Tech & Innovation
TECH & INNOVATION
'Biomimetics' researchers inspired by the animal world
Animal kingdom inspires new breed of robots
By Scott Kirsner, Globe Staff, 6/9/2003
Marc Raibert's dog won't be able to sprint with the greyhounds at Wonderland, or enter the Westminster Kennel Club competition, or irrigate a rose bush. That's because Raibert's Great Dane-sized canine will have a chainsaw engine for a heart and will lap up gasoline instead of toilet water. What BigDog will be able to do, when he's born in 18 months, is run at about 15 miles per hour, surmount meter-high hurdles, and leap over ditches.
The project, now in the final stages of contract negotiations, is being supported by the Defense Advanced Research Projects Agency, the part of the Pentagon that earlier brought you the Internet.
Raibert, the president of Boston Dynamics Inc. and a former MIT researcher, stresses that BigDog is "pure research and development, really pushing where the science and technology [of robotics] can go over the next eight or 10 years." But further down the line, an advanced version of BigDog could serve as a soldier's best friend, carrying small loads -- like communications gear or ammunition -- and running ahead to sniff out chemical or biological danger.
BigDog is just one of several current robotics projects in Boston, funded by government research agencies, that look to the animal kingdom for inspiration. The hope is that nature will help spawn a new generation of robots that are more durable, versatile, and capable than today's machines. "When you watch a snake," says Joseph Ayers, a Northeastern University professor, "it seems simple for it to climb trees, swim, and get into small holes. But how many machines can do that?"
Some call the field "biomimetics," for the efforts to mimic biology. DARPA calls it "biodynotics" -- biologically inspired multifunctional dynamic robots. By either name, researchers are finding that even trying to duplicate the simplest of animals isn't easy.
At Massa Products in Hingham, a seven-pound lobster sits in the center of the conference table. Not only does this creature have the world's toughest shell -- made of an industrial-strength plastic -- but pound for pound, it is also the world's priciest crustacean. Over the last five years, several million dollars have been invested in the RoboLobster, developed in a collaboration between Massa Products and Ayers at Northeastern. Last month, Massa, which builds sonar systems for the Navy, won another $1.3 million contract from the Office of Naval Research to continue its development.
RoboLobster has eight legs, a tail, two claws, and even two clear plastic antennae, used to gauge the direction and strength of water currents. Its job will be to hunt for explosive mines buried in shallow water or on beaches, which pose a hazard to troops trying to land, D-Day style, in amphibious vehicles.
"At first, when we'd go to robotics meetings, we'd be laughed at," says Don Massa, the company's president. But he says that lobsters are well-designed for maneuvering in turbulent, shallow water and on the beach. "Nature is the world's greatest designer," Massa says. "All the bad ideas die off, and the good ones live on. For example, lobsters automatically adjust their posture and position in order to stay put in roiling waters.
But developing the control software that will enable the RoboLobster to navigate and avoid obstacles -- never mind looking for mines -- is a tougher problem to crack. "Making a robot move in the lab is a whole lot different from making it move in the real world, where there are people and obstacles and other things that you can't anticipate," says Jordan Pollack, a robotics researcher at Brandeis University.
One advantage those following a biological example have, though, is that they can turn to real animals for help. Ayers, who is developing the control software for the RoboLobster, uses live lobsters as assistants.
"We have a big outdoor pool in Nahant," he says. "This summer, one thing we'll do is put the robot in a situation where it's surrounded by a field of rocks. If it can't get through, then we'll take a real lobster, and put it in the same situation. We can see how it solves the problem, then build that into the [robot's software]."
(Ayers isn't reluctant to eat his assistants once they've reached the end of their useful life; Massa says Ayers makes "an incredible lobster bisque.")
Massa imagines the RoboLobster being dropped into the water from helicopters, or tossed over the side of a rubber raft piloted by Navy Seals, the service's special operations personnel. "The last thing you want," he says, "is for these highly trained Navy Seals to be blown up looking for mines."
The lobster, in its final version, might be able to drop an ultrasonic "pinger" on the mine, to let humans know where it is. Or the lobster could act as a kamikaze, carrying a pound of plastic explosives inside its shell. Once it identified a mine, it would position itself as close as possible and then detonate, turning instantly into robotic lobster salad -- and taking the mine with it.
Ayers thinks RoboLobster could also prove useful for marine biologists studying lobster behavior, and possibly even for lobstermen looking for the most productive places to set traps. (Although it's more fun to imagine a renegade RoboLobster freeing its brethren from their underwater jails.)
Other local companies, like Icosystem in Cambridge and iRobot in Burlington, are studying "swarming" behavior -- the way that birds, ants, or fish travel together and collaborate on tasks like finding food. The aim is to develop software that would guide large groups of robots on particular missions. These robots might be flocks of drone aircraft or packs of BigDogs.
"To explore a room and look for some kind of danger, or to look for survivors after an earthquake, you might want to have hundreds of robots working together like a swarm of bees or a colony of ants," says James McLurkin, an MIT grad student who also works at iRobot.
Biomimetics is still unproven as a viable approach -- though the current crop of projects could change that. Today's robots, like the tank-like PackBot made by iRobot and deployed in Iraq and Afghanistan on reconnaissance missions, don't look much like animals. And today's unmanned airplanes, like the Predator and Global Hawk, can fly higher and faster than birds -- even though they don't flap their wings.
Raibert at Boston Dynamics believes that biologically-inspired machines will eventually find a useful niche.
While at MIT, Raibert developed some of the world's first robots capable of running and jumping. And DARPA is on the verge of handing Raibert and his research partners several million dollars to work on BigDog over five years. In addition to the military applications, we could see future robotic hounds being purchased for entertainment, like the Sony Aibo, or even as working dogs for the handicapped.
"A large part of the earth's surface is inaccessible to vehicles that have wheels or [tank treads]," he says. "Yet people and animals can go to all of those places."
Raibert's long-term goal? Teaching BigDog to catch a Frisbee.
He's not kidding.
@Large appears every Monday. Send tech news tips to Scott Kirsner at email@example.com.