Skip to main content

Facebook’s new ’embodied A.I’ project aims to build a new breed of robots


Facebook wants to build an army of robot assistants that can wait on us in all kinds of new ways. Well, kind of.

In fact, Facebook A.I. Research, the artificial intelligence research wing of the social networking titan, is hard at work on developing what it calls “Embodied A.I.” that will go way beyond the abilities of present-day voice interfaces like Siri, Alexa, or Google Assistant by carrying out tasks that allow them to operate in a physical environment. While most people think of A.I. agents as being disembodied chatbots, Facebook A.I. aims to change this by building systems that can perceive and act in the real world.

“We are still far from these capabilities, but you can imagine scenarios like asking a home robot ‘Can you go check if my laptop is on my desk? If so, bring it to me,’ or the robot hearing a thud coming from somewhere upstairs, and going to investigate where it is and what it is,” Kristen Grauman, Professor of Computer Science at the University of Texas at Austin, who also works as a research scientist at Facebook A.I. Research, told Digital Trends.

While Facebook’s end goal may be a way off, it has already made impressive progress. On Friday, Facebook showed off some new work it has been doing such as SoundSpaces, an audio simulation tool that can produce realistic audio rendering based on room geometry, materials, and more. This could be used to help future A.I. assistants understand how sound works in the physical world. Another tool is an indoor mapping system that could allow robots to better navigate unexplored terrain.

The next generation of smart assistants

To be clear, this research isn’t just about building physical robot versions of A.I. assistants. The bots Facebook is working on may also be able to sit on smart glasses (imagine a next-next-next gen version of Microsoft’s Clippy avatar), but with way more contextual smarts and understanding than present generation A.I. assistants.

For example, Facebook engineers want users to be able to ask questions like “where did I leave my keys?” or “what was that dessert we had at the restaurant on Friday night?” and receive accurate responses. That means researching and developing capabilities for embodied A.I. agents such as creating and storing memories, navigating from one place to another, understanding gravity and other reasoning about the world, planning next steps, and decoding dynamic human activities.

Facebook isn’t necessarily the first company you think of when it comes to A.I. assistants. It doesn’t presently have one as ubiquitous as those made by Apple, Amazon, or Google. But it — and its founder — have certainly explored this area before. Back in 2016, CEO Mark Zuckerberg announced that he was building an A.I. capable of running his home. It seems those ambitions have rubbed off on the other people at Facebook, if they weren’t there before that is.

“Facebook A.I. is a leader in many of the subfields that Embodied A.I. encompasses, spanning computer vision, language understanding, robotics, reinforcement learning, curiosity and self-supervision, and more,” Dhruv Batra, professor at the Georgia Tech College of Computing and research scientist at Facebook A.I. Research, told Digital Trends. “It’s a significant feat to make advances in each of these sub-fields individually, and combining them in innovative ways enables us to push the field of A.I. even further.”

Facebook’s collaborators on these various projects include the University of Texas at Austin, University of Illinois, Georgia Tech, and Oregon State.

Editors' Recommendations

Luke Dormehl
I'm a UK-based tech writer covering Cool Tech at Digital Trends. I've also written for Fast Company, Wired, the Guardian…
Asus launches a new AMD-powered iMac competitor with a thin, striking design
asus zen aio 24 imac competitor  1

The new 24-inch iMac features a bold new design that has a lot of people talking. Asus has countered with its own all-in-one computer, and it serves as a noteworthy alternative. The Zen AIO packs a 23.8-inch display with ultraslim side bezels, as well as a large chin bezel to match that of the iMac. The thin display takes cues from the iMac design as well, cutting away any unused space behind the screen.

The Zen AIO has a few features the iMac lacks, though. First off, it can be configured with an optional touchscreen layer, giving you another way to interact with your PC aside from keyboard and mouse input. Apple, for its part, stubbornly reserves touch for mobile devices, like the iPad and iPhone, so you won't find a touchscreen on any Mac devices despite the fact that the new iPad Pro is powered by the same M1 processor that's found inside the redesigned iMac.

Read more
How Nintendo could use A.I. to bring 4K gaming to the Switch Pro
nintendo switch pro dlss nvidia 4k

Rumors have been swirling about a forthcoming new console from Nintendo, possibly called the Switch Pro. It is said to be a bit larger and more powerful, but now, we're hearing about a possible feature that could be a changer for Nintendo's handheld/console hybrid.

While Nintendo's console contemporaries -- like the Xbox Series X, PlayStation 5, and modern gaming PCs -- can game in 4K, that's not the case with the Switch. It currently maxes out at the native 720p of the screen when in handheld mode, or up to 1080p when docked and connected to a TV. The screen resolution isn't expected to change with the Switch Pro, but Nintendo may be using an Nvidia technology to better compete with its competitors.

Read more
The BigSleep A.I. is like Google Image Search for pictures that don’t exist yet

In case you’re wondering, the picture above is "an intricate drawing of eternity." But it’s not the work of a human artist; it’s the creation of BigSleep, the latest amazing example of generative artificial intelligence (A.I.) in action.

A bit like a visual version of text-generating A.I. model GPT-3, BigSleep is capable of taking any text prompt and visualizing an image to fit the words. That could be something esoteric like eternity, or it could be a bowl of cherries, or a beautiful house (the latter of which can be seen below.) Think of it like a Google Images search -- only for pictures that have never previously existed.
How BigSleep works
“At a high level, BigSleep works by combining two neural networks: BigGAN and CLIP,” Ryan Murdock, BigSleep’s 23-year-old creator, a student studying cognitive neuroscience at the University of Utah, told Digital Trends.

Read more