Slinky caterpillars usher in a metamorphosis in robotics
by Michael Blanding

Neurobiologist Barry Trimmer cradles the inspiration for his robotics research. © TOM KATES

“Nobody really knows how soft-bodied animals move around,” says Tufts neurobiologist Barry Trimmer, sitting on a rotating stool before a shelf filled with silicone models of larvae. That’s because invertebrates like caterpillars don’t have joints and bones for their muscles to work against. Instead, dozens of muscles crisscrossing beneath their soft skin must each be flexed to the perfect degree at precisely the same time, a job that takes incredible computational skill.

“When you tell engineers to make something completely without joints and to control it precisely, they laugh at you and say it can’t be done,” says Trimmer, an affable Brit with a hawkish nose and close-cropped hair. “And I say, well, the caterpillars do it, and they are the most successful herbivores on the planet. So there must be an answer.”

To find that answer, Trimmer has been building robotic caterpillars in Tufts’ Advanced Technologies Laboratory near the Medford/Somerville campus. So far they exist only in spongy bits and pieces and in the form of a nascent virtual caterpillar on a computer screen. But a working “soft robot” would leave today’s clunky mechanical robots—encumbered with motors and sensors that inhibit easy movement through real-world environments—in the dust. The project recently received a $730,000 grant from the W.M. Keck Foundation that will provide researchers with specialized equipment for their work in soft materials and biomechanics.

Supple soft bodies
Caterpillars are amazingly versatile, able to crawl across flat surfaces, burrow into the ground, scrunch themselves through small holes and scramble through tree branches. “These soft-bodied animals are able to move in a really elaborate three-dimensional environment safely and accurately,” says Trimmer. “We haven’t been able to create robots or machines that do that.” He imagines a host of real-world applications for the kinds of critters he’s constructing, including threading through rubble in search-and-rescue missions, crawling down veins in medical procedures and squirming around battlefields to identify landmines.

Trimmer’s creations fall into the burgeoning field of biomimetics. In short, that means duplicating natural forms by artificial means. Most biomimetic robots conceal conventional components within animal-shaped bodies. Trimmer stands virtually alone in creating an animal that is not only “biologically inspired” but also “biologically informed.” He’s trying to recreate the animal’s movement with the same techniques the animal uses.

“Evolution is constantly doing natural experiments over millions of years, and it comes up with solutions that you and I would never imagine,” says Trimmer. “If you use animals as your guide, you may come up with completely new solutions.”

Trimmer grew up catching bugs and animals in the fields behind the housing complex where he lived in the English Midlands. “Anything dead I found, I’d dissect,” he says. “I’ve always been interested in how things work.” He arrived at Tufts in 1990 after stints at Harvard Medical School and the University of California at Berkeley, where he first started working with Manduca sexta, a caterpillar species native to the southeastern United States and commonly known as tobacco hornworm.

In the biology department, Trimmer shows me a roomful of his babies, dozens of bright turquoise caterpillars arranged in plastic cups on a rack of shelves. “You won’t mind if I do this,” he says, dropping an index finger-sized insect onto my sweater, which it instantly grasps with its legs. Close up, my new friend has a much softer consistency than I imagined, something very close to Jello. As it wriggles under my fingers, I feel the interplay of muscles just beneath the surface with a new appreciation. In a room next door, Trimmer and his students capture those movements with an array of cameras and computers in a process similar to that used for computer-animated films such as “The Polar Express.”

Muscle power
Their breakthrough came in the discovery that in contrast to humans, who rely on superior brainpower to compute fine motor movements, much of the fine calibration in caterpillars is done through the unique shapes and properties of the insects’ muscles and skin. They can move over obstacles and through tight spaces automatically, without additional input from the brain. The tricky part is duplicating those properties in a robot.

Back at the lab, a dissected silicone caterpillar body reveals a tangle of green wires attached to embedded metal springs that crudely simulate muscles. One of the challenges Trimmer faces is finding materials more like those produced by nature. He has been collaborating with two Tufts colleagues: civil engineering professor Luis Dorfmann, an expert on the complex properties of soft materials, and bioengineer David Kaplan, who specializes in weaving genetically engineered biopolymers.

In the process, Trimmer has had to teach himself whole new fields of science, cutting across boundaries that usually keep disciplines separate. “Biologists can’t solve these things by themselves—they need physicists and engineers,” he says. At the same time, roboticists might learn a thing or two from the way nature has solved the problems that continue to vex them.

Michael Blanding is a free-lance writer. This story ran in the May 2007 issue of the Tufts Journal.