Uncovering secrets at the bottom of Earth’s oceans
How a new user interface developed by MIT researchers in collaboration with other institutions can help scientists operate deep ocean submersibles with ease
From sharks that can live for up to 400 years to microbes that could be the evolutionary ancestors of all complex life on Earth: if we’ve learned anything about Earth’s oceans in the past decade, it’s how little we really know. When each year brings a new and unexpected find, one begins to wonder: how much are we missing?
The Shared Autonomy for Remote Collaboration (SHARC) framework could go a long way towards helping scientists answer that question. Developed by researchers at MIT, the Woods Hole Oceanographic Institute (WHOI), and the Toyota Technological Institute of Chicago (TTI), SHARC offers scientists a new, user-friendly means of controlling underwater Remotely Operated Vehicles (ROVs) from afar. ROVs are the mainstays of deep-water oceanography, operating like drones tethered to their parent vessels by an umbilical cord of wires transmitting power and instructions. ROVs allow scientists to collect data from depths that could crush the human body like a tin can due to the water pressure.
They also happen to be terribly expensive and difficult to use. Most ROV operators, or pilots, must train for years before they’re considered qualified to command one of these machines. For busy scientists, ROV control is a skill they don’t have the time to learn, resulting in a bottleneck in the research pipeline, a wall that separates oceanographic scientists from their best research tool. SHARC’s designers, hoping to make an immediate dent in that wall, chose to focus on a small but high-impact aspect of ROV operation: the arm.
Traditionally, a pilot controls the arm with a joystick, similar to a claw machine at a carnival. An ROV’s arm, however, is jointed with several elbows to improve its range of motion. That makes even primitive actions — like picking up a tool — immensely complicated. In an intricate game of domino, an ROV pilot must use “switch mode control” to toggle between joints along the arm, carefully nudging each joint to achieve the desired overall motion. It’s a process known as “inverse kinematics,” and it can turn five seconds on land into five minutes underwater. But when each day a research vessel spends at sea costs a hundred grand or more, “time is money,” says Matthew Walter, Associate Professor at TTI and one of the SHARC developers.
SHARC smooths away this inefficiency by offloading inverse kinematics problems to the ROV’s onboard autonomy. Instead of directing the arm via joystick, pilots assign tasks to the ROV via written and spoken messages—or gestures when using SHARC’s virtual reality interface. Almost instantaneously, they receive a projection of the arm’s requested path. Then, upon the pilot’s approval, the ROV autonomously executes the motion.
SHARC so radically flattens the learning curve that Amy Phung, a candidate in the MIT-WHOI joint Ph.D. program and one of SHARC’s primary developers, says, “it felt like playing a video game. [Only] this ‘video game’ was happening in real life, and the sample I was ‘looking at’ digitally was actually a real sample with scientific value.”
By lowering the barrier for entry, SHARC opens the door to a future where ROVs, and with them the deep ocean, can become “readily available to users worldwide via the internet,” Walter said. Walter imagines a world where student groups watch on the big screen while scientists across the globe provide annotations and suggest mission changes in real-time as the experiment progresses.
Dr. Andrew Bennett, a member of the Ocean Engineering faculty at MIT who was not involved with SHARC, is emphatic about the technology’s potential. “Oil companies sometimes operate ROVs in deep water, and every now and then they come across something strange and they’ll send it to the biologists, and the biologists will go crazy,” he says. “And had they been there, in control of the ROV, we might’ve seen more.” Walter agrees: by allowing scientists to intervene almost immediately, SHARC could get pilots to “take that sample that they might not have taken otherwise.”
When asked if he thinks SHARC is a breakthrough for engineering or accessibility, Bennett says he thinks it’s both. To Bennett, SHARC solves a recurring problem in our fast-accelerating world: “We’ve got the processing power, and we’ve got the sensors; people just couldn’t keep up with the potential.” SHARC could be just the thing to close that gap. “We’re going to discover things we never even realized existed.”