As the pandemic exacerbated the need for less contact and fewer crew members, shipping companies turned to artificial intelligence-assisted navigation.
The Mayflower is undergoing sea trials this month to integrate its AI captain with the rest of the ship, a trimaran made of lightweight aluminum and composite materials, resembling a trident-shaped dagger slicing through the water. Photo: Tom Barnes/IBM
Sometime next April, a 50-foot-long autonomous vessel will break free from the digital shackles of its human controller, scan the horizon with radar, and then sail west across the Atlantic. Unlike the first Mayflower in 1620, which took orders from a captain, the autonomous Mayflower will not take orders from a single person; instead, it will receive orders from an “AI captain” built by programmers at IBM.
The Mayflower's computing system processes data from 30 onboard sensors and six cameras to help the vessel navigate the oceans, follow shipping rules (such as how to pass other ships at sea), and control electrical and mechanical systems such as the engines and rudder. If problems arise, there will be no one on board, although it does have to send daily reports to a human operator in the UK. Today, the Mayflower is a remotely controlled boat sailing near Plymouth Harbour. Converting it into a fully autonomous sailboat by next April will be a big project for Andy Stanford-Clark, IBM's chief technology officer for the UK and Ireland. Stanford-Clark and his colleagues have been building an AI captain for the past few years, training it with over a million images of ships, buoys, and natural features such as cliffs and icebergs. Stanford-Clark says the algorithms that allow it to make navigational decisions may also include some wildcards.
“We don’t know if there will be any emergency actions,” he said. “This is a scientific experiment.”
Stanford-Clark doesn't believe the ship will head out of control towards Rio, but it may begin to behave more like a human, or at least a marine biologist. In fact, the Mayflower autonomous vessel and its silicon-based captain will sample plankton and microplastics in the ocean. This is part of the scientific mission of ProMare, the US nonprofit research organization that sponsors the project and is leading the $1.5 million project alongside IBM and the UK-based maritime technology startup Marine AI. Clark is developing a set of commands to allow the software running each experiment to instruct the AI captain to turn the ship to accommodate what the experiment wants to measure.
“Imagine an interesting scenario where two experiments send competing requests to each other,” he said. “There could be a bidding war where the experiment that is, in some sense, the most ‘important’ gets priority.”
This is precisely the kind of emergency action Stanford-Clark hoped for. The Mayflower is currently undergoing sea trials this month to integrate its AI captain with the rest of the ship, a trimaran made of lightweight aluminum and composite materials, resembling a trident-shaped dagger cutting through the water. The central hull houses the ship's AI brain (a device that operates locally and doesn't require high-bandwidth satellite connections to process data and make decisions), a cargo hold for 1,500 pounds of scientific equipment, and a system to charge lithium-ion batteries to power the engines and other electrical systems. It has a backup diesel engine in case of sun exposure. If all goes well, the Mayflower will conduct sea trials in Dublin, Ireland and Rotterdam, Netherlands in early 2021, before embarking on its transatlantic voyage in April 2021.
Although this ship won't carry passengers or cargo, the artificial intelligence and advanced autonomy it's testing are now slowly being implemented on merchant ships. From the Baltic Sea to Singapore, shipping companies are plugging in new AI-based navigation systems to take humans off the bridge, or at least make their lives on board easier. Like airplanes, almost all large ships have an autopilot that can navigate from one point to another when not much is happening. But the new plan will allow ships to navigate autonomously or even uncontrollably in congested waterways or ports, and react to dangers at sea.
AI-powered tugboats and tenders are mooring cargo ships, self-docking ferries are transporting cars and people through waterways, and soon small cargo ships will be hauling fertilizer and groceries around the Scandinavian fjords with little or no crew. Advances in AI and sensor technology, along with a widespread rethinking of how to transport goods globally with fewer people to save money and avoid pandemics, are driving this maritime transformation. Some vessels are being converted, others are being built from scratch, but smaller vessels are likely to be the first to set sail, steered by AI.
These new maritime systems face the same hurdles as self-driving cars—concerns about liability, uncertain regulations, and cybersecurity. Meanwhile, companies developing AI-based ship and navigation systems are seeing growing demand for their technology. Shippers want to avoid their cargo being stranded at sea or in ports due to crew members contracting the coronavirus. One maritime executive stated that local pilots—those responsible for navigating the last few miles of large vessels—are reluctant to board ships due to the coronavirus risks. This is slowing down what used to be routine as large numbers of vessels enter congested ports or dangerous estuaries.
For these commercial shippers, the near future isn't about self-driving Ubers, but rather driver-assisted Teslas. They don't want artificial intelligence to replace captains of large ships, but rather to enhance their capabilities.
“Sometimes people get really bored on the high seas,” said Hendrik Busshoff, autonomous product manager at Wörtsilä Voyage, a Helsinki-based maritime technology company that also provides high-resolution radar equipment for the Mayflower project, as well as several other autonomous spacecraft.
“We may not remove this person from the ship, but we will move them off the bridge so they can do more high-value work and call on this person when needed,” Buschoff said.
This advanced concept of AI-powered captains is being deployed in Norway today, where Kongsberg Maritime has built a 260-foot autonomous, zero-emission vessel called the Yara Birkeland.
When Kongsberg first began discussions with Yara, the Norwegian-based global fertilizer giant, in 2017, the plan was to build a fully driverless vessel capable of loading, navigating, and unloading cargo between three Norwegian ports without human intervention. However, things changed after Kongsberg officials realized they would have to convince Norwegian and international regulators to draft an entirely new set of rules to allow large vessels like the Yara Birkeland to operate without a human captain.
“In 2017, there is no doubt that [our ships] will be unmanned,” said An-Magritt Tinlund Ryste, Autonomous Project Manager at Kongsberg. “But you also realize that there are no rules and regulations, and they are all based on having people on board. You either need permission or you have to change the rule set.”
So Kongsberg shifted gears. After the Yara Birkeland's hull was built in a Romanian shipyard and launched in February 2020, Kongsberg decided to add a few crew members to the fertilizer transport vessel while maintaining autonomous systems. Humans would act as an insurance policy should problems arise. The Yara Birkeland's sea route would replace 40,000 transports annually, with the company's trucks delivering fertilizer overland around the fjords, reducing diesel pollution and making those rural roads safer. But according to the October 20 earnings call, the pandemic delayed this ambitious project. Author of the call: Yara International CEO Svein Tore Holsether. "Given the complexity of Yara Birkeland and the expertise required to support it, we've decided to slow things down for now," Holsether told investors.
While the Yara Birkeland is out of service, Kongsberg is pushing forward with two more autonomous vessels, each transporting 16 electric grocery delivery trucks from a warehouse on one side of Oslo's main fjord to a distribution point on the other. Kongsberg signed a contract with grocery retailer ASKO on September 1st, and the two vessels are expected to be delivered in 2022. Both vessels will initially have reduced crews before transitioning to unmanned operation. Norwegian maritime officials have expressed their approval and will oversee the project.
Kongsberg's Ryste says that after meeting regulatory requirements, the biggest challenge in building autonomous vessels is planning for problems with weather, the vessel itself, or its communication systems. "How do you meet the minimum risk conditions? How does the vessel operate if you lose all connections to shore, and how do you build that into the architecture?" Ryste says. "How do you make the right decisions? You need to pay attention to the dynamic picture around you."
If an autonomous system malfunctions on a large cargo ship like the Yara Birkland or ASKO cargo ferries, and then collides with smaller vessels or docks, the consequences would be enormous. This is why some marine autonomous companies start small.
According to founder and CEO Michael Johnson, Boston-based Sea Machines has installed its autonomous control systems on more than 50 vessels, including tugboats, oil barges, small workboats, and oil rig support and supply ships along the East Coast and Gulf Coast. These systems automate many functions and allow the human captain on the bridge to identify hazards more quickly. No one is operating them on their own yet.
Meanwhile, the company is testing a new 29-foot workboat for the U.S. Coast Guard, which can be operated remotely from shore or switched to fully autonomous mode. Johnson said the pandemic-related shutdowns have benefited his business. He's receiving calls from shippers wanting to move their cargo without worrying about the latest lockdowns.
“You have large shipping companies that can’t replace their crews, and hundreds of thousands of crew members were stuck on their ships in March because they couldn’t get off flights or because of the overall risk of bringing in new crew,” Johnson said. “Executives at these companies are looking to mitigate the risk.”
With the rise of maritime machinery, regulators are taking notice. The International Maritime Organization (IMO), a UN agency governing international shipping, is undertaking "scope definition work" to develop new rules for transoceanic navigation for autonomous vessels. It has identified four levels of autonomy for vessels: the first is a vessel with automated processes and decision support, with a crew ready to take control; the second is a remotely controlled vessel with a crew; the third is a remotely controlled vessel without a crew; and IMO spokesperson Natasha Brown stated that level 4 is a fully autonomous vessel capable of making its own decisions and determining its actions.
AI Wired Guide
Superintelligent algorithms won't do all the work, but they learn faster than ever before, covering everything from medical diagnosis to ad placement.
After Tom SIMONIT?
Some commercial ship owners would say they would be happy to achieve Level 2 – a pilot-assisted navigation and collision avoidance system that can also save fuel and improve safety.
Omri Dagul, executive vice president of Ray Car Carriers, is testing an AI-based navigation and operating system designed by Tel Aviv-based startup Orca-AI, which operates 55 large vessels that transport new cars from factories in Asia to ports in North America and Europe. He doesn't plan to operate his own robotic fleet, but rather hopes to use AI captains to reduce the number of personnel on the bridge. "It will be very autonomous while there is a crew on board," Dagul says. "Perhaps the crew won't be as highly skilled, but it's still a manned vessel."
While the economic benefits of autonomy seem obvious to shippers like Dagul, the industry still has many issues to resolve. Some are legal, others are technical, and this type of issue has trapped many self-driving cars in a perpetual testing mode. On one hand, companies working on marine autonomous systems are building or modifying their vessels using different types of technologies and communication platforms. Maritime regulators—whether it's the IMO or national officials—may choose one operating system and discard others.
Even the developers of the Mayflower (a small research vessel with no commercial applications) were struggling to figure out how to obtain testing clearance once the ship was ready to sail in open water.
IBM's Stanford-Clark said, "Many existing laws concern human eyes entering ports." "We'll be pioneering this, but when we sail to Dublin or Rotterdam, we'll have to modify some rules to do so. Who's responsible if an accident occurs on an autonomous vessel?"
John Clark, the first human pilot of the Mayflower, never had a problem with lawyers or insurance companies. He did, however, have a problem getting lost. He was supposed to land south of Massachusetts, but a storm forced the crew to head to Cape Cod and then Plymouth. The delay extended the voyage by more than two months, leaving the passengers hungry and cold. It is speculated that a GPS-navigated, AI-controlled ship would have taken a more direct route across the ocean.
