发布时间：2022.04.13 14:05  访问次数：  作者：

say what you will, but the augmented reality (ar) glasses segment has been slow to take off. from an indecisive start in 2013, when google glass was first launched, we have come a long way – but have covered a surprisingly short distance.

it is estimated that ar glasses shipment will reach only  by the end of this year, while vr headset shipments are in the millions.

why, then, is facebook (now meta) investing so heavily in neural biofeedback technology, emg? will emg transform the ar glasses industry as we know it and provide meta with a significant head start?

the signs, for now, are bullish, and emg technology appears to be as close to “mind-reading” level intuitiveness as we can get when it comes to capturing human inputs to control the virtual world.

what is emg and why will it revolutionise meta's ar future?-凯发k8一触即发

electromyography or emg is defined as a hardware technology that uses sensors to detect and record electrical activity from the muscles and convert it into input information for augmented reality wearables.

emg has been around as a medical technique for several years, used to detect anomalies in human or animal movements. it was typically used by neurologists and other life sciences professionals such as physiotherapists and biomedical engineers.

in the context of ar, emg powers a form of human-computer interaction (hci) without resorting to traditional input methods like a mouse or keyboard. in fact, the user wearing the emg device doesn’t even need to move their wrists, hand, or any other limb.

they only need to think about the movement and send an electrical signal to the limb conveying the intent. the device picks up the signal and conveys the message to the ar gear.

there are many applications of emg in . for example, ongoing  proposes that people with cerebral palsy may be able to use an emg-enabled interface to recreate neuromuscular control and coordinate movements in the virtual world.

meta’s own research attempts to leverage emg to make user movements in the virtual world more seamless. the idea is that you’ll wear a pair of ar glasses and an emg wristband.

then, you will think about picking up an object in ar and almost move your hand to do so. the emg device will automatically relay those signals to meta’s ar software and the virtual world will respond to your intent.

meta’s emg journey started in 2019 when it acquired ctrl labs for an undisclosed amount (somewhere between $500 million and $1 billion).

at that time, it was among the few companies doing wmg from a consumer perspective, the other being thalmic labs. thalmic labs has rebranded as north since then and was acquired by google.

by acquiring ctrl labs, facebook took over the company’s important ip in well-trained emg models. ctrl labs had already managed to create custom virtual keyboards that would adapt to a user’s unique typing patterns, quirks, and speed based on electrical activity from the muscles.

this work formed the foundation of what meta is now doing with emg under its facebook reality labs division. here is a quick roundup:

• facebook started work on an input device for ar glasses six years ago, in 2015. it decided that a wrist-based wearable would be the most ergonomic and efficient solution.
• before acquiring emg capabilities through ctrl labs, it had explored other techniques like contextualised ai. but these required several more years of effort.
• the first versions of emg to be launched by meta will include a gesture called a “click”. the click is essentially a pinch-and-release action, which will be meta’s ar equivalent of clicking the mouse button.
• future versions of meta’s emg will have more advanced controls. for instance, you’ll be able to touch and move virtual objects, much like dragging and dropping on a mouse.
• emg will work together with meta’s underlying ai. let’s say that you want to exercise in a virtual world. the ai will surface personalised playlists and you can choose a playlist through intent.
• when you provide input via emg, you should be able to get haptic feedback. meta is working on several prototypes to study wristband haptics, including bellowband and tasbi.

there are several reasons why meta has selected emg, which is a relatively new technology for the company, compared to contextualised ai.

most neurotech is extremely complex and difficult to prepare for consumer-grade commercialisation. in that respect, emg is more viable. emg does not require the insertion of a chip or the need to break the skin barrier, unlike direct brain-computer interfaces.

users can simply take them on or off, and the device can gradually “learn” from the user’s habits through prolonged use.

also, emg and wrist-based wearables are a perfect match. you can easily fit in compute resources, antennas, batteries, and multiple sensors into an emg-enabled smart-watch-like device

it helps that ctrl lab’s exceptional work in this space has placed meta head and shoulders above the competition – even as it works on projects like nazare, which is a pair of futuristic ar glasses that will inevitably rely on something like emg for most of its functionalities.

a few concerns remain on the road towards mainstream emg development and deployment for ar. meta and other companies need to adapt these devices for users with more body fat, which could block sensors.

privacy and security is another challenge, as emg-enabled devices can read your most personal and private electrical impulses. to that end, frl research runs a neuroethics program to identify and address these issues early on.

that said, from 410,000 units in 2021, ar glasses sales are expected to reach 3.9 million by 2024, which is a more than 6x increase. thanks to its recent emg advancements, meta is well-positioned to gain from this opportunity.