gauge ocean acidification caused by increased atmospheric carbon dioxide.
Another experiment used acoustic sensors to collect a stream of underwater audio, searching for marine mammal vocalizations, which were analyzed by an onboard machine-learning model to learn more about the animals’ population distribution in rarely explored locations in the middle of the Atlantic.
Looking ahead, IBM believes the Mayflower’s voyage will be a catalyst for the further advancement of AI and AI-powered automation at the edge in applications far from sea level.
“For us, the Mayflower Autonomous Ship’s challenges — saving time and costs, making trustworthy predictions and solving complex data problems — are not unique,” IBM says. “MAS represents what’s possible when you harness the power of data … and how technology like AI-powered automation can take intelligent data and make it actionable to make informed business decisions, no matter the industry.”