Communications of the ACM
Autonomy and Automation on the Water
The Roboat is a floating platform designed to autonomously cruise the canals of Amsterdam and dock to the canal walls, or other Roboats, as needed.
Credit: MITAMS Institute
Most of the talk about autonomous vehicles has focused on land (self-driving cars, and even snowplows) and air (aerial drones). However, there's plenty of activity on the water, too, as researchers and businesses are developing boats of all sizes that can maneuver autonomously, by remote control, or take over some duties from a human crew.
These vessels range from "floating platforms" the size of a small room to cargo ships 1,000 feet long.
Stepping stones
The floating platforms are the product of aa joint project of the Amsterdam Institute for Advanced Metropolitan Solutions and the Massachusetts Institute of Technology (MIT). "The idea initially was to advance autonomous cars," recalls Fábio Duarte, a research scientist in MIT's Senseable City Lab. "Then we wondered, why would Amsterdam deploy driverless cars if the city has so many bicycles and more than 100 kilometers of canals?" The result was the Roboat, a flat vessel that can not only move people and goods, but also join with other Roboats to create bridges and stages as needed.
Initial Roboat models for testing measured 1 meter by 1.5 meters. The group is now working with models twice that size, and they hope to deploy full-size 4-x-2-meter platforms in January 2020.
The early models are still useful, says Duarte. "With the quarter-scale boat, we've been working on multivessel coordination. Say you have five boats and need them to form a square with one in the center. There are two approaches: one is that each boat is smart and decides where to go, and the other is that only one boat is smart, and it becomes the master and decides how to move each worker."
The platforms have four propellers each, two for moving back and forth and two smaller ones for lateral movements. For navigation, they use a layered approach, Duarte says. "With GPS, it finds the shortest path from point A to point B," explains Duarte. "After that we have a second layer, AIS," the Automatic Identification System, which relies on transponders; all boats in Amsterdam longer than 15 meters already have them, and they're now being installed on all boats entering the canals. "With this technology, the boat can see the traffic in all the canals and find the optimal path, not necessarily the shortest one," Duarte says
"A third layer uses a combination of LiDAR and computer vision to measure possible obstacles and classify them." If the obstacle is an approaching boat, for example, the Roboat determines whether it needs to move right or left to avoid a collision.
A job to do
Further up the size scale are workboats suited to tasks like research, surveying, search and rescue; "dull, dangerous, and dirty jobs," in the words of Travis White, research engineer at the Michigan Technological University Great Lakes Research Center and administrator of the Smart Ships Coalition.
"There are quite a few developers making vessels less than 10 meters in size that are autonomously controlled," says White. "For example, there's the C-Worker 5 made by ASV Global. It's a five-meter fully autonomous vehicle, and it's being used by the oil and gas industry to collect bathymetric data and hydrological data in real time."
Such boats are often used in conjunction with a larger manned vessel as "force multipliers," White says. For example, a crewed survey ship might deploy smaller remotely operated vehicles to increase their coverage area. "That's probably the biggest applications for an autonomous vehicle right now," says White.
Boston-based Sea Machines focuses on this category of ships, too. Sea Machines doesn't build ships, but develops command and control systems that can be installed on existing vessels such as tugboats, fireboats, ferries, and patrol boats. The systems use proprietary technology to connect a vessel's propulsion machinery with its navigation sensors, enabling the ship to follow a preset route, collaborate with other vessels, and detect and avoid obstacles.
"The SM200 would allow remote control of patrol boats, fireboats, anywhere it could help to remove crews from danger," says Amelia Smith, Sea Machines marketing and communications consultant. "The SM300 includes that remote control, plus it has an autonomous command system that allows a shoreside operator or someone on a second vessel to operate the boat from a laptop." The operator can call up an Electronic Chart Display and Information System on a laptop and set "waypoints"—navigational markers akin to pins set on Google Maps—to send the boat in a search grid pattern, or to run back and forth to an offshore aquaculture farm.
For navigation, the Sea Machines systems use input from GPS and AIS, as well as data from infrared and RGB cameras. The company's high-end system, the SM400, also will use LiDAR, says Smith, who adds that the idea is not to take over navigation from a human crew, but to support them. "Humans have difficulty with multitasking. If you've got five, six, seven data feeds all coming in at the same time, a human operator may have difficulty interpreting all of that. A computer can process that data immediately and provide a recommendation to the operator."
Carrying capacity
Buffalo Automation began as a class research project in 2015 at the University of Buffalo. The students decided to focus on watercraft because so much work was already being done on cars and aerial drones. "As the technology got more robust, we decided to bring it to market and start a company," says Thiru Vikram, Buffalo Automation's CEO. "Originally the plan was to sell the product to recreational boaters, but people buy those to drive them, so we transitioned to cargo ships and other bigger vessels. We've got lots of companies here in Buffalo that move cargo, and we saw a very good opportunity."
Ships using Buffalo Automation systems are actively engaged in trade, carrying iron ore, cobalt, and the like. Such ships present different technical challenges than cars and smaller boats, says Vikram; in some respects, easier ones. This type of vessel is "a very large, slow-moving object with a lot of inertia," he says. "Most of the time it's in open water, so the hard parts are when it gets close to port or other crowded areas, and even then, it's so slow that a lot of the challenges that you have with cars simply don't exist."
The systems are installed on existing ships and provide navigation assistance and suggestions, relying on multiple technologies. "With traditional marine autopilots, there's a static map, and then you have GPS to localize within that map," says Vikram. "The big ships also use AIS. We're using all that, and radar lets us update the map in real time with things the static map missed, or new objects such as buoys and other watercraft. LiDAR is more useful for a closer range, like when you're trying to dock." These systems also include cameras, in order to identify the detected objects.
Similar to Sea Machines' approach, Buffalo Automation's system is designed to help a human crew navigate more safely, rather than to take over navigation entirely. The company's current goal is to build a record of reliability for the technology by comparing the suggestions of the assistive system with the decisions the humans chose to make. "Our metrics are aligned towards whether there were any incidents, or if the ship was able to carry more cargo or was it easy to dock," says Vikram. "It really is the experience of the crew that's the defining metric."
Speedbumps and benefits
As with any emerging technology, developing autonomous watercraft presented some unexpected obstacles. In Amsterdam, the Roboat team didn't anticipate how hard it would be to latch two of their platforms together. "When we're docking a boat to a wall, we have a fixed element," says Duarte. "The waves hit the wall and come back, so it's quite complicated, but naval engineering has already solved the problem. Latching two boats, both of them moving against the other, is the most challenging task that we're facing."
For Sea Machines, the surprise lay in the obstacle detection and collision avoidance; it worked too well. "Depending on the sensitivity, radar can pick up waves as an obstacle," explains Smith. "It took hours of experience on the water to find the right threshold. Now the engineers create a 'guard zone' directly around the boat where it ignores waves, so it doesn't react to its own wake."
The technology also holds the promise of benefits beyond safer navigation and relieving crews of repetitive or dangerous tasks. On a cargo ship, for example, "having fewer people on board means that you can take as long as you want to go on a particular journey," says Vikram. "You don't need all the life-support systems." Also, a slower trip means less fuel consumption and fewer emissions.
Some worry that increasing automation will cost jobs, but White disagrees. "It's augmenting what the mariners are already doing," he says. "It's not so much a career-displacing technology, but a career-enhancing technology coming here in the U.S."
However, White continues, it will change the nature of maritime careers. "I think you're going to see automation happening in places like the engine room; taking the operator out of the engine room and putting them behind a computer, where they're monitoring and adjusting electronically, as opposed to physically, in the engine room. And there might a remotely operated ship track on the mariner's academic program."
Jake Widman is a San Francisco, CA-based freelance writer focusing on connected devices and other Smart Home and Smart City technologies.
No entries found