Robots Blur the Lines Between Biology and Engineering
CHICAGO-Scientists have been dreaming up robots modeled on animals for several decades, and now the field has reached a point where robot fish might soon patrol underwater oil rigs and real-life ants could inspire self-driving cars.
Robotics researchers are turning to biologists to figure out how animals have solved (with the help of millions of years of evolution) the challenges of sensing and moving around their environments. And, biologists are building robots to test their theories of animal neuroscience and behavior.
The lines have definitely blurred, said Barbara Webb, a professor of biorobotics at the University of Edinburgh, at the AAAS Annual Meeting. "Certainly in my lab, and probably everybody's lab, we do studies on both animals and robots. I have students that don't know if they're biologists or engineers." (Watch a related news briefing.)
Robots built at Edinburgh University help test theories about the ant brain's role in navigation. | Insect Robotics Lab University of Edinburgh
Webb and her colleagues study how insect brains help the animals see, make sounds and navigate. They build robots to test their hypotheses. "If you think you understand how something works," she explained, "you should be able to build that thing and have it work."
She studies how desert ants navigate along the ground using what they remember seeing in their path. Saddled with poor vision and "a brain that would fit on a pinhead," these ants are guided only by a low-resolution map of their surroundings, Webb said. She and her colleagues followed ants as they wove through weeds and rocks toward seeds. They then tried to replicate the ants' low-tech journeys by building a robot that consisted of a wheel- mounted smart phone with a wide-angle camera lens.
If Webb and her colleagues can learn more about how ants navigate using such minimal input, their findings might aid those who want to build simpler robotic systems for applications like self-driving cars, which for the moment would seem to need massive amounts of data and processing power to operate.
Malcolm MacIver, a professor of biomedical engineering and mechanical engineering at Northwestern University, has been working on the same fundamental problem in biology: "How does our ability to move intersect with our need to acquire information?"
Malcolm MacIver demonstrates the swimming capabilities of the robot inspired by the black ghost knifefish. | Malcolm MacIver
Looking to swimmers instead of crawlers, MacIver has been inspired by the weakly electric black ghost knifefish of the Amazon basin. The name is apt, since a stack of about a hundred of the fish only would provide as much power as a single AA battery. But the fish use the electric current they produce to sense their surroundings in dark waters. They also propel themselves through the murk by way of a ribbon-like fin.
MacIver and colleagues have adopted both of these biological solutions for fish robots that they think have some advantages over conventional remotely controlled underwater vehicles, or ROVs. During the 2010 Deepwater Horizon oil spill in the Gulf of Mexico, ROVs deployed to the wellhead were blinded by silt stirred up by their propellers and proved tricky to maneuver.
MacIver said he hopes that commercial ROV builders might be interested in how nimbly the robot fish swims with its ribbon fin and senses with its electrical field. "The bottom line is that it doesn't get tied up in old fishing nets that have been abandoned or weeds and such things that constantly plague ROVs," and can be precisely positioned underwater, he said.
And then there's the cheetah robot developed by Sangbae Kim, the director of the MIT Biomimetic Robotics Lab, and his colleagues. The cheetah robot can now trot at speeds of up to 13.5 miles per hour and for the first time can change its gait from a trot to a gallop. The components used to build fast-moving robots like the cheetah are steadily improving, Kim said.
The cheetah robot built at MIT can reach speeds of up to 13.5 miles per hour. | Donjin Hyun, Ssangok Seok, Jongwoo Lee, and Sangbae Kim
But he agreed with Webb and MacIver that the field of bioinspired robotics also has matured to the point where researchers have a better understanding of which features they want to "extract" from their natural examples.
Engineers remain inspired, MacIver said, but they no longer seek to "slavishly copy" their model animals down to the last detail of fin or paw.