Chef Bot? Robot Learns Cooking from YouTube Videos

Will cook for master: This robot learned how to cook by watching YouTube videos.

The U.S. military may not be known for its haute cuisine, but it's developing a new robot that can learn how to cook from watching YouTube videos.

Using its brainy programming, therobot is capable of recognizing how kitchen utensils are used in the videos, and can accurately replicate those actions without human intervention, according to the study, which was funded by the Defense Advanced Research Projects Agency (DARPA).

It's not entirely clear why the Army has an interest in robots that can cook, but cooking requires a wide variety of actions that future service robots will need to learn, said researchers at the University of Maryland, College Park, who led the study.

Existing robots are already pretty good at recognizing objects or patterns, but it's much harder to interpret visual information and perform actions based on it, DARPA officials said.

The agency has now "taken the next step" by developing a robot that processes visual information and translates it into actions, Reza Ghanadan, a program manager in DARPA's Defense Sciences Offices,said in a statement.

The team trained the robot by using a database of YouTube cooking videos shot from a third-person perspective. At one level, the robot's "brain" was powered by two learning algorithms, or neural networks: one system for recognizing objects, and one for classifying the type of grip used to handle the object. At a higher level, the robot's software allowed it to use its knowledge to mimic the actions in the videos.

The robot was able to teach itself to follow the cooking videos with a high degree of accuracy, correctly recognizing objects 79 percent of the time, accurately recognizing the way objects were grasped 91 percent of the time, and predicting the correct actions 83 percent of the time, according to DARPA.

In addition, the robot could also remember some of what it learned — for example, which grip type was used with a particular object — and could potentially share that knowledge with other robots, Ghanadan said. "This learning-based approach is a significant step towards developing technologies that could have benefits in areas such as military repair and logistics," she added.

The research was presented Jan. 29 at the 29th meeting of the Association for the Advancement of Artificial Intelligence.

Cheap Holograms Could Give Rise to Glasses-Free 3D TVs

Large and affordable holographic video displays may soon be possible, which could spur the development of glasses-free 3D TVs, researchers say.

A waveguide device for a holographic video monitor being developed by researchers at Brigham Young University in Provo, Utah.

The secret to developing such holographic video displays could be using acoustic waves to control the way a crystal bends light, the scientists added.

Holograms are a special kind of 2D photograph that, when lit up, create the illusion of a 3D image. The pixels making up each hologram scatter light falling onto them in very specific ways, causing these light waves to interact with each other to generate an image with depth.

One potential way to create a moving holographic video — instead of just a static hologram — would involves displays with pixels that can alter the way they bend light. However, these pixels would have to be close to the size of the light waves they are bending, and there is currently no easy or cheap way to create such pixels.

Another technique would use precisely engineered sound waves to squeeze and stretch a transparent material, changing the degree to which it bends light, or its index of refraction. This strategy, called acousto-optic modulation, could create a display that alters the way it scatters light over time, potentially generating holographic videos. However, an early prototype using this method relied on crystals of an expensive transparent material called tellurium dioxide, and could not generate images with a suitable resolution for TVs.

"The prototype used some of the largest crystals of tellurium dioxide every grown, these $25,000 custom pieces of equipment," said study co-author Daniel Smalley, an electrical engineer and media arts and sciences researcher at Brigham Young University in Provo, Utah.

Now, scientists are developing holographic displays using common and much cheaper crystals that could generate high-resolution videos.

"We're using crystals that cost just $2," Smalley told Live Science.

The foundations of this display are crystals of lithium niobate, a material often used in mobile phones and telecommunications networks. Under the surface of each crystal, the researchers etched a microscopic channel, or "waveguide," that serves as a tunnel for light. A metal electrode is then deposited onto each waveguide.

"We can use basic fabrication techniques borrowed from the semiconductor industry to make these devices," Smalley said. "This brings cost down by several orders of magnitude. Hopefully, this means we can make large holographic displays for the first time."

The crystals are piezoelectric, which means that when an electric current is applied to them, they generate acoustic waves that travel like ripples across their surfaces. These acoustic waves can distort the shape of the crystals, which in turn can alter their indexes of refraction and therefore influence any light waves traveling inside them.

To generate a holographic image, lasers of three colors (red, green and blue) are fired into each waveguide. The frequencies of the acoustic waves passing through each crystal determine which colors pass through and which colors are filtered out. The light waves that come out of the waveguides then interact with each other to produce a holographic image.

A holographic video display would stack many of these waveguides alongside each other. Each waveguide is only a few microns wide, and they can be spaced a few dozen microns apart. In comparison, the average human hair is about 100 microns wide.

The researchers suggest they can manufacture a device that can display 50 billion pixels per second. This could provide 3D images at the same kind of resolutions and frame rates seen in standard-definition TV. Now, the challenge will be to develop a computer powerful enough to provide all the data needed to generate these holographic videos, Smalley said.

Smalley and his research assistants Andrew Henrie and Benjamin Haymore detailed their findings online today (Feb. 3) in the journal Review of Scientific Instruments.