It’s no secret that YouTube is a veritable cornucopia of how to, life-hack, and informative cooking videos. We’ve all been there I’m sure. But did you know that YouTube videos can now help robots learn to “sense visual information and turn it into action?”
While it’s generally known that robots can learn to recognize objects and patterns, interpreting and acting on that visual input is an entirely different matter. That is, until now.
Recently, a team of DARPA-funded (MSEE) University of Maryland researchers set out to do just that.
The team’s new system, announced by DARPA on January 28, actually enables robots to watch “how to” cooking videos on YouTube and then “recognize, grab and manipulate the correct kitchen utensil or object and perform the demonstrated task.” The robots’ accuracy was high, and all without any additional input from humans. And they can even teach other robots what they’ve learned. Machine learning? Artificial intelligence? Pretty cool, right?
“This system allows robots to continuously build on previous learning—such as types of objects and grasps associated with them—which could have a huge impact on teaching and training,” said Reza Ghanadan, program manager in DARPA’s Defense Sciences Offices, in a recent statement.
“Instead of the long and expensive process of programming code to teach robots to do tasks, 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, Reza Ghanadan added. “This learning-based approach is a significant step towards developing technologies that could have benefits in areas such as military repair and logistics.”
While I’m pretty certain that a robot-cooked meal is not ready for prime time yet, it’s nice to think that some day in the future I can order a ham and cheese sandwich from my very own robot chef. 🙂
For more information, follow this link to DARPA’s statement.
To read a paper about this work, follow this link.
Image Caption: University of Maryland computer scientist Yiannis Aloimonos (center) is developing robotic systems able to visually recognize objects and generate new behavior based on those observations. DARPA is funding this research through its Mathematics of Sensing, Exploitation and Execution (MSEE) program. Credit: University of Maryland Photo