The spherical Jollbot doesn't resemble a grasshopper, but it owes its ability to jump to these tiny creatures. Insects don't have the muscle action to hop like kangaroos, so they store energy like a compressed spring and release it suddenly to leap. Likewise, when the flexible Jollbot is flattened and then released, it bounds upward roughly 20 inches (50 centimeters) into the air.

Jollbot is an example of a biomimetic machine—one that borrows ideas from nature as inspiration for its appearance, behavior and physical mechanisms. Biomimicry, or biomimetic design, is nothing new (think: Leonardo da Vinci's gliders based on bird wings). But engineers and roboticists are now "proactively looking toward nature for solutions to specific engineering issues," says Jollbot's designer, Rhodri Armour, a PhD candidate in mechanical engineering at the University of Bath in England.

Armour, who did his doctoral work at Bath's Center for Biomimetic and Natural Technologies (created in 2003), sought a mechanism that would allow a bot to explore rough environments, which hinder walking and wheeled devices. After four years and three versions of the machine, he unveiled Jollbot in December. In addition to jumping, when it's beach ball–like body is fully extended and taut, the device can roll along bumpy terrain. Picture it on Mars, where a robot like Jollbot could roll and bounce over areas that wheel and tread-based NASA rovers could not.

Over the past decade, biomimetic design has thrived, according to Mark Cutkosky, co-director of Stanford University's Center for Design Research. Biologists have better tools—such as advanced microscopy for viewing things on the scale of thousandths of nanometers—that allow them to learn more about animals and their physical mechanisms, he says. Ronald Arkin, a Georgia Institute of Technology roboticist, notes a dramatic drop in costs of building robots for his research: Making a bot now costs around $1,000 per unit, rather than $30,000.

Meanwhile, new technologies allow engineers to dream beyond designing glorified mechanical arms: So-called "swarm bots" work together like army ants to move relatively heavy objects; a fire hose–cum-snake robot can slither across the floor before putting out a blaze; and Nissan is developing an avoidance system to prevent car crashes based on bees—which use their compound eyes to see nearly all the way around themselves while buzzing about, changing direction when they sense something in their path.

Robert Full, a biologist at the University of California, Berkeley, insists biomimicry isn't all about plundering the natural world. He has worked with engineers since the mid-1990s when he helped to develop the crab-inspired Ariel, a minesweeping robot made by iRobot Corp. (famous for its Roomba robotic vacuum) that can look for buried explosives in surf zones. Rather than just aping what nature provides, he says, he encourages designs that "use the advantageous principles and analogies that you find in nature and integrate them with engineering to make something better than nature." Evolution, he says, is not the ultimate engineer; rather, it works on the "good enough" principle—making incremental improvements on previous designs instead of starting from scratch to build better ones.

Last year, Full co-founded the Center for Integrative Biomechanics in Education & Research (CiBER) at U.C. Berkeley to foster mutualism between biologists and engineers: The former provide mechanisms that the latter can use on their devices, which then may serve as models to advance biology. Stickybot, which Full collaborated on with Stanford's Cutkosky, probably best exemplifies that principle.

In 2003 the Defense Advanced Research Projects Agency (DARPA), the Pentagon's advanced research arm, spent several million dollars to commission the building of a robot that could climb walls for surveillance purposes. The result was Spinybot, which could ascend rough surfaces like trees and cement walls with the aid of microclaws tipped with tiny spines, a mechanism borrowed from insects like the cockroach. Stickybot, which debuted in 2006 and can walk up smooth surfaces like windows, uses an adhesive inspired by geckos. On their feet, the lizards have millions of setae—essentially hairs with split ends—that use intermolecular forces to accomplish "directional adhesion": If their setae encounter a surface moving in one direction—say, going left to right—they adhere; when going the opposite direction, they peel off. "It's like Scotch tape that you don't have to press down to stick it," Cutkosky says.

Engineers noted that their climbing robots fell of walls if they didn't have a tail. "We thought [geckos] never use their tail," Full recalls, but it turns out they do. The rear appendage helps the reptiles stabilize themselves and keep their heads from moving back, causing them to fall head over tail to the ground.

Determining that little piece of biophysics, however, required the building of a robot. "That's where biomimetics is moving," Full says. "It's more than cursory and superficial advice on design."