•   
  •   
  •   

Technology Tesla's 'Dojo' supercomputer will train its vision-centric autonomous tech

12:56  22 june  2021
12:56  22 june  2021 Source:   engadget.com

One of Autopilot's features was active in fatal Tesla crash

  One of Autopilot's features was active in fatal Tesla crash Tesla said Monday that one of Autopilot's features was active during the April 17 crash that killed two men in Spring, Texas. © Scott Engle Scene video following a crash involving a Tesla Saturday night in Spring, TX Police said no one was in the driver's seat at the time of the crash. Lars Moravy, Tesla's vice president of vehicle engineering, said on the company's earnings call Monday that Tesla's adaptive cruise control was engaged and accelerated to 30 mph before the car crashed.

Tesla has gone all-in on vision-only autonomous driving, to the point of even phasing out radar sensors in some of its EVs. Now at a CVPR 2021 workshop, Tesla senior director of AI Andrej Karpathy has explained how it's planning to do this by using an in-house supercomputer called "Dojo," as TechCrunch has reported.

diagram, engineering drawing

Karparthy explained that with vision-only tech, computers must respond to new environments with the same speed and acuity as a human. However, doing that requires AI training on a massive dataset with a powerful supercomputer to crunch it. Tesla has one of those in house with "Dojo," a next-gen model with 1.8 exaflops of performance and 10 petabytes of NVME storage running at 1.6 terabytes per second.

Tesla pushes ahead with 'full self-driving' amid government investigations

  Tesla pushes ahead with 'full self-driving' amid government investigations Tesla seems ready to move forward with a broader roll out of its autonomous driving software after a federal government investigation raised questions about initial, headline-grabbing findings from a fatal Tesla crash in Spring, Texas, last month. © Scott Engle Scene video following a crash involving a Tesla Saturday night in Spring, TX Tesla CEO Elon Musk teased upgrades and the expansion of his company's driver assistance software in a series of tweets Wednesday.

While the system hasn't been benchmarked, Karparthy figures it would be one of the fastest in the world. "If you take the total number of FLOPS it would indeed place somewhere around the fifth spot,” Karpathy told TechCrunch. “The fifth spot is currently occupied by NVIDIA with their Selene cluster, which has a very comparable architecture and similar number of GPUs."

To train the system, Tesla's supercomputer collects video from eight cameras on Tesla vehicles each running at 36 frames per second. While that generates a huge amount of data, it's more scalable than building and maintaining high-definition maps around the world. However, it also requires nearly instantaneous processing, which needs to be treated as a supervised learning problem.

Insurers ask: Who pays when self-driving vehicles crash?

  Insurers ask: Who pays when self-driving vehicles crash? The advent of self-driving cars is raising questions in the insurance sector about who should pay when the vehicles crash and how insurers will set equitable rates. When cars can operate themselves, the central question will be whether accidents are the manufacturer’s responsibility and therefore covered by product liability insurance or whether the fault would lie […] The post Insurers ask: Who pays when self-driving vehicles crash? appeared first on Roll Call.

So far, the system works well in sparsely populated areas, where cars can drive around with no intervention. However, Tesla has found (like all other autonomous vehicle companies) that navigating densely populated areas is much more difficult. Still, Karpathy said that Tesla's computer has been able to handle new types of traffic warnings, pedestrian collision detections and pedal misapplications, the latter happening when a driver accidentally presses the gas instead of the brakes.

Despite several notorious Tesla accidents where the autonomous driving systems failed to pick up obstacles or correctly track a route, CEO Elon Musk is firmly committed to vision-only. "When radar and vision disagree, which one do you believe? Vision has much more precision, so better to double down on vision than do sensor fusion," he tweeted recently. The company believes a supercomputer will finally help vehicles attain advanced self-driving capability, but it's best to take a wait-and-see attitude as we've heard that tune before.

diagram, engineering drawing: Tesla reveals 'Dojo' supercomputer's role in vision-only autonomous driving © Tesla Tesla reveals 'Dojo' supercomputer's role in vision-only autonomous driving

Best TV shows that take place in small towns .
Stacker looked at the best TV shows of all time to find the 25 best series that take place in small towns. Series are ranked by IMDb user rating.

usr: 3
This is interesting!