Confidence in the future of self-driving cars took a hit earlier this year when a robotic Uber mowed down a woman as she crossed a street in Tempe, Arizona. The ride-sharing service shut down its test program in the state after the March fatality as debate erupted over the technology and how to regulate it.
In the scheme of things, Tempe was a blip in the robot revolution, even with the loss of life. Just last week, in fact, Uber CEO Dara Khosrowshahi announced she still has confidence in the company’s driverless future, and that testing will resume this summer.
If the idea of being in a car with no driver gives you agita, researchers at the Massachusetts Institute of Technology have a plan that might make you seasick—captain-free boats. They’ve developed tech to run autonomous boats that could be used as water taxis or move cargo around urban waterways.
“Imagine shifting some 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 Daniela Rus, director of MIT’s Computer Science and Artificial Intelligence Lab.
Rus is a coauthor of a paper on the boat project presented last month at the IEEE International Conference on Robotics and Automation in Brisbane, Australia. The rectangular-shaped seafaring robots, equipped with computing hardware and GPS systems for navigation, would be able to link up like a freight train for bigger hauls or even to form temporary bridges or platforms for events like festivals or concerts, the researchers say.
The research is part of a project called Roboat, run by MIT’s Senseable City Lab and the Amsterdam Institute for Advanced Metropolitan Solutions. The researchers previously tested an autonomous boat that was able to cruise Amsterdam’s famed canals.
In some respects, developing autonomous boats may be easier than land vehicles, since commercial vessels already carry transponders to signal their positions to other ships, and a conservative approach can involve effectively steering around anything that’s not water.
“I suspect it’s still an easier problem than a self-driving car, because there is a certain degree of uniformity in the water’s surface,” says Paul Kamen, a naval architecture consultant who has written about urban ferry design. “Just to be conservative about it, anything that disrupts the uniformity of the surface—allowing for the known variations with waves or such—anything that’s different generally needs to be avoided.”
But it’s likely that self-piloting technology would be first used to reduce ferry crew sizes and costs through enabling limited features like self-docking rather than immediately eliminate crews entirely, he says. One challenge is that ship and boat crews are traditionally expected to offer rescue to others experiencing emergencies in the water. That, and other unexpected situations, would likely be a challenge for all but the most powerful AI systems.
“There are too many cases where people have to put a wrench on something,” says Christopher Barry, chair of the small craft technical and research committee at the Society of Naval Architects and Marine Engineers.
And if autonomous boats are used to transport cargo, humans and physical infrastructure would likely still be needed to load and unload it, he says. And autonomous boat operators would still have to deal with the politics, infrastructure, and land use issues regulating where passengers and goods could be loaded and unloaded. Owners of property on or near the water might be less than thrilled by an increase in traffic, or find themselves adjacent to a new miniature ferry or cargo terminal.
So far, the MIT scientists developed a prototype that could be produced in about 60 hours using a commercial 3-D printer, then sealed with a few layers of fiberglass. They tested a small prototype of the boat in a pool and in the Charles River, which runs alongside MIT’s campus in Cambridge. The boat was able to track its position down to the centimeter—more accurately then other autonomous surface vessels.
“The controller considers the boat dynamics, current state of the boat, thrust constraints, and reference position for the coming several seconds, to optimize how the boat drives on the path,” says MIT postdoc student Wei Wang. “We can then find optimal force for the thrusters that can take the boat back to the path and minimize errors.”
They plan to further refine their algorithms to deal with weight changes from carrying people and cargo, and with stronger currents and waves that could affect the vessels’ navigation.