Despite many developers’ efforts to teach cars to steer themselves around roads filled with human drivers, pedestrians, and cyclists, the first great wave of autonomous vehicles may not arrive on land. Instead, it might follow the time-honored tradition of running away from tricky problems by heading for the open seas.
Rather than worrying about robo-taxis, groups around the world are working on making ships—large and small—self-piloting, which could save fuel, prevent expensive accidents and groundings, and relieve crews of some of mundanity of life on the water.
Let’s start at the big end of the scale. Container ships and other behemoths may not have to deal with rogue scooters, but they come with their own challenges. Stopping or turning can take miles, so decisions have to be made early, and correctly. Tesla might hope just eight cameras is enough to see its surroundings, but a 1000-foot-long vessel needs hundreds. That requires a serious upgrade in computing power: Four of those cameras can generate 3 terabytes of data a month. So robo-ships don’t have trunks packed with computers. They have server rooms.
The good news is that adding even a few hundred cameras won’t much change the price of a ship that costs millions of dollars, and even the modest investment should pay off. “Six to seven percent of operating costs is crew,” says Thiru Vikram, the CEO of Buffalo Automation. “Eighty percent of accidents are caused by human error, or fatigue, so we can bring down insurance costs, save fuel, and reduce operating costs.”
Buffalo Automation, a startup with close ties to Buffalo University in New York, is developing autonomous boats and ships. Since 2015, it has equipped boats of varying sizes, from a small catamaran to large shipping vessels, with lidar, heat-sensitive cameras, and GPS. It started by making the equivalent of Tesla’s Autopilot, where a computer handles the steering, thruster controls, and propeller pitch, with a human supervising and ready to take control if something goes wrong. Since then the team has progressed to speed boats that don’t need that human backup, at least in certain, constrained conditions. Cameras around the boat remain vigilant for other water users, logs, kayakers, and swimmers.
The startup is now proving its system can handle ships as long as 800 feet in tricky situations. That includes testing its ability to get through a series of locks, and on Cleveland’s well-trafficked, narrow Cuyahoga River.
Some researchers at MIT are thinking smaller. They have designed a 3-D-printable, 13- by 6-foot boat they think could ferry people or goods around cities with rich waterways like Amsterdam or Bangkok. “Imagine shifting some of infrastructure services that usually take place during the day on the road—deliveries, garbage management, waste management—to the middle of the night, on the water, using a fleet of autonomous boats,” says MIT’s CSAIL lab director, Daniela Rus.
The team believes their boats could also use their onboard GPS sensors and inertial measurement units to precisely position themselves in packs, forming instant floating bridges, or stages, or platforms for pop-up food markets on the water. It’s working on the control software needed now, with tests of a smaller vessel on the Charles River in Massachusetts.
The US Navy is also investigating the usefulness of swarming autonomous boats to patrol coastlines or surround a hostile vessel. It’s developing a system that would allow autonomous vessels to talk to each other and collectively decide on the best approach to block a boat’s path or nudge it clear of a restricted area.
Now, automating the way boats move doesn’t mean jobs on container ships will be going away any time soon. “The crew doesn’t focus solely on navigation. Most of what they do today is ship maintenance and things like emergency firefighting,” says Buffalo Automation’s Vikram. “From our perspective, this is a complementary technology.”
He believes the first commercial application could be autonomous tugboats, which currently need crews of up to five people, to help maneuver large ships into tight docks. That’s the sort of task a computer, with appropriate sensors, could handle easily.
So while companies like Waymo, Tesla, General Motors, and dozens of others vie to put robo-rubber on the road, don’t be surprised if some developers set sail. They may not save huge numbers of lives, but dominion over two-thirds of the planet isn’t a bad consolation prize.