Imagine sending a text or playing a game without lifting a finger. Or at least, without moving it very much.
This wristband, developed by Meta’s researchers, might just make that real.
The team recently published a study on a new wrist-worn device that turns the slightest muscle twitches into commands that a computer can understand.
The “neuromotor wristband” is designed to capture the intention behind the movement, even if the user’s hand isn’t touching a screen or surface.
It can detect and decode electrical signals produced by forearm muscles using artificial intelligence (AI).
In a video demonstration, a user writes the words “hello world” in mid-air, and a visualisation of the corresponding text appears alongside.
Meta also showed how people can move a cursor around a screen and play games using only finger taps or tiny hand motions.
"It really is a somewhat breathtaking set of discoveries we have here,” Thomas Reardon, vice president of research at Meta Reality Labs, said in a company-provided video explanation.
How does it work?
Most brain machine interfaces available today require surgery.
Meanwhile, the electrodes in the wristband pick up signals sent from the brain to the muscles, instead of putting the sensors directly in the brain by drilling into the skull.
“We decided to take a different approach and record from the natural output of the brain. We don't need to go into your body to listen to [it]. We can do that from the side of the body," Reardon said.
The wrist and forearm are packed with muscles that control hand and finger movements.
These neuromuscular signals are processed in real time and send commands to computers via Bluetooth.
The Meta team collected training data from thousands of participants to build an AI that works for users with different movement styles.
The team says the wristband will be able to help a wide spectrum of users, from individuals with disabilities to everyday users.
"We took the approach of saying, how would we build something such that it just worked right out of the box with eight billion people?” Reardon said.
“This research that we're publishing demonstrates that there are some inherent, what we call scaling laws, that allow us to build a general model for really all of civilisation, such that people can put on a band and start using their brain directly,” Reardon added.
“They elect those signals from their brain in a general way to control a computer”.
Patrick Kaifosh, director of research science at Meta Reality Labs, said he expects “the technology to go a great deal farther”.
“What you've seen is these scaling curves continue as you get more people, more participants, data, it gets better and better," he added.
Meta says it hopes that the study offers a blueprint to the broader scientific community “to create neuromotor interfaces of their own,” according to a statement.
It has recently released over 100 hours of muscle signal data from more than 300 participants.
For more on this story, watch the video in the media player above.
