Underwater robots swarm the ocean

Researchers tackle a big problem for tiny ocean robots

By Evan Lubofsky | August 21, 2019

(Illustration by Tim Silva, Woods Hole Oceanographic Institution)

Underwater robots do a lot these days. They can be programmed to go to remote, dangerous, and often previously unexplored parts of the ocean to measure its key characteristics—from salinity and temperature to the speed and direction of currents. They map the seafloor and benthic environments in outstanding detail. And, they find things—from shipwrecks and downed war planes to the billowing plumes of hydrothermal vents in the deep sea.

These smart workhorses of the ocean have become so useful, in fact, that some ocean scientists like Erin Fischell can’t get enough of them.

“Instead of using just a single, larger and more expensive underwater robot to cover an area of the ocean, we want to have hundreds or even thousands of smaller, lower-cost robots that can all work in sync to give us a more complete picture of what’s happening,” said Fischell, who develops autonomous underwater vehicle (AUV) technology at Woods Hole Oceanographic Institution (WHOI). “This will give us better spatial and temporal coverage in the areas we’re trying to study, and provide us with much richer and robust data sets in far less time.”

Getting in sync

For underwater robots to become the vast and coordinated ocean monitoring network that Fischell envisions, they need to form underwater ‘swarms’ that move en masse through the ocean—not unlike schools of fish. Swarming has already shown promise in disaster rescue missions and other applications on land, but we haven’t yet been able to extend the capability to the ocean due to basic physics. Radio signals, such as those generated by GPS navigation systems, travel at the speed of light. But the absorption of light in water is 10 trillion times greater than that in air, so radio signals can’t go very far underwater and thus aren’t viable for underwater communications.

To overcome these limitations, underwater robot navigation has traditionally relied upon different kinds of technologies such as high-power inertial navigation sensors (INS) that cost hundreds of thousands of dollars. Fischell says that while these pricey peripherals would be impractical for low-cost underwater vehicles, a robot attempting to navigate without INS technology can quickly run into severe issues.

“These low-cost ocean robots can drift hundreds of meters every 10 minutes, so if you leave one down there for an hour, it can end up being a kilometer off from where you think it is,” she said. “It doesn’t take long for them to lose their way.”

And that’s just a single unit. If you want a fleet of robots moving collectively in the same direction, the lack of accurate, controllable navigation becomes a show stopper.

Do you hear what I hear?

Fischell has been working with MIT professor Henrik Schmidt and Nicholas Rypkema, a researcher at MIT and former MIT/WHOI Joint Program graduate student, to bridge this capability gap with the development of a new acoustic-based navigation system. It includes a series of small, economical underwater robots known as SandSharks that eavesdrop on a pulsing underwater speaker—or acoustic beacon—that transmits sound into the ocean every second. The robots listen in with help of underwater microphones that pick up the acoustic signals, enabling the system’s control software to determine the distance and angle of each robot relative to the beacon. In this sense, the beacon acts as a common reference point for each robot, allowing them to navigate collectively in a swarm.

“Once we know range and angle, we can figure out where a robot is relative to the sound source,” Fischell said.

SandSharks on the Charles River Dock
Three ready-to-deploy SandShark robots on a dock before engineers launch them into Boston's Charles River. (Photo by Nick Rypkema, Massachusetts Institute of Technology)

Field proven

The system was recently field tested in Boston’s Charles River. A kayak acted as command central, with an acoustic beacon mounted off its side and a shoe-box sized control box placed inside the cockpit.

“The controls are very easy for the user—there’s a simple four-way switch for Follow, Sample, Return, and Abort,” Fischell said.

Three low-cost robots were launched from shore and cruised at roughly 2 meters below the surface. Once “Follow” mode was selected during the field tests, the robots began receiving the acoustic bleeps and moments later, all three were moving in the same pattern relative to the beacon attached to the kayak as it glided around the river.

“As the beacon moved though the water, all of the robots followed along, which made it easy for the user to understand what was going on,” Fishchell said. “We put strobe lights on the robots so we could see them moving in sync through the water.”

The system also offers a great deal of flexibility with respect to controlling the swarm. Depending on the mission, the robots can cruise in various lines, angles, and circles relative to the beacon, and line up together in virtual arrays.

“The robots’ course can easily be changed if the ocean features that are being monitoried—like fronts or plumes, for example—shift in space,” Fischell said.

Scaling up

The team appears to have solved a big problem for tiny robots—and one that oceanographers have grappled with for quite some time, according to Rypkema.

"Swarms of underwater vehicles have been a dream for researchers for decades, and the nature of the ocean means that these swarms can have an enormous impact on understanding its properties and dynamics,” he said.

Getting a fleet of three robots to work cooperatively is a huge win, but Fischell and her colleagues are thinking much bigger. They ultimately want a scalable network of hundreds of robots roaming the ocean.

“There are a lot of underwater robots out there in the sub-$10,000 price range,” Fischell said. “So as we optimize our technology and methods, we can hopefully get a lot of these vehicles working together for less cost than some of the larger and more complex robots we’ve had to rely on.”

This work is supported by the Office of Naval Research (ONR), Battelle, the Defense Advanced Research Projects Agency (DARPA) and the Reuben F. and Elizabeth B. Richards Endowed Fund at WHOI.