Wheels have served humankind for thousands of years. But they’re overrated, says Robert Full. How many creatures in nature get around by rolling like a wheel? Full counts one, if that: a shrimplike creature called a stomatopod that occasionally curls itself into a hoop and rolls around the beaches of Panama. For the rest of us landlocked animals—except, it seems, for the NASA eggheads who persist in sending wheeled rovers to Mars—it’s legs or bust.
“Everyone thinks the wheel is the most efficient form of locomotion,” Full exclaims. “Wrong!”
Full is a professor of biomechanics at the University of California at Berkeley. His interest in the field goes back to his childhood, when he used to watch ghost crabs scuttling on the beach in Florida during family vacations. “I want to go to graduate school and watch weird animals move around,” he later told his parents.
This he does. As director of the Poly-PEDAL Laboratory, a cross-disciplinary team of researchers—many of them undergraduates—Full is devoted to forming general principles from the study of animal motion. (Polypedal means “many footed,” and PEDAL is an acronym for Performance, Energetics, and Dynamics of Animal Locomotion.) His team’s activities have included counting the hairs on the toes of geckos and building brainless multilegged robots. But their research hit full stride a few years back, when they began using high-speed digital cameras to film cockroaches, centipedes, and other organisms running on small treadmills. A roach moving at top speed, Full found, will tip up on its hind legs and run like a person. (“I still find them disgusting,” he says.)
The Poly-PEDAL team went on to build an insect runway, outfitted with high-speed video, that can measure the exact force, placement, and timing of an insect’s every tiny footstep. The platform consisted of a photoelastic material sandwiched between two polarizing filters and a light that shone through all three layers. When a roach walked across the stage, its legs deformed the material to a tiny yet quantifiable degree. (Initially the team used orange Jell-O as the medium, but the insects were more interested in eating their runway than walking on it.)
Time-lapse studies showed that a roach’s leg—or any kind of leg, in fact—is nothing more than a spring rod, like a pogo stick. If you graph the force it exerts during any one step, the result looks like a sine wave: It increases to a maximum during the push-off, decreases to zero as the foot hovers in midair, and then goes negative during the landing and repeats. A step is just a bounce. Walking or running is simply a matter of bouncing forward from one pogo stick to another, with some side-to-side motion for stability. Dogs bounce on two legs at a time; roaches (and, amazingly, centipedes) bounce on three legs; crabs bounce on four.
This rule is universal and simple to model mathematically. Full’s robot videos helped inspire the insect movements in the animated film A Bug’s Life. But the implications of his work became most clear when he decided to build a walking robot. Most robots with legs are designed to move as if they had wheels, he says. Their motion is preternaturally smooth, with little up-and-down motion, and they keep at least three legs on the ground at all times for stability. But that’s not how real legs work. Animals are “dynamically stable”—unbalanced from one moment to the next but fully stable over a complete cycle of motion. Stability is a state that legged animals are forever falling into.
And that’s the least of it. Although legs come in many shapes and sizes, they’re all equally springy. Every leg in the animal kingdom has exactly the same relative stiffness, Full has found, and all organisms generate the same amount of mechanical energy—one joule to move 2.2 pounds of body weight 3.3 feet—when walking or running on land. “It’s the same for everything,” Full says. “Dogs. Cockroaches. You. It’s crazy!”
In short, Full’s cockroach work revealed a natural sweet spot: an ideal, easy-to-calculate springiness for any manner of limb. From that realization, only a small conceptual leap was needed to design the first efficient walking robot. Working with scientists at the University of Michigan and McGill University, Full went on to build Rhex (for robot hexapod), a giant mechanical roach that acts like a bread box bouncing on six pogo sticks. Rhex has no touch sensors, no feedback mechanisms, no programmed navigational ability, no grasp whatsoever of the physical world—just six bouncing legs. Yet it works marvelously, traversing rough terrain without a stumble. Most incredibly, when given a sideways poke, it quickly regains its composure. Self-stabilization, Full realized, doesn’t require neural wiring; it’s a free by-product of bouncing. “It’s built into the mechanical form,” he says.
Insects may be stable for the same reason, Full suggests. On his laptop, he has a video of a cockroach racing over a field of uneven blocks. “How do they run over this terrain?” he muses. “They don’t even slow down! Are they sensing everything, feeding it back through their brains? It can’t be. It’s too fast.”
To test his theory, Full sent his live roach back to the treadmill. This time, his team attached a miniature sideways-pointing jet pack on its back. Partway through its run, the jet pack would fire, knocking the roach off balance with a known force. To Full’s astonishment, the roach recovered its balance in less than 10 milliseconds—faster than any neural reflex. The brain wasn’t wasting energy on the body’s stability. The roach was staying upright without thought.
Because the brain doesn’t need to monitor stability, the insect is free to perform higher-order activities, like sensing its whereabouts and plotting a course. “The antennae need only give a general command,” Full says. “They don’t have to speak to the legs.” With that in mind, Full teamed up with Stanford University’s Center for Design Research to build Sprawl, a small six-legged robot with a sensor antenna and a simple course-correcting algorithm. Then came Ariel, a crablike robot built in collaboration with the Massachusetts firm iRobot. Ariel is six-legged, amphibious, and the first robot capable of remaining stable in shallow, rough surf. If it flips over, the legs reorient so that the top of the body becomes the bottom, allowing Ariel to move upside down. Full, eagerly funded by the Office of Naval Research, is now engineering Ariel to perform such tasks as finding underwater mines.
Full insists that his designs don’t mimic nature. They’re just inspired by it. Natural selection does not produce optimal systems, he says. Real limbs and their functions are constrained by the developmental process—they must grow and change over time—and by the fact that they perform multiple tasks. “Natural selection is not engineering,” Full says. “It’s tinkering with what you have.” The key in designing Ariel was realizing that its joints required just two degrees of motion, not the countless possibilities found in living crabs. “Crabs fight with each other, mate, jump on each other,” Full says. “We don’t need to put that in. Robots aren’t reproducing—yet.”
Meanwhile, Full is thinking extraterrestrially. For years he has been trying to get the rover builders at NASA to see the error of their ways: Wheels are slow, unstable, and marginally maneuverable on rocky terrain, he says. Rhex would have a field day on Mars. (When Full showed NASA a video of Rhex scrambling down a rocky hill, he
tinted the hill orange, for a Mars-like effect.) Back in January, when the Mars rover was stuck on its landing
The world’s most efficient climbing robot may one day be modeled on the gecko. Geckos stick to walls with five-toed feet that are covered by half a million microscopic hairs per foot. The hairs are so tiny that they’re attracted to surfaces by van der Waals forces, which govern objects of molecular size.
platform, Full couldn’t help but chuckle: The wheels of progress move slowly. “We’re still trying to convince them that legs are the way to go,” he says.
Discuss this article in the Discover Forum