The Washington PostDemocracy Dies in Darkness

This robot at U-Md. taught itself to cook by watching YouTube

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. (Photo by John T. Consoli)
Placeholder while article actions load

Finally. A robot that can not only do something completely great — like make you a salad — but doesn’t need to be taught how to do it.

It can just figure it out by watching YouTube.

At the University of Maryland, researchers are designing robots that can learn how to do things by watching people. Instead of being intricately programmed step by step to make a series of movements to accomplish a goal, these robots can figure out how to do it themselves.

Here, their robot (aptly named Julia) shows off cooking skills:

University of Maryland researchers are investigating autonomy in robotics that includes action recognition. (Video: University of Maryland Institute for Advanced Computer Studies)

The researchers —  Yiannis Aloimonos, a computer science professor , Cornelia Fermüller, an associate research scientist, Yezhou Yang, a doctoral student in computer science, and Yi Li, a U-Md. PhD who is now a senior researcher in computer vision at the National Information Communications Technology Research Centre of Excellence in Australia  — chose cooking as a goal since it’s something everyone can understand. (Even if they can’t do it.)

It’s also something that takes a series of fairly precise steps to do well. And who wants to take the time to teach their robot how to peel a carrot?

Aloimonos predicted a future in which robots not only automate complex manufacturing, as they do now, but can take over the dull tasks of everyday life so that we humans are freed up to do the more interesting stuff.

We just need to be careful what they see on YouTube…