See How They Run

Two legs are a lot to move, and four legs are considerable. But six legs are a quandary, and forty-four formidable.

By Carl Zimmer|Thursday, September 01, 1994
Here, according to Nilesh Patel, is how you paint the legs of an ant: Begin by finding the right paint. Don't use Wite-Out (it flakes off), and don't use oil-based paints (they're so thick that they make ants walk as if they're wearing casts). Acrylic is best, says Patel. And be sure to get the thinnest, finest brush your local art supply store has to offer.

Next, chill the ant well. Patel, a 23-year-old senior at the University of California at Berkeley, takes his ants to the integrative biology department's temperature-controlled room, which he cools to 40 degrees. Inevitably the cold-blooded ants become sleepy. To completely immobilize them, Patel first tried trapping them under a staple pushed gently into Styrofoam, but they managed to wriggle free. Now he Scotch- tapes three of their six legs to a table. Gently pulling the free legs straight with a pair of tweezers, he dabs them with his brush. "The more practice you have, the better you get at it," says Patel. "When I first started, it took me about seven minutes, but now I can get an ant out in two."

Patel learned to paint ants at the bidding of postdoctoral researcher Rodger Kram, whose work was spurred by a simple question: Do ants run?

The science of animal locomotion is full of such simple questions; getting the answers has always been the difficult part. In 1872, for example, Leland Stanford, a railroad tycoon and the founder of Stanford University, got into a heated debate as to whether all four legs of a horse leave the ground when it trots. Legend has it that he even bet $25,000 that they did. In any case, he did bankroll famed landscape photographer Eadweard Muybridge to find out. Muybridge had horses trot down a path strung with threads connected to a row of cameras; when the horses snapped the threads, the cameras snapped the pictures. It took Muybridge years to perfect a shutter fast enough and film sensitive enough to capture the images (he also needed some time off to defend himself--successfully-- against the charge that he had murdered his wife's lover), but in 1877 he was finally able to give Stanford his answer: a series of pictures of a horse in motion, one of which showed the animal seemingly levitating, with all its legs in the air.

Like Muybridge, Kram wanted to take a series of pictures of an animal in motion--albeit a much, much smaller animal. And, like Muybridge, he had some technical considerations. Because even his high-speed video cameras couldn't easily distinguish a moving ant's six legs, he decided that the three facing the camera had to be painted white. Next, he and Patel built a narrow plastic chute for the painted ants to dash through, coating the walls with liquid Teflon to keep the ants from climbing up its sides. Patel then spent several weeks videotaping ants traveling down the chute. A prod from a pair of forceps or a puff of air was enough to get an ant moving. Patel would reach over to the camera and switch it on in time to tape the insect's movements.

One morning last fall, one ant gave Patel some trouble. "As soon as I touched him he took off," he recalls, "and I wasn't fast enough turning on the camera." By simply turning the camera on before prodding the ant, however, he succeeded in capturing the animal's movements on tape. Later, watching the tape in slow motion, he and Kram saw something remarkable. In previous trials, the ants had used their normal gait, known as the alternating tripod, in which two legs on one side and one leg on the opposite side move in concert--the ant would raise, say, its middle left leg and its front and back right legs, bring them forward, plant them on the ground, then repeat the movement with its other three legs. But the tape of Patel's swift ant showed that at one point all its legs actually left the ground. It was the first image of a running ant ever captured. "Unfortunately," says Kram, "nobody had bet any money on it."

Kram and Patel, along with the dozen other Berkeley researchers who make up the PolyPEDAL lab (PEDAL stands for Performance, Energetics, and Dynamics of Animal Locomotion), are modern Muybridges. But instead of exposing the hidden complexities in the movement of tetrapods--animals like ourselves, with four limbs--the PolyPEDAL lab specializes in the even more complicated mysteries of arthropods, creatures that move on six or more legs and have external skeletons. Among the lab's menagerie, in addition to the trotting ants, you'll find galloping crabs, undulating centipedes, and a cockroach so fast it's made the Guinness Book of World Records.

"We don't study arthropods because we like them," says Robert Full, the 36-year-old physiologist who directs the lab. "Many of them are actually disgusting. But they tell us secrets of nature that we can't find out from studying one species, such as humans."

They are also telling the researchers secrets that may soon be applied to some very unnatural creatures. For decades engineers and computer scientists have been convinced that walking, insectlike robots would be ideal for moving over rough terrain. Thus far, though, the slow, awkward machines they've built embody none of the speed and grace of insects. That, Full says, is because they've based most of their designs on old assumptions about how insects move, assumptions the PolyPEDAL lab has shown are false.

Until the PolyPEDAL lab came along, researchers looking at animal locomotion concentrated on tetrapods even though there are several hundred times more arthropod species on this planet. It's hard to blame the researchers, though. After all, it's much easier to film and analyze the four heavy legs of a running dog than the six nearly weightless limbs of a skittering roach.

After completing his famous studies of horse movement, Muybridge went on to photograph many other tetrapods, including humans, and showed that, as a rule, they all lifted their legs off the ground simultaneously while running. That complete lack of contact with the ground, in fact, came to define the act of running. It's a fairly crude definition, though--it tells you only about a brief moment in an animal's gait and doesn't describe the rest of the motion. These days researchers are constructing better definitions of walking and running by focusing on two simple models: pendulums and pogo sticks.

A pendulum can swing for a long time because it continually recovers its energy. On its downward stroke, it's powered by the force of gravity; when it reaches the lowest point of its arc it has so much energy that it can counteract gravity and swing upward. When you walk, your body behaves like an upside-down pendulum: the foot you plant in front of you is the pendulum's axis, your center of mass the hanging weight. In the beginning of your stride you work against gravity, vaulting your center of mass upward with your leg until you reach your highest point. Gravity then takes over, and your body swings downward until your other leg hits the ground. The next stride is even easier. You can use the energy given to you by gravity to vault yourself into your second and all successive steps, just as a pendulum reclaims its energy in each swing.

When you run, however, you stop behaving so much like a pendulum and begin behaving more like a pogo stick. Now when you first plant your leg, your body sinks down on it instead of rising up. Your leg actually acts as a brake for your body, and so your center of mass is at its lowest point when your acceleration is lowest. Meanwhile your tendons are acting as springs. As they stretch and snap back, they store and release energy, just like the spring in a pogo stick, and propel you upward and forward. Thus running, like walking, recovers energy and cuts down on your expenditure.

One intriguing implication of the pogo-stick model is that you don't have to have both feet in the air to be running. In 1986 Thomas McMahon, a biomechanicist at Harvard, videotaped six runners trying to imitate that miracle of locomotion, Groucho Marx. They deliberately bent their legs deeply as they ran, so that they always kept one foot on the ground, yet their bodies still behaved like pogo sticks. So much for Muybridge.

At the time, Full was working in the laboratory of McMahon's collaborator, Dick Taylor, investigating the intricacies of crab metabolism. One of the things he needed to know was what the crabs were doing with their limbs: Were they using any kind of energy-saving gait? The conventional wisdom at the time--which, as is often the case with conventional wisdom, had no real evidence to back it up--was that if tetrapods were pendulums and pogo sticks, then arthropods were wheels. Like a wheel, it was thought, an arthropod's center of mass moved forward at a constant speed, never rising or falling, and so the animal never recovered energy as it moved.

To study crab locomotion, Full wanted to use one of the lab's force plates--a device that measures the impact of an animal's footsteps. Generally a force plate consists of a sheet of wood or metal sitting atop a grid of crisscrossing beams. Inside each of the beams are gauges that are squeezed by the pressure of the animal's feet as it walks overhead; some gauges measure the up-and-down impacts while others sense forward-and- backward or side-to-side forces. But no one had ever tried to use a force plate to detect the steps of a half-ounce crab before. To do so, Full and a fellow postdoc gave the plate more sensitive gauges and a more complicated electronic configuration, then sent crabs scuttling over its surface.

When Full saw the force patterns the crabs were creating, he had a feeling of déjà vu. Far from moving like a wheel, the crab's center of mass rose and fell, and the crab itself slowed down and sped up rhythmically. At low speeds it walked like a pendulum, with gravity aiding its forward movement; and at faster speeds it ran like a pogo stick, with energy clearly being stored and released in some unidentified springlike structures. The crabs rarely became airborne, which meant that they, too, could run like Groucho. "I said, 'My goodness, there are so many similarities to the general model--it can't happen!' " says Full.

To see if perhaps crabs were the exception rather than the rule among arthropods, Full decided to study cockroaches. But the insects, at a twentieth of an ounce, were too light for even the state-of-the-art Harvard force plates. Full had to wait until he came to Berkeley later that year, where he set up the PolyPEDAL lab. It took almost 12 months, but with the help of an undergraduate named Michael Tu he managed to build and calibrate a force plate that could measure the impact of a cockroach's steps. The construction wasn't easy. "You have 24 strain gauges made out of silicon slivers and gold wires about twice the thickness of a hair, and you have to glue all the gauges on the beams and solder the wires to terminals by hand," Full explains. This was all then covered with a layer of wood a half-millimeter thick. It was worth the effort, however: Full and Tu promptly discovered that cockroaches walk and run like crabs--and horses and humans--by behaving like pendulums and pogo sticks.

But though an arthropod moves like a tetrapod, each of its many legs behaves uniquely. With student Lena Ting, Full measured the force of each individual cockroach leg by covering the lab's force plate with a second thin sheet of wood, suspended just slightly above the first, with two cutout patches the size of a roach's foot. Into these holes went cardboard plugs that sat on the lower plate. The force of the roach's footsteps was thus detected only when the insect hit the plugs and indirectly struck the plate below. Using this device, Full and Ting found that when a running roach's rear legs strike the ground, they push back against it to propel the roach forward; the front legs act like a set of brakes, decelerating the roach; and the middle legs act like human legs-- like pogo sticks--to both accelerate and decelerate the body.

It was during these experiments that Full and his students discovered that the American cockroach can run five feet per second, a speed the Guinness Book of World Records recognized as the fastest of any insect on Earth. This means the roach travels 50 body lengths in a second. A human would have to run 200 miles an hour to match that pace.

It was also during these experiments that the researchers noticed something truly peculiar: the roach was running with its body tilted up 23 degrees. "The film wasn't very good, and we couldn't tell what the hell the animal was doing," Full remembers. "And then we looked at the force platform." Periodically it registered no forces at all--in other words, there were times when the roach had taken all its feet off the ground. Taken together with the images of the roaches tilting upward, this suggested that the bugs were getting up on their two long rear legs and running. "We thought the plate was broken, or maybe there had been a gust of wind, so we made the roaches do it again." Eventually they looked at 40 roaches, using high-speed video with better resolution. In every case, they could see the roaches tilt up their bodies and sprint on their two back legs, just like humans.

Roaches turn bipedal at top speeds, Full suspects, because running on six becomes counterproductive. Their legs, he notes, are moving back and forth 27 times a second, which is probably as fast as their muscles can work. "The only way to go faster, then, is to take bigger steps," says Full, "and since the front legs are a third shorter than the rear ones, they can't possibly have the same stride length."

Still, it was hard to accept that cockroaches could run on two legs, particularly because it seemed as if they were breaking the laws of physics. "When we calculated it, it seemed that they should just fall forward because they're tipped so far over," says Full. Their secret lies in their aerodynamics, a peculiar consequence of the insect's small size and high running speed. Berkeley biomechanics expert Mimi Koehl teamed up with Full to do wind-tunnel experiments and showed that when the roach is sprinting, the air pushes against it so hard that it keeps the roach from falling down.

It's difficult enough to decipher the mysteries of six-legged cockroaches and eight-legged crabs. What about a 44-legged centipede? Researchers have long assumed that a crawling centipede, at least, moves like a wheel--after all, it certainly looks as if it's gliding as smoothly as a chair on casters. They also thought a centipede's movement was rather uneconomical. Many centipedes bend like snakes when they move quickly, apparently wasting energy in undulations that could be used to move the animal forward. At fault, supposedly, was the centipede body plan. Centipede legs extend diagonally down from a long, flexible body. As each leg twists at the joint to propel the animal forward, the force also inadvertently swings the body around the joint.

To test these assumptions, PolyPEDAL lab member Bruce Anderson turned to the common Arizona centipede, a species with a particularly pronounced undulating walk. It's a chilling creature, six inches long, with an orange body, 44 yellow legs, giant pincer jaws, and a painful bite-- indeed, Anderson will handle it only with a gigantic pair of tweezers.

When he videotaped the centipede in motion, Anderson found what PolyPEDALists have almost come to expect: rather than moving smoothly forward like a wheel, the centipede's body accelerates and decelerates as it travels. He also found that it's a much more economical creature than was once thought. Anderson implanted electrodes in the centipede's trunk muscles and found that whenever a segment of its body was curved out to the left, contracting muscles would then begin forcing that segment to curve to the right. When the segment was curved out to the right, the opposite action would occur. In other words, the centipede was not passively bending as a result of its anatomy; it was actively trying to undulate. And the bending seemed to be saving it energy--oxygen-consumption experiments showed that the centipede actually needed less fuel to move than most other animals of its weight.

To see how it was saving energy, though, meant seeing what each of the creature's 44 legs was doing, and force plates just weren't up to the task. For that the lab needed another device, known to the PolyPEDALists as the Jell-O track. The creation of engineering student Angela Yamauchi, it's a clear plastic chute with a long, shallow tray of transparent gelatin sitting between two filters, one below the gelatin, the other just above it. Polarized light shines up from the bottom of the track.

Polarized light consists of photons that have been filtered so they all vibrate in the same plane. If the light encounters a second filter oriented in the same direction, it can pass right through. But if you rotate the second filter 90 degrees, as is the case with the filter above the gelatin, all the light is blocked.

Viewed through the filter above the track, therefore, the clear gelatin looks black. But when an object--say, the foot of an insect--rests on it, the gelatin's structure is transformed. It begins to interfere with the light passing through, changing its polarization. Photons that had been blocked by the second filter are then able to pass through. When an insect ambles down the track, little splotches of white light thus appear around its feet. By looking at the size of a splotch, Yamauchi can calculate the size of the force, and by analyzing its shape, she can tell you in what direction the force is moving.

When Anderson set his centipede on the Jell-O track, he discovered something remarkable. At any moment, most of the centipede's legs are not on the gelatin--in fact, on average, only four touch the ground at a time. These four are the legs at the centermost point of the concave side of the four curves in the centipede's body, two on the left side and two on the right. As the centipede undulates forward and the next foot in line reaches the center of a bend, the centipede lifts up the planted foot and sets the next foot down on the same spot. As different legs reach the center of the curves, they make the entire animal speed up or slow down, with the middle legs supplying most of the propulsive force.

How does this help the centipede save energy? "I made stick figures of these things until I was bleary-eyed," says Anderson, but eventually he came up with a theory. Since the centipede is moving only four legs at a time, rather than 44, it takes less energy overall to operate its legs. This could pose a problem, of course; fewer legs on the ground means less support for the centipede's long, flexible body. But by bending the body in the first place--which requires flexing its body muscles--the centipede keeps its body tight and stops it from sagging.

Of the many arthropods that have made their way into the PolyPEDAL lab, only one actually moves anything like a wheel. In 1979 Berkeley biologist Roy Caldwell encountered a species of stomatopod--a small, shrimplike creature--that had washed up on a beach in Panama. Lying on its back, it pulled its tail to its head to form a hoop and did a backward somersault, letting its back flop onto the ground. Then it pulled its tail up again and rolled on, slowly making its way back into the ocean.

When Full and his students videotaped some stomatopods in the lab in 1992, they found that for 40 percent of each "stride," when the creature forms a hoop, it indeed moves exactly like a wheel. But during the other 60 percent--when it's raising its body up and flopping it down--its center of mass rises and falls, slows down and speeds up. "It's the ultimate example of something that is close to a wheel," says Full, "but even there, it has forces that look like a leg--except that now the leg is the body."

It's not only in the animal world that legs seem to make more sense than wheels. Engineers have long wanted to make robots with legs for traveling over natural terrain, and robots with six legs or more make the most sense. There is already a long shopping list of uses for a walking robot, including exploring the moon and Mars and inspecting toxic waste sites. But today's crop of robots is a long way from any moon walk. One of the most famous walking robots, a six-legged machine named Attila, from the laboratory of MIT engineer Rodney Brooks, is a case in point: it's slow and it uses too much energy. "Attila can go three feet in 30 seconds," explains Mike Binnard, a member of Brooks's lab. "If you watch a video, it looks okay in fast-forward, but at regular speed it's pretty boring."

The problem, according to Full, is that engineers rely on the old conception of insects as having simple, wheellike locomotion. But Full has been preaching his new biomechanical gospel to robot designers for three years now, and they're beginning to take his suggestions to heart. Binnard, for example, is in the process of building an Attila-style robot with cockroach legs. Using Full's extensive data on the insect, he replaced its six identical limbs with three differently shaped pairs of legs: small front legs that act like brakes, longer middle legs that push in both directions, and even longer rear legs that push the robot forward. The savings in energy should help it travel five times faster than Attila.

Binnard and the other robot designers are not the only ones benefiting from this collaboration. The questions they ask have led Full and his PolyPEDAL students into entirely new areas of inquiry. For instance, Binnard recently asked Kram how he might teach his robot to climb over objects. It's something robots are particularly bad at: they slowly plant one foot at a time, continually check their balance, and when they finally pull themselves up, they often strip their gears as they go. Cockroaches, on the other hand, can race over almost anything. Faced with the question, Kram realized he didn't actually know how cockroaches do it. So he and Yamauchi built a step for the Jell-O track, ran roaches over it, and discovered that as the roaches approach the step, they tilt their front ends up with their middle legs and put their front legs on the ledge. But they don't use their front legs to pull themselves up; instead they boost themselves over with a strong push of their back legs.

While Full loves the idea of a scurrying roach robot, his ultimate dream is a mechanical crab. Not only can a crab race across dry land, it can plunge into crashing surf and continue running underwater. A robot with a crab's skills could do a number of remarkable things. Consider Marines trying to land on a beach laced with mines. Crab robots could jump out of the landing craft, run onshore, hunt the mines down, and set them off. Rockwell International recently gave a contract to Full and IS Robotics, a Massachusetts robotics company, to design such a robot for the Navy. A robot crab could also distinguish itself in civilian duty. Researchers surveying the ocean floor or engineers inspecting underwater construction use propeller-driven, tethered machines. In strong currents or crashing waves these devices get tossed around and are often rendered useless. A robot with a crab's stability, on the other hand, could easily work in these inhospitable places.

To build such a robot, however, the researchers need to understand better just how crabs do what they do. Marlene Martinez, another Berkeley grad student, is therefore collecting data on real crabs. She's seen crabs that live on rocky beaches wrap their legs around small rocks to resist the crashing waves, and she's heard of species that live on sandy beaches vibrating their legs to liquefy the sand so they can jam their limbs down into it and hold on tight. She's observed how, when a crab walks underwater, it adjusts the angle at which its body points so the water will lift and support it. This added buoyancy means the crab doesn't need to keep as many legs on the ground for stability. A crab robot would do well to copy any or all of these strategies.

But no matter how deeply involved the PolyPEDALists get in robot design, their main focus is still on exploring how flesh-and-blood animals- -or chitin-and-hemolymph arthropods, to be more precise--move. Despite having discovered the first airborne running ant, for example, Kram is convinced that ants don't have to leave the ground to run. He thinks they, like crabs and humans, may be able to imitate Groucho Marx. To prove his hypothesis, he's going to have to design the most sensitive force plate yet; an ant weighs a thousandth of an ounce. But when he's done, chances are this research, like much of the other work that goes on in the PolyPEDAL lab, will uncover even more hidden similarities between ants and ourselves.

"I think what our research has said is that evolution is constrained," says Full. "There are not an infinite number of ways to move, even though they can look very different. To say that a model of a pogo stick could apply to this kind of diversity--I can tell you, I never would have believed it unless I'd collected the data myself."
Comment on this article

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar

Log in to your account

Email address:
Remember me
Forgot your password?
No problem. Click here to have it emailed to you.

Not registered yet?

Register now for FREE. It takes only a few seconds to complete. Register now »