© 2024 Wyoming Public Media
800-729-5897 | 307-766-4240
Wyoming Public Media is a service of the University of Wyoming
Play Live Radio
Next Up:
0:00
0:00
0:00 0:00
Available On Air Stations
Transmission & Streaming Disruptions

UW Professor Helps Robots Adapt Like Animals

Antoine Cully/UPMC.

 A University of Wyoming professor is part of a research team that has come with a groundbreaking way for damaged robots to adapt and continue to function.

The study was published Thursday in the science journal Nature—and is titled ‘Robots That Can Adapt Like Animals.’

UW Computer Science professor Jeff Clune says that widespread use of artificially intelligent robots will be the norm someday soon—in fields like manufacturing, search and rescue, health care, transportation, and even planetary and deep sea exploration.

Clune’s latest research could be applied to robots doing anything from assembling commercial products to those cleaning up after nuclear disasters. But, so far, there’s been one big problem.

“The second that they become damaged—typically robots become inoperable,” says Clune. “So what we want are robots that can continue to perform their mission even if they become damaged.”

Credit Antoine Cully/UPMC
The undamaged robotic arm used in the experiments.

Clune and his team developed a ‘trial-and-error’ algorithm that allows a damaged robot to draw on its past experiences and performance —which Clune calls a ‘simulated childhood’—to test and come up with strategies to compensate for damage and  finish its job.

“Much as you would do if you were injured out walking in the forest or something—they try something to see if that works—and if it doesn’t work, they try something completely different until they find something that works,” says Clune. “And we’ve found that they can do this in about 1 to 2 minutes and be on about their business."

The team tested their algorithm on a six-legged robot with a damaged leg and a mechanical arm with a broken motor. You can watch a video describing those tests below. 

Clune works on related research projects within UW’s Evolving Artificial Intelligence Laboratory

Related Content