Nvidia's New 'Robot Brain' Goes On Sale (cnbc.com) 33
Nvidia has launched its Jetson AGX Thor robotics chip module, a $3,499 "robot brain" developer kit that starts shipping next month. CNBC reports: After a company uses the developer kit to prototype their robot, Nvidia will sell Thor T5000 modules that can be installed in production-ready robots. If a company needs more than 1,000 Thor chips, Nvidia will charge $2,999 per module. CEO Jensen Huang has said robotics is the company's largest growth opportunity outside of artificial intelligence, which has led to Nvidia's overall sales more than tripling in the past two years. "We do not build robots, we do not build cars, but we enable the whole industry with our infrastructure computers and the associated software," said Deepu Talla, Nvidia's vice president of robotics and edge AI, on a call with reporters Friday.
The Jetson Thor chips are based on a Blackwell graphics processor, which is Nvidia's current generation of technology used in its AI chips, as well as its chips for computer games. Nvidia said that its Jetson Thor chips are 7.5 times faster than its previous generation. That allows them to run generative AI models, including large language models and visual models that can interpret the world around them, which is essential for humanoid robots, Nvidia said. The Jetson Thor chips are equipped with 128GB of memory, which is essential for big AI models. [...] The company said its Jetson Thor chips can be used for self-driving cars as well, especially from Chinese brands. Nvidia calls its car chips Drive AGX, and while they are similar to its robotics chips, they run an operating system called Drive OS that's been tuned for automotive purposes.
The Jetson Thor chips are based on a Blackwell graphics processor, which is Nvidia's current generation of technology used in its AI chips, as well as its chips for computer games. Nvidia said that its Jetson Thor chips are 7.5 times faster than its previous generation. That allows them to run generative AI models, including large language models and visual models that can interpret the world around them, which is essential for humanoid robots, Nvidia said. The Jetson Thor chips are equipped with 128GB of memory, which is essential for big AI models. [...] The company said its Jetson Thor chips can be used for self-driving cars as well, especially from Chinese brands. Nvidia calls its car chips Drive AGX, and while they are similar to its robotics chips, they run an operating system called Drive OS that's been tuned for automotive purposes.
AI robots? (Score:1)
Well, we will see when the first one kills somebody. Because AI is not suitable to implement machinery safety.
Re: (Score:2)
Re: (Score:2)
If you count self driving cars it has already happened.
Exactly this. An autonomous vehicle --is-- a robot.
Re: (Score:2)
Self-driving cars use a _lot_ more tech besides AI.
Re: (Score:2)
Because what people call AI isn't really AI this is a difficult question to parse.
Re: (Score:3)
Re: (Score:2)
Well, we will see when the first one kills somebody. Because AI is not suitable to implement machinery safety.
End-to-end AI is problematic to use for safety-critical applications. However, no reputable player in the autonomous vehicle arena uses end-to-end. Instead, AI modules are used for certain functions such as part of perception or planning, where the inputs and output are lower-level and more tractable for specification and validation.
Re: (Score:2)
It is a pity that the story is not about autonomous vehicles at all.
Robots ARE Ai. (Score:2)
" robotics is the company's largest growth opportunity outside of artificial intelligence"
??
Re: (Score:2)
" robotics is the company's largest growth opportunity outside of artificial intelligence"
??
This is a very significant statement. The implication is that in Nvidia's eyes, robotics is a nearer-term prospect than autonomous vehicles. In my opinion, both are still long-term prospects. Robotics still needs to add a bunch of functionality, especially if we're talking autonomous mobile robots. Autonomous cars already have most of the required functionality but are still significantly lacking in safety validation, which is an industry wide challenge that currently has no practical solution. Without
Re: (Score:2)
" robotics is the company's largest growth opportunity outside of artificial intelligence"
??
This is a very significant statement. The implication is that in Nvidia's eyes, robotics is a nearer-term prospect than autonomous vehicles. In my opinion, both are still long-term prospects. Robotics still needs to add a bunch of functionality, especially if we're talking autonomous mobile robots. Autonomous cars already have most of the required functionality but are still significantly lacking in safety validation, which is an industry wide challenge that currently has no practical solution. Without that safety validation, we get either Tesla that is willing to sell a wink-wink "safe" Level 2 car masquerading as a Level 4/5 car or Waymo that is so legally cautious that we don't know when when we're go beyond here and there robotaxis.
BTW, solving the safety validation problem in my eyes is equivalent to solving the system validation problem for complex software systems. Whoever can solve that problem will get super rich. I don't expect a solution in my lifetime for either autonomous vehicles or arbitrary complex software systems.
I have been saying for about 10 years -- no one will solve the problem of autonomous vehicle safety while sharing streets and roads with autonomous humans. We will have autonomous vehicles operating at a significant scale when we ban human drivers and pedestrians from using those same spaces. Public opinion isn't there yet, but at some point the "think of the children" and "if it saves even one life it will have been worth it" folks will be galvanized to push it through. Several more high profile terrorist/
Re: Robots ARE Ai. (Score:2)
The article is mixing two different products up and itâ(TM)s confusing. Jetson AGX is the basic platform for robotics and AI computing. Drive AGX utilizes the same chip, but there are other features that do not exist in Jetson, like ASIL-B compliance for the big ARM cores when running QNX, and ASIL-D compliance when running an AUTOSAR stack on the functional safety island (FSI) based on Cortex-R cores. The FSI could also run FreeRTOS to avoid costs with FuSa software but still be a hard realtime enviro
Re: (Score:2)
Digital twin platforms bridge the gap between AI and robotics, and NVDA's story about that is that they are uniquely positioned to make high fidelity digital twins of real world situations robots will find themselves in, and using these high fidelity digital twins, they can train the robot's brain faster than real-time and back test it, simulating life-like sensor data from cameras and the like, without any danger of it damaging something in the real world if the robot screws up while doing it.
Re: (Score:2)
Excuse me? (Score:2)
The company said its Jetson Thor chips can be used for self-driving cars as well, especially from Chinese brands.
Why does country of origin matter when it comes to self-driving? I would think, outside of slight law variances, self-driving cars are self-driving cars. :Or are they just saying they have more interest in Chinese brand self-driving cars than in US brand self-driving cars? And if so, why? Would that be due to pricing? Regulation? Some form of 'ism' that I can't wrap my head around? Or just trying to get in on what they see as the biggest potential market for data-aggregation via in-service hours?
Re: (Score:3)
Based on their press release [nvidia.com], it looks like it's just the case that Chinese automakers are using them more. On their "ecosystem" list, you don't see any Waymo/Jaguar, Tesla, or other companies making cars for the U.S. market (except Volvo).
Re: (Score:2)
Based on their press release [nvidia.com], it looks like it's just the case that Chinese automakers are using them more. On their "ecosystem" list, you don't see any Waymo/Jaguar, Tesla, or other companies making cars for the U.S. market (except Volvo).
So, they just wrote that sentence in a weird way that implies the chips are better suited to Chinese brands, rather than saying they are being used more in Chinese brands. Neat. And they say journalism is dead.
Re: (Score:2)
Re: (Score:2)
T for Terminator (Score:3, Funny)
Re: (Score:2)
Re: (Score:2)
T5000 ? (Score:2)
Re: (Score:2)
Re: (Score:2)
Skynet was the product, Cyberdyne was the company that created Skynet, so they'd need to change to that.
Ive seen this movie. (Score:2)
And theyre actually calling it the T5000? LOL.
Wait... wut? (Score:2)