Using lessons learned from harbor seals and artificial intelligence, engineers in California may be on to a new way to track enemy submarines.
The idea started with research published in 2001 on the seals.
Scientists at the University of Bonn in Germany showed that blindfolded seals could still track a robotic fish. The researchers concluded that the seals did this by detecting the strength and direction of the whirling vortex the robot created as it swam through the water.
Subsequent research showed that the seal used its whiskers as sensors to detect the flow patterns.
Eva Kanso, a professor of aerospace and mechanical engineering at the University of Southern California, is interested in how animals use water flows to guide their behavior. It's an academic puzzle for Kanso, but a very real, very practical question for a harbor seal.
"The animal wants to understand — is it a prey that created this vortex, or is it a predator that created this flow pattern?" she says.
Kanso and her colleagues have been trying to emulate the seals' ability to make those distinctions.
At first, she tried to do this using the physics of flow patterns, but she didn't get very far — the physics is complex and computationally intense.
So she turned instead to artificial intelligence. Just like a computer can be trained to recognize an apple from a collection of pixels in an image, she and her colleagues have trained a computer to recognize various objects based on the flow patterns they create in the water.
For now, the computer program isn't all that smart. It can only sense fairly simple patterns. But Kanso expects it will get better.
And she acknowledges that the software could be used for more than understanding harbor seals' behavior.
She can imagine a day, she says, when the tool could be useful to submarine hunters, since submarines, too, can leave distinctive flow patterns.
It's probably not a coincidence that the Navy helps pay for Kanso's research.
ARI SHAPIRO, HOST:
Engineers in California have a new idea for how to track enemy submarines. Their idea was inspired by lessons learned from research on artificial intelligence and harbor seals, as NPR science correspondent Joe Palca reports.
JOE PALCA, BYLINE: Just about anything moving through the water creates a flow pattern called a vortex. Ask Eva Kanso what a vortex is, and you get a pretty simple answer.
EVA KANSO: It's a fluid that's moving around. That's what I mean by a vortex.
PALCA: Kanso is a professor of aerospace and mechanical engineering at the University of Southern California. So let's say you see a vortex swirling in a murky pool of water.
KANSO: How would you know what is it that created this vortex?
PALCA: That's an academic question for Kanso, and she's spent some time working on the problem. But she says it's a very real, very practical question for a harbor seal.
KANSO: The animal wants to understand. Is it a prey that's created this vortex, or is it the predator that created this flow pattern?
PALCA: Turns out seals can figure that out. They can even track a tasty fish by the vortex it generates. A seal doesn't use its eyes to see the vortex. Instead, there's evidence it uses its whiskers as sensors that can measure the shape and intensity of the vortex. Now, Kanso is an engineer, not a marine biologist, but she figured there were lessons to be learned from the seals.
KANSO: Going from local measurement, from what you can measure where you are with your senses, to extract information globally about the whole flow field - that's the question that's interesting to us.
PALCA: In other words, she and her colleagues wanted to know how much they could deduce about what's moving around in a broad swath of water just from the flow patterns. Her approach was to use artificial intelligence. Just like a computer can be trained to recognize an apple from a collection of pixels in an image, she and her colleagues have trained a computer to recognize various objects based on the flow patterns they make in the water.
For now the computer program isn't all that smart. It can only make sense of simple patterns. But she expects it will get better, and she sees a day when it could be useful for submarine hunters since submarines can leave distinctive flow patterns.
KANSO: So they would be able to say, OK, I think there was a submarine that was passing through this location at this speed.
PALCA: And maybe someday the computer could provide even more information.
KANSO: OK, now that I sense this flow pattern and I can kind of guess who created it - and I can track that flow to its source.
PALCA: Being able to track submarines this way is years off, but there's evidence someone believes it might be possible. The Navy helps pay for Eva Kanso's research. Joe Palca, NPR News. Transcript provided by NPR, Copyright NPR.