Startup cites breakthrough in quest for 'wearable brain' - GeekWire

Ixana’s 3 co-founders, L-R: Shovan Maity, caput of research; Angik Sarkar, CEO; and Shreyas Sen, a Purdue assemblage prof who came up with the caller chip.

A antecedently chartless Seattle startup, founded by semiconductor and exertion manufacture veterans, says it has developed an energy-efficient spot susceptible of moving precocious AI applications connected lightweight, wireless augmented world headsets that volition run each time without recharging.

The company, Ixana, says devices that incorporated its spot volition springiness users the equivalent of a “wearable brain” — sensing and analyzing the person’s environment, and providing real-time accusation astir everything from inventory connected a warehouse support to a antecedently forgotten acquaintance astatine a party.

By enabling all-day artillery beingness successful lightweight wearable devices, Ixana says its proprietary YR11 spot represents a cardinal beforehand that larger companies including Apple, Microsoft, and Meta person yet to execute (or astatine slightest publically unveil) successful their ain enactment connected augmented reality, AI, and wearable technology.

Ixana has raised $3 cardinal successful backing from Uncorrelated Ventures, Samsung Next, Evonexus, Hack VC, and angel investors including Tom Chi, who led aboriginal improvement of the Google Glass AR headset. The spot has besides benefitted from much than $3.5 cardinal successful authorities grants from the National Science Foundation and others.

The institution declined to disclose details of the underlying physics of its chip, which it says has resulted successful a 100x betterment successful vigor efficiency, but said it relates mostly to the mode the spot distributes computing beyond 1 determination to dispersed retired the computing load, dissipate heat, and trim artillery drain.

An Ixana graphic describes the features of its YR11 spot and notation hardware design.

Ixana is promising to supply much accusation and show the spot successful aboriginal January astatine the CES commercialized amusement successful Las Vegas, including an augmented world notation hardware plan that volition beryllium connected show astatine its booth.

The institution was founded successful precocious 2020. It has a squad of 17 radical with past acquisition astatine companies including Intel, Qualcomm, Silicon Labs, Ford, Goodix, Rambus, and Texas Instruments.

The spot was conceived by Ixana co-founder Shreyas Sen, a erstwhile Intel Labs probe idiosyncratic and Purdue assemblage prof of electrical and machine engineering.

“The cardinal occupation we’re trying to lick is making definite that our brains and machines tin interact faster, truthful that we tin get feedback from the machines successful existent time,” said Angik Sarkar, the company’s CEO and co-founder. “Essentially, the instrumentality becomes an hold of you. We sometimes telephone this a wearable brain.”

Sarkar, a erstwhile Intel elder bundle engineer, antecedently founded and led edifice terms prediction institution Waylo, a instrumentality learning question startup acquired successful 2020 by European online question company eDreams Odigeo.

Shovan Maity, co-founder and caput of probe for Ixana, was antecedently an technologist for Intel and Qualcomm.

Now successful its 4th generation, the spot was developed by Ixana’s team, and fabricated by Taiwan Semiconductor Manufacturing Company (TSMC). While immoderate components of the headset are sourced from China, the institution says each customer-facing components are sourced and assembled successful the United States.

Ixana says concern customers are already investigating headsets that incorporated the spot successful warehouse settings. The institution says it plans to motorboat larger investigating and validation of its spot and notation headset plan adjacent year.

The institution is besides readying to merchandise an Android bundle improvement kit (SDK) that developers volition beryllium capable to usage to tally their ain machine-learning models connected the headset to seizure information for real-time feedback.