Which, of course, is the tricky bit. Potentially one of the most useful embodiments of natural design is the bio-inspired robot, which could be deployed in places where people would be too conspicuous, bored to tears, or killed. But such robots are notoriously difficult to build. Ronald Fearing, a professor of electrical engineering at the University of California, Berkeley, has taken on one of the biggest challenges of all: to create a miniature robotic fly that is swift, small, and maneuverable enough for use in surveillance or search-and-rescue operations.
If a blowfly had buzzed into Fearing’s office when we first sat down on a warm March afternoon, the windows flung wide to the garden-like Berkeley campus, I would have swatted it away without a second thought. By the time Fearing finished explaining why he had chosen it as the model for his miniature aircraft, I would have fallen on bended knee in admiration. With wings beating 150 times per second, it hovers, soars, and dives with uncanny agility. From straight-line flight it can turn 90 degrees in under 50 milliseconds —a maneuver that would rip the Stealth fighter to shreds.
The key to making his micromechanical flying insect (MFI) work, Fearing said, isn’t to attempt to copy the fly, but to isolate the structures crucial to its feats of flying, while keeping a sharp eye out for simpler—and perhaps better—ways to perform its highly complex operations. “The fly’s wing is driven by 20 muscles, some of which only fire every fifth wing beat, and all you can do is wonder, What on Earth just happened there?” says Fearing. “Some things are just too mysterious and complicated to be able to replicate.”
After CalTech neurobiologist Michael Dickinson used foot-long plastic wings flapping in two tons of mineral oil to demonstrate how the fly’s U-shaped beat kept it aloft, Fearing whittled the complexity of the wing joint down to something he could manufacture. What he came up with resembles a tiny automobile differential; though lacking the fly’s mystical 20-muscle poetry, it can still bang out U-shaped beats at high speed. To drive the wing, he needed piezoelectric actuators, which at high frequencies can generate more power than fly muscle can. Yet when he asked machinists to manufacture a ten-milligram actuator, he got blank stares. “People told me, ‘Holy cow! I can do a ten-gram actuator,’ which was bigger than our whole fly.”
So Fearing made his own, one of which he held up with tweezers for me to see, a gossamer wand some 11 millimeters long and not much thicker than a cat’s whisker. Fearing has been forced to manufacture many of the other minute components of his fly in the same way, using a micromachining laser and a rapid prototyping system that allows him to design his minuscule parts in a computer, automatically cut and cure them overnight, and assemble them by hand the next day under a microscope.
With the microlaser he cuts the fly’s wings out of a two-micron polyester sheet so delicate that it crumples if you breathe on it and must be reinforced with carbon-fiber spars. The wings on his current model flap at 275 times per second—faster than the insect’s own wings—and make the blowfly’s signature buzz. “Carbon fiber outperforms fly chitin,” he said, with a trace of self-satisfaction. He pointed out a protective plastic box on the lab bench, which contained the fly-bot itself, a delicate, origami-like framework of black carbon-fiber struts and hairlike wires that, not surprisingly, looks nothing like a real fly. A month later it achieved liftoff in a controlled flight on a boom. Fearing expects the fly-bot to hover in two or three years, and eventually to bank and dive with flylike virtuosity.
To find a biomimetic bot already up and running—or at least ambling—one need only cross the bay to Palo Alto. Ever since the fifth century B.C., when Aristotle marveled at how a gecko “can run up and down a tree in any way, even with the head downward,” people have wondered how the lizard manages its gravity-defying locomotion. Two years ago Stanford University roboticist Mark Cutkosky set out to solve this age-old conundrum, with a gecko-inspired climber that he christened Stickybot.
In reality, gecko feet aren’t sticky—they’re dry and smooth to the touch—and owe their remarkable adhesion to some two billion spatula-tipped filaments per square centimeter on their toe pads, each filament only a hundred nanometers thick. These filaments are so small, in fact, that they interact at the molecular level with the surface on which the gecko walks, tapping into the low-level van der Waals forces generated by molecules’ fleeting positive and negative charges, which pull any two adjacent objects together. To make the toe pads for Stickybot, Cutkosky and doctoral student Sangbae Kim, the robot’s lead designer, produced a urethane fabric with tiny bristles that end in 30-micrometer points. Though not as flexible or adherent as the gecko itself, they hold the 500-gram robot on a vertical surface.
But adhesion, Cutkosky found, is only part of the gecko’s game. In order to move swiftly—and geckos can scamper up a vertical surface at one meter per second—its feet must also unstick effortlessly and instantly. To understand how the lizard does this, Cutkosky sought the aid of biologists Bob Full, an expert in animal locomotion, and Kellar Autumn, probably the world’s foremost authority on gecko adhesion. Through painstaking anatomical studies, force tests on individual gecko hairlets, and slow-motion analysis of lizards running on vertical treadmills, Full and Autumn discovered that gecko adhesion is highly directional: Its toes stick only when dragged downward, and they release when the direction of pull is reversed.
With this in mind, Cutkosky endowed his robot with seven-segmented toes that drag and release just like the lizard’s, and a gecko-like stride that snugs it to the wall. He also crafted Stickybot’s legs and feet with a process he calls shape deposition manufacturing (SDM), which combines a range of metals, polymers, and fabrics to create the same smooth gradation from stiff to flexible that is present in the lizard’s limbs and absent in most man-made materials. SDM also allows him to embed actuators, sensors, and other specialized structures that make Stickybot climb better. Then he noticed in a paper on gecko anatomy that the lizard had branching tendons to distribute its weight evenly across the entire surface of its toes. Eureka. “When I saw that, I thought, Wow, that’s great!” He subsequently embedded a branching polyester cloth “tendon” in his robot’s limbs to distribute its load in the same way.
Stickybot now walks up vertical surfaces of glass, plastic, and glazed ceramic tile, though it will be some time before it can keep up with a gecko. For the moment it can walk only on smooth surfaces, at a mere four centimeters per second, a fraction of the speed of its biological role model. The dry adhesive on Stickybot’s toes isn't self-cleaning like the lizard’s either, so it rapidly clogs with dirt. “There are a lot of things about the gecko that we simply had to ignore,” Cutkosky says. Still, a number of real-world applications are in the offing. The Department of Defense’s Defense Advanced Research Projects Agency (DARPA), which funds the project, has it in mind for surveillance: an automaton that could slink up a building and perch there for hours or days, monitoring the terrain below. Cutkosky hypothesizes a range of civilian uses. “I’m trying to get robots to go places where they’ve never gone before,” he told me. “I would like to see Stickybot have a real-world function, whether it’s a toy or another application. Sure, it would be great if it eventually has a lifesaving or humanitarian role...”
His voice trailed off, in a wistful, almost apologetic tone I had heard undercutting the optimism of several other biomimeticists. For all their differences in background, temperament, and ultimate aims, most practitioners conclude their enthusiastic discourses on their bio-inspired invention with a few halfhearted theories on how it may someday make its way into the real world. Often it sounds like wishful thinking.
For all the power of the biomimetics paradigm, and the brilliant people who practice it, bio-inspiration has led to surprisingly few mass-produced products and arguably only one household word—Velcro, which was invented in 1948 by Swiss chemist George de Mestral, by copying the way cockleburs clung to his dog’s coat. In addition to Cutkosky’s lab, five other high-powered research teams are currently trying to mimic gecko adhesion, and so far none has come close to matching the lizard’s strong, directional, self-cleaning grip. Likewise, scientists have yet to meaningfully re-create the abalone nanostructure that accounts for the strength of its shell, and several well-funded biotech companies have gone bankrupt trying to make artificial spider silk. Why?
Some biomimeticists blame industry, whose short-term expectations about how soon a project should be completed and become profitable clash with the time-consuming nature of biomimetics research. Others lament the difficulty in coordinating joint work among diverse academic and industrial disciplines, which is required to understand natural structures and mimic what they do. But the main reason biomimetics hasn’t yet come of age is that from an engineering standpoint, nature is famously, fabulously, wantonly complex. Evolution doesn’t “design” a fly’s wing or a lizard’s foot by working toward a final goal, as an engineer would—it blindly cobbles together myriad random experiments over thousands of generations, resulting in wonderfully inelegant organisms whose goal is to stay alive long enough to produce the next generation and launch the next round of random experiments. To make the abalone’s shell so hard, 15 different proteins perform a carefully choreographed dance that several teams of top scientists have yet to comprehend. The power of spider silk lies not just in the cocktail of proteins that it is composed of, but in the mysteries of the creature’s spinnerets, where 600 spinning nozzles weave seven different kinds of silk into highly resilient configurations.
The multilayered character of much natural engineering makes it particularly difficult to penetrate and pluck apart. The gecko’s feet work so well not just because of their billions of tiny nanohairs, but also because those hairs grow on larger hairs, which in turn grow on toe ridges that are part of bigger toe pads, and so on up to the centimeter scale, creating a seven-part hierarchy that maximizes the lizard’s cling to all climbing surfaces. For the present, people cannot hope to reproduce such intricate nanopuzzles. Nature, however, assembles them effortlessly, molecule by molecule, following the recipe for complexity encoded in DNA. As engineer Mark Cutkosky says, “The price that we pay for complexity at small scales is vastly higher than the price nature pays.”
Nonetheless the gap with nature is gradually closing. Researchers are using electron- and atomic-force microscopes, microtomography, and high-speed computers to peer ever deeper into nature’s microscale and nanoscale secrets, and a growing array of advanced materials to mimic them more accurately than ever before. And even before biomimetics matures into a commercial industry, it has itself developed into a powerful new tool for understanding life. Berkeley animal locomotion expert Bob Full uses what he learns to build running, climbing, and crawling robots—and they in turn have taught him certain fundamental rules of animal movement. He has discovered, for example, that every land animal, from centipedes to kangaroos to humans, has precisely the same springiness in its legs and generates the same relative energy when it runs. Kellar Autumn, the gecko-adhesion specialist and a former student of Full’s, regularly borrows bits of Cutkosky's Stickybot to compare them with the animal’s natural structures and to test central assumptions about gecko biology that cannot be learned from the geckos themselves.
“It’s no problem to apply a 0.2 Newton preload to a patch of gecko adhesive and drag it in a distal direction at one micron per second,” Autumn says. “But try asking a gecko to do the same thing with its foot. It’ll probably just bite you.”