Of course, the invented evolution mammals that hover 200 feet in the air giant skin shreds and crabs 3 feet wide that climb trees, but has he ever invented a four-legged animal with telescoping limbs? No he didn’t. Biology can’t really work that way. But robots certainly can.
Discover the Dynamic Robot for Embodied Testing, aka DyRET, a machine that change the length of her legs on the fly– not to scare humans, but to help robots of all stripes not to fall too much. Writing today in the newspaper Artificial intelligence of nature, Norwegian and Australian researchers describe how they got DyRET to learn how to lengthen or shorten its limbs to tackle different types of terrain. Then, once they let the shape-changing robot get lost in the real world, he used that workout to effectively tread surfaces he had never seen before. (That is, he managed not to collapse into a heap.)
“We can actually take the robot, take it out, and it will just start to adapt,” says computer scientist Tønnes Nygaard of the University of Oslo and the Norwegian Defense Research Establishment, the lead author of the article. “We saw that he was able to use the knowledge acquired before.”
Walking animals do not have stretchy limbs because, above all, this is just not biologically possible. But it is not necessary either. Thanks to millions of years of evolution perfecting our bodies, humans, cheetahs and wolves all move with incredible agility, constantly scanning the ground in front of us for obstacles as we run.
Robots, on the other hand, need help. Even an ultra-sophisticated machine like the Boston Dynamics Spot robot dog struggles to navigate complex terrain. Giving robots telescoping legs both improves their stability when moving over different surfaces and increases their energy efficiency. Stumbling consumes a lot of battery power, and a restless robot could injure itself or humans nearby. “I think it’s a really good idea to have an adjustable body,” says Francisco Valero-Cuevas, an engineer at the University of Southern California. which develops quadrupedal robots but was not involved in this new research. “That’s what’s going on here. An adjustable body makes it a more versatile robot. “
Nygaard and his colleagues formed DyRET by literally building experimental sandboxes first. In the lab, they filled long boxes of concrete, gravel, and sand, representing a range of different terrains the bot could find in the real world. Concrete is the easiest – beautiful, flat, and predictable. Walking in the sand is much more uncertain, because with each step the robot’s legs would sink in a unique way. Gravel is a physically hard surface, like concrete, but it’s also unpredictable, as rocks can move around, making DyRET’s footsteps difficult. “By having the three examples of terrain, with different hardness and roughness, you get a pretty good representation of some sort of general interaction between morphology, or body, and the environment,” says Nygaard.