Nature-inspired robots swim, crawl, and scuttle like animals
Loading...
| Cambridge, Mass.
When it comes to designing a new robot, some scientists are finding a visit to the zoo more helpful than hours spent at the drawing board. Rather than invent new ways for a robot to navigate a forest or crowded city street, they are copying how animals already do it.
If the goal is to make robots capable of surviving any environment, then nature almost always provides a template, says Joseph Ayers, a professor of biology at Northeastern University in Boston. 鈥淎nimals have evolved to operate in every environment,鈥 he says.
The movement is part of an emerging field known as biomimetics, which takes designs from the natural world and applies them to everything from architecture to textiles.
In robotics, engineer-biologist teams are finding that automatons based on the same design principles as a lobster or a gecko can overcome many of the challenges that have kept robots from entering the real world. This new school of thought is changing many assumptions about how robots operate, by emphasizing simple mechanics that can replace complex software and sensors.
鈥淲e鈥檙e in the age that robots are moving out of the factory, out of rather structured environments, into a wider range of applications where the environment is less predictable, and there鈥檚 a higher premium on being both physically and operationally robust,鈥 says Mark Cutkosky, professor of mechanical engineering at Stanford University in Palo Alto, Calif. He has invented cockroach- and gecko-inspired robots, among others. 鈥淥ne of the things that natural organisms do wonderfully well is cope with unexpected variations in the environment.鈥
While engineers have incorporated some natural design techniques over the centuries 鈥 thorn bushes, for example, are believed to have inspired barbed wire 鈥 humans have mostly overlooked natural solutions.
After thousands of years of human innovation, 鈥渢here鈥檚 only a 12 percent overlap between the way humans have solved problems and the way the rest of the natural world has solved problems,鈥 says Janine Benyus, cofounder of the Biomimicry Guild, a consulting firm in Helena, Mont. 鈥淸That] means that almost 90 percent of the time when you鈥檙e looking to the natural world for a solution, it鈥檚 going to be novel.鈥
Most of today鈥檚 robots work like machines, not animals. While advantageous in some respects, it鈥檚 an immense constraint in others.
鈥淚f you鈥檙e controlling a robot with a computer program, unless you鈥檝e anticipated every possible situation it鈥檚 going to get into, it will eventually get into a situation where it has no escape strategy and it will be stuck. Animals never get stuck,鈥 says Dr. Ayers. 鈥淲hat animals do is they wiggle and squirm [until they escape].鈥
Many autonomous robots use complex sensing networks to interact with their environment. They must sense everything that鈥檚 around them before making any movement, a serious computational task that animals do instinctively. Biomimetics, however, is beginning to provide robots with the ability to move without carefully plotting every step.
Instead of using complex, expensive sensor networks, Ayers copied the basic physics of how a lobster moves to create a robo-lobster. It reacts with the surrounding environment on something of an intuitive level, he says, not stopping to process what it鈥檚 doing or exactly what lays ahead. As a result, the robo-lobster never responds to an obstruction in the same manner. If it gets trapped, it keeps moving until it breaks free.
Dr. Cutkosky鈥檚 gecko-inspired robot, Stickybot, also relies on simple mechanics to get around. He and his research team copied the tiny hairs on a gecko鈥檚 feet that allow it to seamlessly scale just about any surface. Now his robot can do the same.
鈥淲hat we鈥檝e been able to learn with the newer bio-inspired approaches to locomotion is that it doesn鈥檛 have to be that hard.鈥 They don鈥檛 have to be terribly expensive, and they don鈥檛 have to have terribly powerful computers and expensive sensors everywhere just to make them run,鈥 says Cutkosky. 鈥淚nsects don鈥檛 have terribly powerful brains and yet they cope pretty well.鈥
For decades, scientists have tried to re-create the human hand for robots, seeing it as the ultimate manipulator. But the vast majority of robots are still unable to pick up even a cup of coffee, says Harvard University engineering professor Robert Howe. He describes these efforts as an 鈥渦nmitigated disaster.鈥
Taking a philosophy that he describes as directly inspired by biomimetics, Howe and graduate student Aaron Dollar, built a hand in 2006 that looks nothing like anything in the animal kingdom, but is capable of picking up anything a human can. Unlike a human hand, which requires 34 muscles to move the fingers and thumb, their robotic hand has only one motor. It relies on springy plastic fingers to naturally grasp an object with the appropriate force. And yes, it can pick up a cup of coffee.
鈥淭he idea there is that you put the smarts into the mechanics,鈥 says Howe, describing one of the central concepts for biomimetic robots. 鈥淚nstead of using sensing and control and computers, you try to build the mechanism so it does the right thing without any of that higher-level supervision.鈥
Most proponents of biomimetics agree that it鈥檚 less about creating carbon copies of nature and more about using it as a starting point.
鈥淓volution works on the just-good-enough principle, not on the perfecting principle,鈥 says Robert Full, a professor of integrated biology at the University of California at Berkeley.
鈥淪o you need biologists who understand not just how organisms work, but how they came to be. Otherwise, you鈥檙e going to be getting advice that may be ineffective, that may be not as good as you just doing it with your own engineering solution,鈥 he says.
When Dr. Full was building an underwater robot based on a crab, he initially suggested that each leg have 18 motors because a crab uses the same number of muscles to control each limb. After some consideration though, he realized that those 18 motors could be replaced by two and it would still move like a crab.
鈥淵ou might say, 鈥榃hy does the crab have all those other joints? What is it doing?鈥 The crab also fights with other crabs, it mates with other crabs,鈥 says Full, 鈥淩obots don鈥檛 do that yet, and maybe we don鈥檛 want them to do that.鈥