Advertisement

Robot acquires chef skills via YouTube instructional vids

Robots are being programmed to continuously build on prior information and learning, accelerating the evolution of artificial intelligence.

By
Brooks Hays
Researchers test our their cooking robot. Photo by MSEE/University of Maryland.
Researchers test our their cooking robot. Photo by MSEE/University of Maryland.

COLLEGE PARK, Md., Jan. 30 (UPI) -- The Internet is filled with useless info, but buried in the rubble are nuggets of information -- whole instructional videos, in fact. Building off of previous artificial intelligence work, scientists have engineered a robot capable of acquiring and demonstrating cooking skills learned by watching YouTube videos.

Computer engineers have previously constructed processing systems capable of recognizing objects and patterns -- the basic ability to acquire new information of the system's own volition. But robotics experts at the University of Maryland were able to go a step further, developing software that allows a robot to not only process visual information, but act on it too.

Advertisement

Researchers used a Baxter robot, typically used in manufacturing, and developed a system that allowed it to process the visual stimuli of the YouTube video, recognize and grab corresponding kitchen utensils and/or food objects and manipulate them in a manner true to the culinary task being performed in the video.

"The MSEE program initially focused on sensing, which involves perception and understanding of what's happening in a visual scene, not simply recognizing and identifying objects," Reza Ghanadan, a program manager at DARPA, said in a recent news release. "We've now taken the next step to execution, where a robot processes visual cues through a manipulation action-grammar module and translates them into actions."

That robots are becoming more domestic is not the takeaway from this newest development. What's revelatory is that robots are being programmed to continuously build on prior information and learning.

"Instead of the long and expensive process of programming code to teach robots to do tasks," Ghanadan said, "this research opens the potential for robots to learn much faster, at much lower cost and, to the extent they are authorized to do so, share that knowledge with other robots."

The new research, which scientists hope will be applied to military training and logistics, was presented this week at the annual meeting of the Association for the Advancement of Artificial Intelligence.

Latest Headlines