Oregon State University

College of Earth, Ocean, and Atmospheric Sciences

Robots That Think Like a Fish

OSU glider deployment, Yaquina Head

Deploying an undersea glider at Yaquina Head, Oregon

February 24, 2015

By Mark Floyd, 541-737-0788; mark.floyd@oregonstate.edu

Source: Jack Barth, Kelly Benoit-Bird, Geoff Hollinger

CORVALLIS, Ore. – Oregon State University researchers have received a $1 million grant from the W.M. Keck Foundation that will allow them to outfit a pair of undersea gliders with acoustical sensors to identify biological "hot spots" in the coastal ocean.

They also hope to develop an onboard computing system that will program the gliders to perform different functions depending on what they encounter.

In other words, the scientists say, they want to outfit a robotic undersea glider to “think like a fish.”

"We spend all of this time on ships, deploying instrumentation that basically is designed to see how ocean biology aggregates around physical features – like hake at the edge of the continental shelf or salmon at upwelling fronts,” said Jack Barth, a professor in OSU's College of Earth, Ocean, and Atmospheric Sciences and a principal investigator on the project. "But that just gives us a two-week window into a particular area.

"We already have a basic understanding of the ecosystem,” Barth added. "Now we want to get a better handle of what kind of marine animals are out there, how many there are, where they are distributed, and how they respond to phytoplankton blooms, schools of baitfish or oceanic features. It will benefit a variety of stakeholders, from the fishing industry and resource managers to the scientific community.”

Barth is a physical oceanographer who knows the physical processes of the coastal ocean. He'll work with Kelly Benoit-Bird, a marine ecologist, who specializes in the relationships among marine organisms from tiny plankton to large whales. Her work utilizes acoustics to identify and track animals below the ocean surface – and it is these sensors that will open up a new world of research aboard the gliders.

"Our first goals are to understand the dynamics of the Pacific Northwest upwelling system, find the biological hotspots, and then see how long they last,” Benoit-Bird said. "Then we'd like to learn what we can about the distribution of prey and predators – and the relationship of both to oceanic conditions."

Using robot-mounted acoustic sensors, the OSU researchers will be able to identify different kinds of marine animals using their unique acoustical signatures. Diving seabirds, for example, leave a trail of bubbles through the water like the contrail left by a jet. Zooplankton show up as a diffuse cloud. Schooling fish create a glowing, amoeba-shaped image.

"We've done this kind of work from ships, but you're more or less anchored in one spot, which is limiting," Benoit-Bird said. "By putting sensors on gliders, we hope to follow fish, or circle around a plankton bloom, or see how seabirds dive. We want to learn more about what is going on out there."

Programming a glider to spend weeks out in the ocean and then "think" when it encounters certain cues, is a challenge that falls upon the third member of the research team, Geoff Hollinger, from OSU's robotics program in the College of Engineering. Undersea gliders operated by Oregon State already can be programmed to patrol offshore for weeks at a time, following a transect, moving up and down in the water column, and even rising to the surface to beam data back to onshore labs via satellite.

But the instruments aboard the gliders that measure temperature, salinity and dissolved oxygen are comparatively simple and require limited power. Using sophisticated bioacoustics sensors that record huge amounts of data, and then programming the gliders to respond to environmental cues, is a significant technological advance.

"All of the technology is there," Hollinger said, "but combining it into a package to perform on a glider is a huge robotics and systems engineering challenge. You need lots of computing power, longer battery life, and advanced control algorithms."

Making a glider "think," or respond to environmental cues, is all about predictive algorithms, he said.

"It is a little like looking at economic indicators in the stock market," Hollinger pointed out. "Just one indicator is unlikely to tell you how a stock will perform. We need to develop an algorithm that essentially turns the glider into an autonomous vehicle that can run on autopilot."

The three-year research project should benefit fisheries management, protection of endangered species, analyzing the impacts of new ocean uses such as wave energy, and documenting impacts of climate change, the researchers say.

Oregon State has become a national leader in the use of undersea gliders in research to study the coastal ocean and now owns and operates more than 20 of the instruments through three separate research initiatives. Barth said the vision is to establish a center for underwater vehicles and acoustics research – which would be a key component of its recently announced Marine Studies Initiative.

The university also has a growing program in robotics, of which Hollinger is a key faculty member. This collaborative project funded by Keck exemplifies the collaborative nature of research at Oregon State, the researchers say, where ecologists, oceanographers and roboticists work together.

"This project and the innovative technology could revolutionize how marine scientists study the world's oceans," Barth said.


See more Feature Stories