Connect with us


Mobileye CEO clowns on Nvidia for allegedly copying self-driving car safety scheme – TechCrunch



While creating self-driving car systems, it’s natural that different companies might independently arrive at similar methods or results — but the similarities in a recent “first of its kind” Nvidia proposal to work done by Mobileye two years ago were just too much for the latter company’s CEO to take politely.

Amnon Shashua, in a blog post on parent company Intel’s news feed cheekily titled “Innovation Requires Originality, openly mocks Nvidia’s “Safety Force Field,” pointing out innumerable similarities to Mobileye’s “Responsibility Sensitive Safety” paper from 2017.

He writes:

It is clear Nvidia’s leaders have continued their pattern of imitation as their so-called “first-of-its-kind” safety concept is a close replica of the RSS model we published nearly two years ago. In our opinion, SFF is simply an inferior version of RSS dressed in green and black. To the extent there is any innovation there, it appears to be primarily of the linguistic variety.

Now, it’s worth considering the idea that the approach both seem to take is, like many in the automotive and autonomous fields and others, simply inevitable. Car makers don’t go around accusing each other of using the similar setup of four wheels and two pedals. It’s partly for this reason, and partly because the safety model works better the more cars follow it, that when Mobileye published its RSS paper, it did so publicly and invited the industry to collaborate.

Many did, and as Shashua points out, including Nvidia, at least for a short time in 2018, after which Nvidia pulled out of collaboration talks. To do so and then, a year afterwards, propose a system that is, if not identical, then at least remarkably similar, and without crediting or mentioning Mobileye is suspicious to say the least.

The (highly simplified) foundation of both is calculating a set of standard actions corresponding to laws and human behavior that plan safe maneuvers based on the car’s own physical parameters and those of nearby objects and actors. But the similarities extend beyond these basics, Shashua writes (emphasis his):

RSS defines a safe longitudinal and a safe lateral distance around the vehicle. When those safe distances are compromised, we say that the vehicle is in a Dangerous Situation and must perform a Proper Response. The specific moment when the vehicle must perform the Proper Response is called the Danger Threshold.

SFF defines identical concepts with slightly modified terminology. Safe longitudinal distance is instead called “the SFF in One Dimension;” safe lateral distance is described as “the SFF in Higher Dimensions.”  Instead of Proper Response, SFF uses “Safety Procedure.” Instead of Dangerous Situation, SFF replaces it with “Unsafe Situation.” And, just to be complete, SFF also recognizes the existence of a Danger Threshold, instead calling it a “Critical Moment.”

This is followed by numerous other close parallels, and just when you think it’s done, he includes a whole separate document (PDF) showing dozens of other cases where Nvidia seems (it’s hard to tell in some cases if you’re not closely familiar with the subject matter) to have followed Mobileye and RSS’s example over and over again.

Theoretical work like this isn’t really patentable, and patenting wouldn’t be wise anyway, since widespread adoption of the basic ideas is the most desirable outcome (as both papers emphasize). But it’s common for one R&D group to push in one direction and have others refine or create counter-approaches.

You see it in computer vision, where for example Google boffins may publish their early and interesting work, which is picked up by FAIR or Uber and improved or added to in another paper 8 months later. So it really would have been fine for Nvidia to publicly say “Mobileye proposed some stuff, that’s great but here’s our superior approach.”

Instead there is no mention of RSS at all, which is strange considering their similarity, and the only citation in the SFF whitepaper is “The Safety Force Field, Nvidia, 2017,” in which, we are informed on the very first line, “the precise math is detailed.”

Just one problem: This paper doesn’t seem to exist anywhere. It certainly was never published publicly in any journal or blog post by the company. It has no DOI number and doesn’t show up in any searches or article archives. This appears to be the first time anyone has ever cited it.

It’s not required for rival companies to be civil with each other all the time, but in the research world this will almost certainly be considered poor form by Nvidia, and that can have knock-on effects when it comes to recruiting and overall credibility.

I’ve contacted Nvidia for comment (and to ask for a copy of this mysterious paper). I’ll update this post if I hear back.

Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *


This 22-year-old builds chips in his parents’ garage



Enlarge / Sam Zeloof completed this homemade computer chip with 1,200 transistors, seen under a magnifying glass, in August 2021.

Sam Kang

In August, chipmaker Intel revealed new details about its plan to build a “mega-fab” on US soil, a $100 billion factory where 10,000 workers will make a new generation of powerful processors studded with billions of transistors. The same month, 22-year-old Sam Zeloof announced his own semiconductor milestone. It was achieved alone in his family’s New Jersey garage, about 30 miles from where the first transistor was made at Bell Labs in 1947.

With a collection of salvaged and homemade equipment, Zeloof produced a chip with 1,200 transistors. He had sliced up wafers of silicon, patterned them with microscopic designs using ultraviolet light, and dunked them in acid by hand, documenting the process on YouTube and his blog. “Maybe it’s overconfidence, but I have a mentality that another human figured it out, so I can too, even if maybe it takes me longer,” he says.

Zeloof’s chip was his second. He made the first, much smaller one as a high school senior in 2018; he started making individual transistors a year before that. His chips lag Intel’s by technological eons, but Zeloof argues only half-jokingly that he’s making faster progress than the semiconductor industry did in its early days. His second chip has 200 times as many transistors as his first, a growth rate outpacing Moore’s law, the rule of thumb coined by an Intel cofounder that says the number of transistors on a chip doubles roughly every two years.

Zeloof now hopes to match the scale of Intel’s breakthrough 4004 chip from 1971, the first commercial microprocessor, which had 2,300 transistors and was used in calculators and other business machines. In December, he started work on an interim circuit design that can perform simple addition.

Zeloof says making it easier to tinker with semiconductors would foster new ideas in tech.
Enlarge / Zeloof says making it easier to tinker with semiconductors would foster new ideas in tech.

Sam Kang

Outside Zeloof’s garage, the pandemic has triggered a global semiconductor shortage, hobbling supplies of products from cars to game consoles. That’s inspired new interest from policymakers in rebuilding the US capacity to produce its own computer chips, after decades of offshoring.

Garage-built chips aren’t about to power your PlayStation, but Zeloof says his unusual hobby has convinced him that society would benefit from chipmaking being more accessible to inventors without multimillion-dollar budgets. “That really high barrier to entry will make you super risk-averse, and that’s bad for innovation,” Zeloof says.

Zeloof started down the path to making his own chips as a high school junior, in 2016. He was impressed by YouTube videos from inventor and entrepreneur Jeri Ellsworth in which she made her own, thumb-sized transistors, in a process that included templates cut from vinyl decals and a bottle of rust stain remover. Zeloof set out to replicate Ellsworth’s project and take what to him seemed a logical next step: going from lone transistors to integrated circuits, a jump that historically took about a decade. “He took it a quantum leap further,” says Ellsworth, now CEO of an augmented-reality startup called Tilt Five. “There’s tremendous value in reminding the world that these industries that seem so far out of reach started somewhere more modest, and you can do that yourself.”

Computer chip fabrication is sometimes described as the world’s most difficult and precise manufacturing process. When Zeloof started blogging about his goals for the project, some industry experts emailed to tell him it was impossible. “The reason for doing it was honestly because I thought it would be funny,” he says. “I wanted to make a statement that we should be more careful when we hear that something’s impossible.”

Zeloof’s family was supportive but also cautious. His father asked a semiconductor engineer he knew to offer some safety advice. “My first reaction was that you couldn’t do it. This is a garage,” says Mark Rothman, who has spent 40 years in chip engineering and now works at a company making technology for OLED screens. Rothman’s initial reaction softened as he saw Zeloof’s progress. “He has done things I would never have thought people could do.”

Zeloof’s project involves history as well as engineering. Modern chip fabrication takes place in facilities whose expensive HVAC systems remove every trace of dust that might trouble their billions of dollars of machinery. Zeloof couldn’t match those techniques, so he read patents and textbooks from the 1960s and ’70s, when engineers at pioneering companies like Fairchild Semiconductor made chips at ordinary workbenches. “They describe methods using X-Acto blades and tape and a few beakers, not ‘We have this $10 million machine the size of a room,’” Zeloof says.

Zeloof had to stock his lab with vintage equipment too. On eBay and other auction sites he found a ready supply of bargain chip gear from the 1970s and ’80s that once belonged to since-shuttered Californian tech companies. Much of the equipment required fixing, but old machines are easier to tinker with than modern lab machinery. One of Zeloof’s best finds was a broken electron microscope that cost $250,000 in the early ’90s; he bought it for $1,000 and repaired it. He uses it to inspect his chips for flaws, as well as the nanostructures on butterfly wings.

Continue Reading


Google Labs starts up a blockchain division



Here’s a fun new report from Bloomberg: Google is forming a blockchain division. The news comes hot on the heels of a Bloomberg report from yesterday that quoted Google’s president of commerce as saying, “Crypto is something we pay a lot of attention to.” Web3 is apparently becoming a thing at Google.

Shivakumar Venkataraman, a longtime Googler from the advertising division, is running the blockchain group, which lives under the nascent “Google Labs” division that was started about three months ago. Labs is home to “high-potential, long-term projects,” basically making it the new Google X division (X was turned into a less-Google-focused Alphabet division in 2016). Bavor used to be vice president of virtual reality, and Labs contains all of those VR and augmented reality projects, like the “Project Starline” 3D video booth and Google’s AR goggles.

Just like “algorithms,” “AI,” and “5G,” “blockchain” is often used as the go-to buzzword for rudderless tech executives hoping to hype up investors or consumers. A blockchain is really just a distributed, P2P database, sort of like if BitTorrent hosted a database instead of pirated movies and Linux ISOs. The database is chopped up into blocks, and each new block contains a cryptographic hash of the previous block, forming a chain of records that protect each other against alterations. On a traditional database, transactions are verified by the database owner, but on a blockchain, nobody owns the database, so each transaction needs to be verified by many computers. This is the big downside of blockchains: everyone’s constant transaction verifications use a massive amount of electricity and computing power.

The decentralized nature of blockchains means nobody can take down your database, which cryptocurrencies like Bitcoin leverage to make a wealth transaction system that no government controls. But it’s not always clear why you would add all the complication and energy usage of a blockchain to your project.

Not much is known about the group, except that it is focused on “blockchain and other next-gen distributed computing and data storage technologies.” Google’s growth into a web giant has made it a pioneer in distributed computing and database development, so maybe it could make some noise in this area as well.

Continue Reading


The reviews are in: AMD’s mining-averse RX 6500 XT also isn’t great at gaming



Enlarge / The Sapphire AMD Radeon RX 6500 XT, yet another GPU that you probably won’t be able to buy. (credit: Sapphire)

When AMD announced its budget-friendly RX 6500 XT graphics card at CES early this month, the company suggested that the product had been designed with limitations that would make it unappealing to the cryptocurrency miners who have been exacerbating the ongoing GPU shortage for over a year now. But now that reviews of the card have started to hit, it’s clear that its gaming performance is the collateral damage of those limitations.

Reviews from Tom’s Hardware, PCGamer, TechSpot, Gamers Nexus, and a litany of other PC gaming YouTube channels are unanimous: The RX 6500 XT is frequently outperformed by previous-generations graphics cards, and it comes with other caveats beyond performance that limit its appeal even further. (Ars hasn’t been provided with a review unit.)

The core of the problem is a 64-bit memory interface that limits the amount of memory bandwidth the card has to work with. Plus, the card has only 4GB of RAM, which is beginning to be a limiting factor in modern games, especially at resolutions above 1080p. Many tests saw the RX 6500 XT outperformed by the 8GB variant of the RX 5500 XT, which launched at the tail end of 2019 for the same $199 (and you could actually find and buy it for that price).

Read 6 remaining paragraphs | Comments

Continue Reading