Robot Fish Could One Day Conquer Seas

June 10, 2008 — -- If it swims like a fish and acts like a fish, then it must be a fish, right? Nope, it's robofish, the newest contraption in a lengthening list of electro-mechanical devices based on the secrets of the best designer of all: nature.

Scientists and engineers aren't exactly stealing from nature. They're just borrowing from some of its greatest hits.

Kristi Morgansen, assistant professor of aeronautics and astronautics at the University of Washington, calls it "bio-inspired." As an undergrad she had researched "bio-inspired locomotion," and after earning her doctorate she looked around for a "fun application of things I had been doing," she said in an interview.

One thing led to another, and now she has a tub of water with three mechanical fish that can do something that some engineers thought might be impossible. They can "talk" to each other, under the water, and even coordinate their activities, like swimming together or swimming apart.

It's no minor achievement because other robotic fish have to surface from time to time. Not for oxygen, like real fish, but to send back their data and get new instructions. Morgansen's robofish can communicate with each other underwater using acoustic signals that can actually transmit data through pressure waves. She got that idea from nature, too. She teamed up with fisheries scientist Julia Parrish to study how fish actually communicate with each other in the real world.

That led ultimately to the creation of her mechanical menagerie, consisting at this point of just three robofish, each about the size of a 10-pound salmon (she doesn't have enough room for more). Some day, she said, her school of fish could number in the hundreds and they could be sent off to the far reaches of the planet to monitor pollution, keep tabs on undersea volcanoes and, of course, track enemy submarines.

There's lots to be learned first. Her robofish can maintain their structural integrity only down to about eight feet, as far as is known, because that's the depth of her laboratory tub. To be really useful, they will need to work at much greater depths, and that's a bit of a challenge. A leaky robofish will soon be a dead robofish.

Her creatures don't look like real fish, although some other researchers have accomplished that. But they work like real fish. Sort of. Instead of a propeller, which is noisy, inefficient and easily snarled in weeds and ropes, her fish are pushed along by a fin that flaps back and forth. They were patterned after a kind of fish, including the speedy tuna, that uses the rear third of its body for locomotion while the forward two-thirds remains fairly rigid.

"From an engineering point of view that's helpful because you can put all your electronics and stuff up in a rigid compartment and you don't have to worry about it being flexible, which is harder to engineer," she said.

Her robofish take turns "talking" to each other.

"When it's someone's turn to talk it will send out a signal which will usually have an identifier of what vehicle it is, what time it is, and the date," she said. "It could also include its position or information about some object they are tracking. So the information gets sent out and anybody that's close enough can decode that data and use it. They take turns doing that."

That progress makes it possible for Morgansen's fish to accomplish what might seem like a pretty simple task at this point. They can either swim together, or swim in different directions. So it's not yet ready for prime time, but it's a significant advance. They are also able to do some tracking.

In the future, she said, her swarm could include different fish with different capabilities, thus broadening their application. It could be a school of hundreds, but the problems get tougher as the numbers go up.

"It will take longer for them to converge on the answer (where is that submarine?) because they have to exchange that much more information," she said. "There are going to be limits in what is computationally feasible," and they will need a better source of power than the hobby-shop batteries that are used now.

And she still has to return to nature for more info. Morgansen and her colleagues trained some live fish to respond to a stimulus by swimming to a feeding area. It turns out that all the fish don't have to know what to do. Some apparently are leaders, others followers, so if even a few of the fish knew it was time to head for the lunch counter, all of them made it.

"That has implications for what will happen in a group of vehicles," she said. "Can one vehicle make the rest of the group do something just based on its behavior?"

Her robofish suggest that's most likely the case. Some of the fish received only about half of the data that was sent, but they were still able to accomplish their task of swimming together, or swimming apart.

In a couple weeks, Morgansen and her students will give their robofish a serious challenge. They will be asked to track a remote controlled toy shark, like the ones available in the toy section of your neighborhood mall.

It will be a modest test of their tracking abilities. Really modest in the beginning because of the size of the tub in Morgansen's lab.

"We probably will not have enough room," she said. "That's the reason we have just three of them. We will try to get into the university's swimming pool."

She's not yet ready to try her luck in nearby Lake Washington.

"We know we can keep them from leaking at eight feet, but there's not been much done to optimize them for anything deeper than that," she said. "And the lake is really, really deep. They are a little expensive and difficult to replace, so we don't want to lose any of them."

But some day, no doubt, the seas will be their oyster.

Lee Dye is a former science writer for the Los Angeles Times. He now lives in Juneau, Alaska.