Robots can apparently adapt like animals, new research finds, a trait that could provide tremendous benefits to society such as in search-and-rescue missions and putting out forest fires.
Using incredible instincts, robots can be programmed to automatically recover from damage in less than two minutes. For example, during the study, published in the journal Nature, a six-legged robot managed to keep walking with two broken legs while a robotic arm learned how to correctly place an object even with several broken motors.
Animals are known for their amazing ability to adapt to injury. That's why there are many three-legged dogs that can catch Frisbees, for instance. Humans also come up with biological strategies for dealing with anything from a sprained ankle to a broken wrist. And until now, the majority of today's robots couldn't say they had such animal-like instincts.
"When injured, animals do not start learning from scratch," senior author Jean-Baptiste Mouret said in a press release. "Instead, they have intuitions about different ways to behave. These intuitions allow them to intelligently select a few, different behaviors to try out and, after these tests, they choose one that works in spite of the injury. We made robots that can do the same."
During the study, before the researchers had the robot go on its merry way, it would use a computer simulation of itself to create a detailed map of the space of high-performing behaviors. This map represents the robot's "intuitions" about different behaviors it can perform and their predicted value. If the robot becomes damaged, it can use these intuitions to guide a learning algorithm (called "Intelligent Trial and Error") that conducts experiments to quickly find a behavior that works despite the damage. Meaning, the robot can adapt much like an animal would after injury. (You can watch a video of this process in action here).
"Once damaged, the robot becomes like a scientist," explained lead author Antoine Cully. "It has prior expectations about different behaviors that might work, and begins testing them. However, these predictions come from the simulated, undamaged robot. It has to find out which of them work, not only in reality, but given the damage."
"Each behavior it tries is like an experiment and, if one behavior doesn't work, the robot is smart enough to rule out that entire type of behavior and try a new type," she continued. "For example, if walking, mostly on its hind legs, does not work well, it will next try walking mostly on its front legs. What's surprising is how quickly it can learn a new way to walk. It's amazing to watch a robot go from crippled and flailing around to efficiently limping away in about two minutes."
This new technique will help scientists develop more robust, effective, and autonomous robots that may open doors to all sorts of practical applications.
For more great nature science stories and general news, please visit our sister site, Headlines and Global News (HNGN).