Researchers develop a robot leg that learns to walk by itself

Researchers at the University of Southern California have come up with a robotic leg that learns to walk by trial and error, much like an animal.

The team, USC Viterbi School of Engineering, thinks to be the first in the world to create a leg powered by animal-like tendons, which can even regain a foothold after being stumbled for not falling.

The operating algorithm of the leg can learn a new walking task after just five minutes of "play" – by performing random movements that build a map of the limb and its interactions with its environment.

"Nowadays, it takes the equivalent of several months or even years of training for a robot to be ready to interact with the world, but we want to achieve rapid learning and adaptations. that we see in nature, "said Professor Francisco J Valero-Cuevas, on the project with PhD students Ali Marjaninejad, Darío Urbina-Meléndez and Brian Cohn.

Walk

The researchers hope that the technology could have several uses, including to help us understand human movement and disability, and to create robots capable of navigating in difficult and rapidly changing environments (for tasks such as recovery after disaster).

"If you leave these [new] The robots learn from the relevant experience, then they will eventually find a solution that, once found, will be used and adapted as needed, "said Marjaninejad, the lead author of the research paper.

"The solution may not be perfect, but will be adopted if it suits the situation.Everyone of us does not need or want – or can not spend time and effort – to win an Olympic medal. "

Via Nature