Robochef (Photo : DARPA)
The United States Defense Advanced Research Projects Agency (DARPA), more famous as a designer of futuristic technology such as GPS for the U.S. military, has just built the brains of a robot that can do just one thing -- cook.
Now, does anyone seriously believe DARPA would spend millions in taxpayer dollars to perfect the perfect robot cook? Or is there something much deeper to this project?
As part of its Mathematics of Sensing, Exploitation, and Execution project, DARPA revealed it funded research that led to the development of a mathematical language so innovative it allows a robot to learn just by watching videos.
The new language will give advanced sensors to be embedded in robots the ability to discern which things it hears or sees are important while discarding input it considers trivial.
"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," said program manager Reza Ghanadan of DARPA's Defense Sciences Offices to Tech Times.
This kind of artificial intelligence has been proven possible. A robot built by researchers at the University of Maryland learned how to handle kitchen tools by watching humans do it in YouTube videos.
This robot had two electronic "neural systems". One system for recognizing objects and the other for tracking movements and creating a mathematical model of that movement that allows the robot to copy it.
Ghanadan said the research is a significant step in robotics development.
"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," he said.
A robot that can learn to cook by watching videos can also learn to fire a machine gun, a rifle or any weapon just by watching videos. And it can teach other robots the same thing.
