Popular culture has been inundated with stories about autonomous or semi-autonomous cars. Sometimes the stories are about a bright future in which car crashes are a thing of the past. More often, the story is about autonomy that failed in some way, with tragic human consequences.
In the meantime, the same technologies that have made autonomous road vehicles a reality are also permeating the maritime field, leading to developments such as Yara Birkeland, an autonomous containership that received $16 million in project funding from the Norwegian government. Currently being built by Vard for Yara Birkeland AS and due to enter service in 2020, initially with a crew, the all-electric ship is designed to be fully autonomous and could take to the waves without humans as early as 2022.
But what happens when something goes wrong, as it inevitably will? Regulations generally imply that a vessel owner or operator should have control of a vessel and have the ability to deal with mishaps when they occur. In fact, owners and operators have sometimes taken liberties with this vision, satisfying themselves — if not the authorities — by having a plan and the expectation that they can hire someone to help when a problem arises.
The new world of autonomy, though, raises the stakes. With the potential for no humans to be present to work their way out of a mess, the question of who or what is in charge becomes even more urgent.
“It is a major overhanging topic for 2019,” said Cole Callihan, a partner with the New Orleans-based legal firm Adams and Reese. “There are questions as far as litigation and liability and the application of Coast Guard regulations for these vessels.”
“There is an international group, including the U.S. Coast Guard, presently scoping the navigation rules to see if any changes are needed,” said Grady Hurley, lead attorney of the maritime litigation and arbitration team at Jones Walker LLP in New Orleans.
Hurley said if a collision occurred involving two autonomous vessels, the owner of the ship responsible would be liable just as if a captain were at the wheel. There may be arguments of both negligent operation and unseaworthiness if the artificial intelligence (AI) equipment “was not reasonable, fit and maintained,” he added. Likewise, the situation wouldn’t be all that different if there were a collision between a manned vessel and an autonomous vessel. The fundamental assumptions would be the same. Hurley said that many ships currently have AI features such as autopilot, so the problem isn’t entirely novel.
But the classic concepts of liability and fault may need refinement when human “operators” are potentially thousands of miles away from the vessel. Furthermore, the fault may turn out to be in logic or assumptions built into a computer program. Does the person who created the code then bear responsibility, and is that different from an unreliable radar unit or a failed shaft bearing, which also could cause a disaster?
Rolls-Royce’s Blue Ocean designers have come up with a range of concepts for autonomous cargo ships, including an LNG-fueled short-sea vessel. The early costs of insuring such ships is likely to be high until the technology proves reliable, experts say.
Someone is ultimately responsible and has made a mistake, but what is unknown is how that will be traced back. The current responsibility regime is focused on the ship and its crew as the primary players.
“From our perspective, I see as a first application, essentially phase one, things like transportation of cars on ferries on inland waters, or containers on short-sea voyages,” said Lars Gustafson, managing director and U.S. marine practice leader at Arthur J. Gallagher & Co., an insurance brokerage and risk management company headquartered in Rolling Meadows, Ill. And there is a big leap from there to phase two.
Larger vessels would need “a lot of sign-offs from marine insurance companies,” and the insurers would have to be convinced that the technology has been tested in real-life situations, Gustafson said. Those insurers thrive on accurate data and will insist on it before they agree to underwrite anything that is this new or of this magnitude, he added.
Yara Birkeland fits Gustafson’s description of a “phase one” project. The parent company of Yara Birkeland AS is a fertilizer manufacturer. Once in service, the ship is expected to eliminate up to 40,000 truck journeys per year to distribute the company’s bagged products.
“For early adopters actually moving onto the open sea, insurance costs will be high,” Gustafson said, drawing a comparison with the introduction of liquefied natural gas (LNG) tankers in the 1970s. “At the time, that was untested technology and the insurance premiums were very high, probably about eight to 10 times higher in relative terms than they are today.” Eventually the technology proved to be reliable and claims were few, so insurance become much more competitive, he said.
While regulation is an issue, Gustafson said the most crucial factor for determining whether autonomous vessels take to the waves and whether they become insurable is the role of the classification societies.
“The IMO (International Maritime Organization) represents the flag states of each country, (but) it is a division of the U.N. with an inexact legal standing because it is up to the individual states to accept or reject its recommendations,” he said.
On the other hand, it is the classification societies that will be looking at the technical side of the vessels and will sign off on the feasibility and legal ramifications. Of the dozen or so classification societies that examine 90 percent of the world’s maritime tonnage, Gustafson said the leaders for autonomous technology and vessel design have been DNV GL and the American Bureau of Shipping (ABS).
Before being tested outdoors, a scale model of Yara Birkeland plies SINTEF Ocean’s test tank in Trondheim, Norway, in September 2017. How the full-size ship and others like it handle scenarios outside a controlled environment is driving regulatory discussions at the IMO and elsewhere.
Lt. Amy Midgett, a spokeswoman for the U.S. Coast Guard, said autonomous technology is being discussed by various IMO committees that are beginning to wrestle with many of the potential implications. While there are no answers yet on the path toward regulation, the weighty importance of the questions is well recognized.
“When it comes to charting a course to develop standards and regulations, the advancement of vessel autonomy presents several new challenges for ensuring the continuity of safety and security in the marine transportation system,” she said. “These challenges include consideration of stakeholder concerns, the regulatory framework, industry best practices, international standards and eventually specific regulatory standards. These are complex challenges with many factors to consider in order to ensure the safe deployment and implementation of autonomous technology.”
While it is clear that the neither the Jetsons nor their robotic friends will be running the shipping business in the near future, changes will be coming. Even assuming that “computers do not make mistakes,” the liability picture changes very little.
“Vessels can be liable in rem,” said Hurley, citing the legal concept that assumes the property involved in an action is the primary focus, not a particular person. “There are presumptions of negligence under the law and strict liability.”
Thus, there can be negligence in programming or calibrating, or equipment failure. Likewise, there could be in error in programming, or various sensors malfunctioning could affect the data received. So, Hurley added, “A ship can be held liable in rem and its owners and operators in personam.”
“The maritime industry is flexible and has adjusted with advances in technology from oars to sails to propulsion, and from compass to autopilot and radar,” he said. “The basic rules of the road will stay in place and be adapted to technology.”
Callihan noted that shipping “is a traditional industry, so it is going to take time. Trucking will probably be the first with autonomy.”