Discover more from PATENT DROP
PATENT DROP: Amazon takes to the sky
Plus: Baidu fixes the uncanny valley; Microsoft puts the AI in advertising
Happy Monday and welcome back to Patent Drop!
Today, we’re taking a look at a filing from Amazon for a new drone designed to cut costs and complexity; Baidu’s tech to make AI-generated faces look less uncanny valley; and Microsoft’s AI that aims to stretch advertising budgets.
But first, Today’s Patent Drop is sponsored by AcreTrader 一 the team that helps you get farmland assets into your portfolio. Their platform enables accredited investors to invest in the real, tangible asset of farmland, which has a lengthy track record of positive returns. A corn farm in Iowa or a citrus ranch in California can help you diversify your portfolio. Add farmland to your portfolio with AcreTrader.*
Let's jump in.
#1. Amazon drones on
Amazon wants to touch the clouds… or at least soar a few hundred feet above the ground.
The company is seeking a patent on a new kind of drone design and manufacturing that may make it possible for it to expand its delivery fleet. The system includes easily replaceable components and fewer parts, aiming to cut down on cost, weight and complexity while keeping the vehicle robust.
For one, the frame of Amazon’s drone is made of “bonded components,” (a.k.a., pieces that aren’t separable), such as “struts, bulkheads, a tail section, wing sections, brackets or joints, and/or motor mounts.” This reduces the number of components included in the frame by having them be fused, and improves the stiffness and strength of the drone overall.
The design also incorporates “assembled components,” which include the “fuselage, motors, propellers, motor pod fairings, stabilizer fins, and/or landing gear.” These components are removable and replaceable, making maintenance easier and quicker while reducing the possibility of needing to chuck a drone entirely if one part breaks down.
Reducing the cost of the drone by creating replaceable pieces and making it less complex overall could allow manufacturers to push out these vehicles more quickly. Creating a stiffer drone, meanwhile, allows for it to remain lightweight while still carrying heavy payloads.
“Design, manufacturing, and assembly of aerial vehicles may be complex, difficult, and expensive,” Amazon noted in its patent. “Accordingly, there is a need for aerial vehicle designs to lower weight, increase stiffness, reduce costs, and facilitate fabrication, assembly, and maintenance.”
Amazon’s interest in delivering your packages via skymail isn’t new. In late 2013, then-CEO Jeff Bezos said in a “60 Minutes” interview that he hoped to launch Prime Air drone delivery for packages under five pounds within four to five years. But the company’s drone ambitions faced a number of hurdles over the years, including safety concerns and FAA regulations.
Amazon started drone deliveries at the tail end of 2022 in the cities of Lockeford, California, and College Station, Texas. Prior to the launch, the company debuted a drone design in November that could fly longer in a wider range of weather conditions, with the goal of “enabling customers to choose drone delivery more often.”
But despite the company's high hopes, Prime Air has yet to live out the tale of success that Bezos spun in 2013. Despite projections in January that the company would make 10,000 Prime Air deliveries by the end of 2023, as of April, the company had only completed 100 deliveries in the markets it launched in. It also cut a significant amount of the team working on drones as a part of its wider layoffs in late January.
The commercial drone delivery market is still nascent. The tech in this patent could be part of its drone comeback plan: Bulking up its fleet with low-cost, easily assembled drones that require less effort to maintain could breathe life into Amazon’s Prime Air ambitions, growing steadily as the industry itself picks up speed.
#2. Baidu flaps its lips
For an artist, lips are often the most difficult thing to draw. AI video programs, it seems, are facing the same problem.
Baidu is working on tech to help mouth movements leap over the uneasy, uncanny valley feeling that AI-generated 3D videos commonly give. Baidu’s system uses two neural networks: One network to first create a 3D video from audio data, and a second network to correct and refine the lip movements created by the first network. The Chinese search giant noted that the neural networks are trained in a host of different AI concepts, such as “computer vision, augmented/virtual reality and deep learning.”
This tech relies on a machine learning concept called “principal component analysis” or the process of converting “high-dimensional data into low-dimensional data” to essentially extract the main features of the data. This system uses a neural network to perform this kind of analysis on audio data to create a 3D video from it.
That analysis then goes through a second neural network, which basically checks the work of the first network and makes corrections. The analysis is translated into a 3D video of lip movement specifically, and applied to a “pre-constructed 3D basic avatar model to obtain a 3D video with a lip movement effect.”
Baidu said in its filing that this tech allows for 3D facial renderings to be done without the use of “blendshape,” or a common way to animate facial expressions that relies on scans of a person’s face to get realistic facial movements.
“Lip movement is an important part of the 3D video generation process,” Baidu noted in its filing. “3D video generation is usually limited by Blendshape, which is mostly man-made, so the generated lip movement is lacking in expressiveness and details.”
Mouth movements are incredibly tricky to render using AI. This is because AI has not yet been able to capture the “microexpressions” in human faces, or the small movements that convey a person’s nonverbal communication and emotion, Jake Maymar, VP of Innovation at The Glimpse Group, told me. It’s the reason that animators and VFX designers so often use human subjects as models, he said: “It’s part of the recipe for getting that right.”
“It's a very, very hard thing to capture the soul of an individual,” Maymar said. “They're called microexpressions because they're almost subliminal. Your brain picks up on them. That light in the eyes, that life behind a person — to capture that and then recreate it digitally is very hard.”
But developers toppling that barrier could open the door for more than just AI-generated VFX, said Maymar. An AI-based system that captures the subtleties of nonverbal communication and translates them to a VR or AR avatar in real time could mean the difference between a VR environment that people want to spend time in and one that people don’t.
“Once we unlock this capability, this will be the tipping point for remote communication,” Maymar said. “The reason I think Baidu is focused on this video technology is it's sort of the first step … before getting into spatial or 3D representations of people.”
This isn’t the first time we’ve seen Baidu take an interest in AI-based video. The company is also seeking to patent AI-based tech that generates videos from a reference photo (though the example photos in the filing were mysteriously devoid of faces). Given the company’s previous patent activity, this latest filing likely won’t be the last seeking to perfect AI-generated facial movements and body gestures, Maymar said.
Baidu has been working hard to keep pace with competitors in the AI race overall. While the company’s initial launch of its AI chatbot Ernie was a flop, founder Robin Li announced in late May that Baidu would soon launch an upgrade to the large language model behind the chatbot. The company also announced a $145 million venture fund to back AI application startups in early June.
But standing out among key players like Google, Microsoft and OpenAI is no easy feat. Making AI-based video that actually looks good may be part of Baidu’s plan to do so.
SPONSORED BY ACRETRADER
Are You Investing In Real Assets?
Did You Know You Can Invest in Wine Grapes?
Wine grapes. Pistachios. Almonds. Avocados. Soybeans. Cotton. Corn.
These are just a few of the many crops grown on farmland investment opportunities offered by AcreTrader, the premier platform for investors to earn potential income on farmland.
Double Duty: Your money works overtime when you invest in farmland, with return potential from both annual rent and land appreciation.
You don’t need to be a fifth-generation farmer to reap the benefits of owning a farm. Explore an investment in farmland today with AcreTrader.*
#3. Microsoft’s AI ad play
Microsoft may be leveraging its AI strength to boost its ad business.
The company is seeking to patent a machine learning model that offers “incrementality estimation” to predict how a piece of sponsored content is going to perform. The company defines incrementality as how the budget given to an ad impacts its performance.
Microsoft’s machine learning model is trained to forecast how a piece of content is going to perform based on the budget applied to it, in order to “judge the effectiveness of the budget towards performance.” The outcome of this model could be based on a number of different performance metrics, including the number of clicks, impressions, and in the case of job postings, the number of applications received, the company said.
The company said in its filing that common machine learning models used to estimate ad performance often make the mistake of conflating higher budgets with better performance, and therefore “suffer from inaccurate recommendations due to inaccuracies in the estimate of how much additional advancement will be accomplished by increasing the budget a certain amount.”
“More particularly, these machine-learned models rely upon historical information in which biases can appear in the data” Microsoft said in the filing.
Here’s where Microsoft’s model differs: The model is trained with an approach called “asymmetric budget split,” where it learns from two separate sets of training data: one for a high-budget set and another for a low-budget set. Microsoft claims this method “ensures unbiasedness in the machine-learned model,” as it is not pulling answers from one large data set, therefore not taking into account the performance of high-budget ads to judge low-budget ones (and vice versa).
Advertising is far from Microsoft’s core business. While the company made $3 billion in ad and search revenue in the most recent quarter, up 3% year over year, its cloud and business productivity offerings make the bulk of its money. Compared to competitors like Meta and Google, which made $28 billion and $61 billion on ads in the latest quarter, respectively, Microsoft's ad business looks like small potatoes.
Ads may not necessarily be the company’s forte, but Microsoft subsidiary LinkedIn is showing signs of faster growth, reporting revenues of $3.6 billion, up 8% compared to last year. Taking this into account, it’s no wonder that the patent filing focuses on how its measurements can improve job posting performance, as well as ads.
If the company is seeking to grow its ad practice, implementing AI into its strategy is a no-brainer. Microsoft is a leader in the AI space, and with this tech, the company is essentially applying its strengths to its weaknesses. (Though it may have some competition in this area, too, as Google, Meta and Amazon are also toying with AI-generated ads.)
Microsoft’s patent also has the potential to negate the need for advertising A/B testing, said Dan Ratner, CEO of Australian ad agency uberbrand. “If AI can help me determine which ads are going to be more effective before I actually publish, and if it could become super reliable, my advertising would be much more effective.”
While this filing gives a peek into how Microsoft thinks about advertising, it also shows how AI has and will continue to impact the world of digital advertising. Arun Kumar, EVP of data and insight at Hero Digital, said AI’s impact will be felt in three main areas: content creation, advertising delivery, and effectiveness measurement.
“The burning question among people now is — will AI eliminate careers for humans or put creative organizations out of business,” Kumar said in an email. “No, but what it will do is provide better efficiencies by reevaluating and reallocating where humans are spending their time. Instead of spending time on baseline activities they have now freed up time for other value-ad projects.”
Check out a few more fun patents before you head off.
Nvidia wants to give you gaming tips. The company is seeking to patent what it calls “playstyle analysis,” which aggregates profiles of games that a user typically plays to recommend similar ones.
EBay wants to give buyers the full picture. The company wants to patent tech that can automatically generate a video of an item listing based on photos taken of it by a seller.
Google wants to help you get really good at trivia. The company is seeking a patent on tech that provides “unique facts” based on an entity in a topic you search. For example, if you look up “When was Einstein born,” it may pull up interesting facts about him alongside answering your question.
What else is new?
Salesforce is doubling it’s investment fund for AI startups from $250 million to $500 million. The company will also starting pushing generative AI features in it’s products.
OpenAI, Google DeepMind and Anthropic will give the U.K. government early access to it’s AI models to support it’s researching into the tech’s safety.
Is your portfolio prepared to weather a recession? Any financial history buff understands that all bets are off when the recession bells toll. An alternative in any economic scenario is an investment in the asset responsible for our much-needed food supply. With AcreTrader, you can invest in parcels of farmland across our nation, which have historically acted as a natural inflation hedge and portfolio diversifier. Start investing in farmland with AcreTrader.*