Captained by A.I., This New ‘Mayflower’ Will Cross the Atlantic This Spring | Innovation | Smithsonian Magazine

On September 6, 1620, the Mayflower set sail from Plymouth, England, bearing 102 passengers and about 30 crew members. After a perilous 66-day journey across the North Atlantic and a harsh winter, the surviving Pilgrims and crew of the Mayflower encountered the Wampanoag, who were familiar with Europeans as traders, kidnappers, and agents of plague. The Wampanoag have lived in what is now southeastern Massachusetts for more than 12,000 years, and the two communities coexisted for about 50 years before war began.

The European immigrants prevailed in that war, as well as in a long series of conflicts with other tribes. On this land taken from Indigenous Peoples, a new nation was eventually born, largely built by those whose ancestries traced back to the Old World via immigration and slavery.

As the country grew, inventions like the telephone, airplane, and Internet helped usher in today’s interconnected world. But the inexorable march of technological progress has come at great cost to the health of the planet, particularly because of global dependence on fossil fuels. The United Nations declared in 2017 that a Decade of Ocean Science for Sustainable Development would be held from 2021 to 2030. This Ocean Decade calls for a worldwide effort to reverse the oceans’ degradation.

The dawn of this decade, 2020, also marked the 400th anniversary of the Mayflower’s journey. Plymouth 400, a cultural nonprofit, has been working for more than a decade to commemorate the anniversary in ways that honor all aspects of this history, said spokesperson Brian Logan. Events began in 2020, but one of the most innovative launches is still waiting in the wings—a newfangled nautical craft, the Mayflower Autonomous Ship, or MAS.

Built and tested over the past five years, MAS will chart a new path by retracing an old one. This year it will travel from Plymouth, England, to Plymouth, Mass. Throughout the journey, it will collect data that might provide insight into ocean and marine mammal health while gathering information about a sustainable energy source—the ocean’s waves—that could help power our world.

Aye, A.I., Captain!

When discussion began about what a new Mayflower might look like, Brett Phaneuf, cofounder of the marine research nonprofit ProMare, advocated for designing and building an innovative ship instead of a replica. At the time, militaries, private companies, and research agencies were all focusing on uncrewed systems, he said. “Everybody was looking at what the next iteration of technology [would be]…and it’s autonomy—true autonomy.”

Phaneuf remembered watching the IBM supercomputer Deep Blue defeat world chess champion Gary Kasparov in the 1990’s. With the opportunity to build a new Mayflower, he thought, “we must be able to bring some of this technology to bear…how hard can this be?” Pretty hard, as it turns out. Creating an artificially intelligent captain to navigate the ocean proved trickier than navigating the famously difficult strategies demanded by chess. The ship must have agency to make decisions by itself, said Phaneuf, “so even when you don’t have communication with it…it can still function safely and achieve a goal.”

To navigate coastal waterways and open oceans alike, MAS uses artificial intelligence (A.I.) developed by IBM and ProMare. The A.I. Captain uses its camera-based vision system to recognize potential hazards (learned from perusing pictures ranging from ships to seagulls). Two weather stations on board, plus a link to IBM’s The Weather Company, provide the A.I. Captain with real-time meteorological information. Sensors that notice how waves roll by inform the A.I. Captain of sea conditions. With all these inputs, the A.I. Captain must make and execute educated decisions to avoid collisions and stay upright and safe.

To help with those decisions, MAS uses IBM’s Operational Decision Manager, a tool used by financial institutions to determine, for example, whether certain people qualify for loans. In this case, the rules aren’t financial but, instead, focus on avoiding impacts. The International Regulations for Preventing Collisions at Sea (COLREGs), published by the International Maritime Organization, serve as the “rules of the road” for the ocean, according to Lenny Bromberg, program director for automation, intelligence, and decision management at IBM. Because the COLREG framework bounds the Operational Decision Manager, he said, the A.I. Captain can decide how to safely and legally proceed when anything from dolphins to debris to destroyers appears within its sights.

A diesel generator combined with batteries and solar panels drives these systems and sensors, as well as the ship itself, said Phaneuf. When the batteries are low, the diesel-powered engine starts, charges the batteries, and shuts itself off, allowing the ship to run via battery and solar. “If you want to cross an ocean, we could probably build a [totally] solar-powered ship,” he said. “But then you’d need to take out about 99 percent of all the science.”

“My Humans May Want This!”

Without a physical captain or crew, there’s no need for sleeping quarters, a galley, or anything else humans may require. Eliminating these compartments freed ProMare and its partners to design a lightweight trimaran whose innards can be devoted mostly to science. The energy-efficient payload sits in a space of about 2 cubic meters—slightly larger than a refrigerator box, said James Sutton, a software engineer at IBM who helped build the systems running the MAS science package. The ship can hold 700 kilograms (1,500 pounds) of scientific equipment.

To sample seawater, said Sutton, the ship has an intake pipe that sucks water out of the ocean and into one of several sensor systems. To keep the intake from getting clogged with large detritus like seaweed, grills and filters armor this crucial entrance. Flowmeters ensure that tubing inside the ship remains clear. With this design, he said, “we don’t have to worry about sticking lots of sensors on the outside of the ship.”

One slug of water heads into a fish tank–like box equipped with sensors that test for temperature, salinity, pH, and oxygen content. A fluorometer optically observes whether anything in the seawater fluoresces, which can be used as a proxy for quantifying chlorophyll-rich algae, said Sutton. A satellite link uploads data from the fish tank to the cloud in real time whenever possible, he said.

Also on board is a version of HyperTaste, IBM’s A.I.-assisted “tongue” designed to rapidly test the chemical composition of liquids. This system, based on a tool originally used to identify counterfeit brandy, according to Phaneuf, includes four separate sensors that measure the presence of various molecules and ions in seawater. The brandy-sniffing version takes about a minute to test, whereas the MAS version takes about 15.

HyperTaste begins each cycle by sampling from a bucket of artificial seawater on board that serves as a baseline reading, followed by a rinse of deionized water and a taste of the ocean, said Sutton. When it’s done, like a wine connoisseur, the system spits the seawater back into the ocean and cleanses its palate for the next sip. HyperTaste will measure quantities of calcium, magnesium, and other markers of the ocean’s biogeochemistry about every 15 minutes, said Sutton.

A holographic microscope will photograph water samples from several different directions to build 3-D images of any microbes or microplastics present in the water, said Sutton. Unfortunately, because the photographic data sets are too big to send over the satellite link, they will be saved on the 12 terabytes of solid-state hard drive storage available on board, he said.

To complement the holographic microscope, a robotic sampler will collect and refrigerate about 20 liter-sized flasks of seawater that will be given to a local university upon arrival in Plymouth, Massachusetts, said Phaneuf. To choose where water samples come from, he said, the hope is to teach the A.I. system to recognize anomalies and say to itself, “My humans may want this!”

Chemical oceanographers take these same kinds of measurements—temperature, salinity, oxygen, pH, fluorescence, and others—not only at the surface but also in vertical depth profiles, said Hilairy Hartnett, an oceanographer at Arizona State University. These metrics help scientists monitor water density, nutrient content, and the ocean’s health. “What we lack in oceanography is lots and lots of coverage,” she said. “The oceans are freaking huge!”

That MAS can collect such detailed information about the surface ocean is appealing, Hartnett continued, but “until we see the data, it’s going to be hard to know what we can do with it.”

The Next Wave of Sustainable Energy?

MAS’s inertial measurement unit records acceleration in all three axes 300 times per second, painting an incredibly high-resolution picture of how waves rise and fall in the open ocean, said Sutton. Six cameras mounted around the central mast of the ship keep watch as the waves come toward MAS, allowing researchers to use the pixel velocity to calculate the size and speed of each wave. By linking inertial measurements to individual waves, said Sutton, researchers can calculate how many joules of energy each wave carries. Such information could eventually help place wave energy harvesting systems in the ocean.

Wave energy harvesting, said Michael Webber, a professor of mechanical engineering at the University of Texas at Austin, “is global, sustainable, nondepletable, and emission free.” To get usable electricity, Webber said, you need either rotational motion or an electrical potential—a difference between charges. The most basic way to make electricity is by burning fossil fuels to turn water into steam, which spins a turbine like a backward fan, he said. That turbine rotates a shaft, which rotates magnets in a generator. “That’s the basis for just about all of our electricity,” he said.

In the ocean, the rise and fall of waves can be converted into rotational motion, for example, via a buoy that bobs up and down or a gate on the seafloor that rocks back and forth, Webber explained. “Earth does the heavy lifting for you,” he said. However, as powerful as waves can be, “it takes a lot of equipment to convert that into useful energy, and you have to worry about marine ecosystem impacts.”

“I would be interested to know what the wave structures are like around the oceans,” said Webber, but mapping wave energy by boat is difficult because the oceans are large and constantly changing. Satellite imagery, he said, could be very helpful. Data from MAS, said Jyotika Virmani, executive director of the Schmidt Ocean Institute, can help verify and calibrate such satellite data. “It’ll be interesting to see how this autonomous ship’s information meshes with what we can do from satellite data,” said Hartnett.

Bigger Ships, More Science

Choosing which aspects of the ocean MAS would explore for its inaugural voyage, “was kind of organic,” said Phaneuf. At some point, “I went, ‘Stop! We have enough stuff right now.’”

The science team had to focus on what could be done without the watchful eye of a crew, said Sutton. For example, a single hydrophone mounted on a pipe near the center of MAS pokes out from the bottom of the ship to listen for songs of marine mammals like whales and dolphins. In a typical marine acoustic experiment, however, “the gold standard would actually be to have it tethered on a line behind the ship, quite some distance,” he said. “But we didn’t want to risk the line getting caught.”

When Scripps Institution of Oceanography geophysicist Vashan Wright goes to sea, his goal is to image the subsurface as he searches for faults, submarine slides, and paleoseismic deposits. “I can’t imagine [an autonomous vehicle] dragging a 5-kilometer streamer behind it, and an air gun array, and having no problems,” he said. Doing this type of science autonomously “would take a lot of creative thinking.”

A notable absence in the MAS science suite is sonar, especially considering that one of the goals of the Ocean Decade is a comprehensive digital atlas of the ocean. “Right now, we have first-order bathymetry for the world oceans from orbital gravity [data], but that’s pretty coarse compared to what you get from a hull-mounted sonar,” said Robert Stern, a professor of geosciences at the University of Texas at Dallas. Sonar helps scientists map the bathymetry of the ocean in exquisite detail, but current coverage is patchy. “[Autonomous ships] would be perfect for mapping large swaths of the oceans,” said Stern.

However, mapping bathymetry many fathoms deep with sufficient resolution would require equipment that simply can’t fit on MAS as it is currently configured, said Phaneuf. Plus, he said, the power requirement for such equipment is extreme. The ship is simply too small.

The autonomous Mayflower descendants are already on their way, said Phaneuf, speaking of the next two ships in the pipeline. The first will be named after Oceanus Hopkins, born to the Pilgrims during the Mayflower’s crossing. The second child born on the original Mayflower, Peregrine White, will give his first name to the second of MAS’s offspring. Construction of Oceanus, expected to be nearly twice the size of MAS, is scheduled to begin in late 2022 or 2023. These future vessels will have more endurance, said Phaneuf, “and much more payload for science.”

Accessible Oceans

Autonomous research ships could help bring the oceans to those who cannot currently access the world of oceanography. At the moment, admitted Hartnett, “it’s not a super accessible field.”

“[Autonomous ships] would expand access to people who don’t know how to swim or are afraid if something goes wrong,” said Wright. “Sometimes, those are people from historically excluded groups,” he added.

Stern knows this better than most. “I’ve got a degenerative nervous disease called Charcot-Marie-Tooth syndrome, and it affects my motor nerves,” he said. “I get around on a scooter, and I don’t do any fieldwork anymore.” To maintain involvement with marine research, he relies on ships with Internet, which have become more common in the time of Covid-19. “I can’t handle any rocks on the ship, obviously, but I can participate in a much better way than just waiting until [my team gets] back.”

Other life circumstances may not allow people to devote weeks at a time to an expedition, said Allison Fundis, chief operating officer for the Ocean Exploration Trust. This group includes parents and expectant mothers. “For that reason, it’s more important for us to provide that portal to people, so they can experience [the sea] without having to physically be on the ship themselves.”

Some scientists simply don’t have the funding or time to go to sea, said Virmani. With MAS, they could potentially get the data they need to continue their work.

MAS also presents tremendous teaching opportunities, said Hartnett, especially for landlocked universities. “I love being able to find ways for students in my oceanography classes to deal with real-time oceanographic data,” she said. “The ability to help [students] see the kinds of data that we collect and use…is very powerful.”

An Uncertain Future

Uncrewed research vessels like MAS may expand opportunities for scientists but complicate the careers of people who work indirectly with science. “Many people make their livelihood at sea,” said Wright. They are the cooks, the able seamen, and technical staff whose careers require going to sea. “When we think about automation, we have to think about…what happens to them.” Fundis acknowledged this concern but said that replacing crewed missions with autonomous ships “is very much not the case.” Instead, she described autonomous vehicles taking on tasks not suitable for crewed ships, like sailing during risky weather windows and making long transits across remote regions.

Virmani noted that uncrewed vessels could greatly expand the ability to monitor particularly dangerous situations, like the 2011 Fukushima nuclear power plant meltdown in Japan that resulted in radioactivity contaminating parts of the Pacific. “You don’t have people on board, so it’s pretty safe to send something like this to assess what’s going on,” Virmani said of MAS.

A Sputnik Moment

In October 1957, when the Soviets launched Sputnik into orbit, walking on the Moon a mere 12 years later may have seemed like an outrageously unattainable target. Early competitors in the Space Race may never have imagined astronauts living in space or tourists popping into orbit. Similarly, said Stern, MAS may be a Sputnik moment for oceanography.

After its 2020 launch was rescheduled because of the Covid-19 pandemic, MAS attempted its first transatlantic voyage on June 15, 2021. However, a mechanical failure forced Phaneuf and his team to recall the ship to England after three days. It’s now back in the water and ready to try again in spring.

“All Sputnik did was, it went around the Earth, and it beeped,” said Stern. “It didn’t collect any data at all but still revolutionized humans’ relation to space.” An autonomous ship, he said, is like Sputnik. “It doesn’t really have to do much, as long as it can do what it’s designed to do, which in this case is roam the oceans.”