PATENT DROP: Apple has its eyes on the prize
Discover more from PATENT DROP
PATENT DROP: Apple has its eyes on the prize
Plus: Intel’s edge computing ambitions; Airbus’s high-flying AI
Happy Thursday and welcome to Patent Drop!
Today, we’re taking a peek at Apple’s eye-tracking neural networks that may be a part of its rumored VR headset. We’ll also check out Intel’s plan to make edge computing an everyday reality, and tech from Airbus to make AI fly high.
Let’s jump right in.
#1. Apple’s mixed-reality vision
Apple wants to see what you’re seeing.
The company is seeking to patent tech for “event camera-based gaze tracking” which utilizes neural networks in a “controller,” a.k.a. an AR or VR headset. Here’s how it works: An event camera – or a vision sensor that picks up on “changes in light intensity” – collects data on eye movements, which are then fed to a neural network.
The neural network then generates a “gaze characteristic,” which includes the user's “initial pupil center” and any gaze adjustment from tracking “location-focused input,” or the minuscule movements of just your pupil, not the entire eye.
Apple notes that this kind of eye tracking works more quickly and efficiently in guiding user interactions, offering “foveated rendering (or better resolution at the point of focus), and cutting down on geometric distortion in rendering 3D objects. It also eats up far less processing power than traditional methods, preventing excessive power consumption and overheating in the user’s headset.
“Existing gaze tracking systems often include a camera that transmits images of the eyes of the user to a processor that performs the gaze tracking,” Apple said in its filing. “Transmission of the images at a sufficient frame rate to enable gaze tracking requires a communication link with substantial bandwidth.”
An AR/VR headset has long been rumored to be in the works at Apple. This patent filing may reveal another detail about how this headset will work.
It also isn’t the first time we’ve seen patents from Apple indicating an interest in eye tracking. The company previously sought to patent tech that tracked eye movement as a means of controlling a “virtual agent” to perform tasks in an augmented or virtual reality environment, which we discussed in Patent Drop a few months ago. Taken together, it seems that eye tracking and control could be a major feature of its buzzy upcoming device.
Adding advanced features like this could be Apple’s plan to get consumers to shell out despite the steep price tag: The headset will reportedly cost $3,000, Bloomberg’s Mark Gurman reported in late March. The headset also comes with a number of disadvantages, Gurman reported, such as an external battery that needs to be charged every few hours, no companion app and limited media content at launch. It could ship out as soon as the end of this year and is expected to debut at Apple’s Worldwide Developers Conference starting June 5.
The headset would be Apple’s first step in competing in the artificial reality ring. Microsoft currently offers three versions of the HoloLens headset, ranging from $3,500 to nearly $5,000. Meta, meanwhile, offers several versions of the mixed-reality Meta Quest headset, starting at $1,000.
With Apple’s core business being consumer electronics, the company’s best shot at competing in the market could be offering state-of-the-art features that appeal to die-hard Apple fans.
#2. The edge of glory computing
Intel wants to be on the bleeding edge of edge computing.
The company filed a patent application detailing several ways that it aims to deploy edge computing. For reference, edge computing is a type of computing in which data is processed as close to its point of origin as possible, ergo, on the edge of the network. Essentially, rather than data being transmitted to a data center for whatever work needs to be done, the processing, analysis, or other operation is completed wherever the data is created.
Intel’s patent essentially details how to string these edge computing devices together, Kevin Gordon, VP of AI technologies at NexOptic, told me. For example, say you have a network of thousands of sensors, cameras, or other devices that are all doing data processing on the devices themselves. Rather than transferring relevant data to the rest of the device network through a cloud or physical server, Intel aims to connect these devices so that they can share data amongst themselves.
This tech works with emerging 5G networks by essentially creating “subnetworks” of physical devices, Gordon noted.
“This patent is saying (Intel’s) going to build edge devices and coordinate these edge device networks … and figure out how to make all these edge devices communicate to each other in a meaningful way,” Gordon said.
While this sounds esoteric, Intel laid out a few examples of how it may show up on the user end, including industrial controls and autonomous driving, with “ultra-low latency.”
Edge computing is both young and old. A widespread example of this paradigm is the common smartphone. These devices collect user data, process it and offer functions using that data on the device itself. What’s nascent is the application of this tech to a much broader array of functions, said Gordon.
Devices for security, manufacturing, agriculture and more are growing in sophistication, so that more data processing, analytics and operations are done on-device, limiting reliance on central servers. “You won’t even need the central computers anymore, and you can do away with infrastructure there,” Gordon said.
It’s no secret that Intel has an interest in edge computing. The company offers edge implementations for manufacturing, retail and communications, with a client roster including Audi, CenturyLink and Etisalat. Intel’s interest in creating devices with higher-level processing makes sense: Its main money maker is hardware and chips, and it’s “getting beat up in the data center space” by cloud heavyweights like AWS and Microsoft Azure, Gordon said.
Edge computing, meanwhile, is predicted to see a higher growth rate than data centers in the next decade, he added. “I see them positioning themselves to take advantage of the growth in edge here.”
Plus, Gordon said, with 5G networks still in the early days, Intel may be “putting its stake in the ground on how you can use 5G with these edge computing applications.”
#3. Airbus’s AI-rplane
“If I’m back here, and you’re back here, then who’s flying the plane?”
Airbus is seeking to patent tech that uses AI to determine a flight plan. First, this tech, which Airbus dubbed a “routing engine,” receives a request to create a flight plan, which includes information regarding the aircraft’s destination and criteria on how the requester wants to optimize the flight, such as using the least amount of fuel or getting to the destination as quickly as possible.
Then, using machine learning, this tech calculates a flight route that includes the longitude and latitude of an aircraft, the altitude and time, and potential constraints on performance, such as structure stress or fuel consumption, based on what the requester wants to prioritize.
For example, if a pilot or flight controller wants to get a flight from Los Angeles to New York at the lowest possible cost, this routing engine could calculate how to make this flight using the least amount of fuel, taking into account things like environmental factors, number of passengers, and how it may impact the aircraft.
“Typically, in mission planning systems for aircraft it is required the intervention of a human operator in order to define a flight plan through a dedicated human machine interface,” Airbus said in its filing. “Steps are performed by a human operator, rendering optimization tasks long and iterative by nature, unable to provide high quality solution(s) in a reasonable timeframe.”
As Airbus noted in this, flight plans are usually human-generated. Adding AI into the mix adds another example to the ever-growing list of the ways this tech is weaving its way into processes that are generally done by people, for people.
Creating a flight plan involves thinking critically about the pilot's priorities for a flight and the potential obstacles that an aircraft could encounter. Leaving this job to AI could limit human errors, making the job of a pilot or air traffic controller much easier and minimizing any potential risks of that flight.
A sizable chunk of Airbus’ business comes from manufacturing commercial aircraft, seeing revenues of nearly $9 billion in its commercial segment in the first quarter. The company also has a smaller defense segment, which made roughly $2.5 billion in that time period.
But with revenues down in both segments year over year – around 5% and 6% respectively – Integrating a proprietary AI could be a way to attract airlines to its aircraft over competitors. That AI could also lend a critical hand amid the shortage of pilots, both in the commercial airline industry and the U.S. Airforce.
The other thing to consider, however, is what’s at risk if this AI makes a mistake. It’s the same argument for slowing the pace of implementing AI other into human-centered industries, like healthcare: Given that passengers’ safety is in the hands of Airbus’ machine learning routing engine, the tech needs to be ironclad before making its way into customers’ aircraft.
Extra Drops
Some other fun patents we wanted to share.
Adobe wants to catch posers. The company is looking to patent tech that uses AI to detect counterfeit objects in images by checking the authenticity of logos, trademarks and designs.
Microsoft doesn’t want to spoil the ending. The company is seeking to patent tech for “spoiler prevention” in digital media, which allows you to put certain apps or browsers into “spoiler-free” mode to keep the juicy details of your favorite show a secret.
Nike is digging deeper into Web 3.0. The company is looking to patent a method for creating an avatar corresponding to “athletic activity performed by a corresponding user.” (So if you want your avatar to be ripped, you better get to work.)
What else is new?
Microsoft is axing salary bumps for full-time employees this year, with CEO Satya Nadella blaming “global macroeconomic uncertainties” in an email to staff.
Softbank posted $32 billion in losses from its Vision Fund segment in 2022. The record losses come roughly a year after the company went into “defense” mode due to turmoil in the tech sector.
Mikko Salovaara, CEO of Revolut Finance, is leaving the fintech company after two years in the position, saying his departure is for “personal reasons.”
Have any comments, tips or suggestions? Drop us a line! Email us at admin@patentdrop.xyz or shoot us a DM on Twitter @patentdrop.
 | A guest post by
|