SoylentNews is people
SoylentNews
Tesla promises 'one million robo-taxis' in 2020
Submitted via IRC for ErnestGoesToSpace
Tesla Promises Investors 'One Million Robo-Taxis' by 2020
To kick things off, the company shared that it had built its very own computer for self-driving cars. The neural network chip was built from the ground up; the project started back in 2016. Each computer (which is stored behind the glove box) has redundancy so that if one chip fails, the second chip can take over.
This is the company's first time building its own silicon. CEO Elon Musk was quick to boast that Tesla " which has never designed a chip, designed the best chip in the world."
Musk reiterated what he's said before about the hardware available in Teslas. "All Tesla cars right now have everything necessary for self-driving available today. All you need to do is improve the software."
That hardware includes the company's reliance on cameras and radar. When the subject of LiDAR (Light Detection and Ranging) came up, Musk said "LiDAR is a fool's errand. Anyone that's relying on LiDAR is doomed." He later added that "it's fricking stupid. It's expensive and unnecessary."
Source: https://www.engadget.com/2019/04/22/tesla-elon-musk-self-driving-robo-taxi/
Tesla Vaunts Creation of 'The Best Chip in the World' for Self-Driving
At its "Autonomy Day" today, Tesla detailed the new custom chip that will be running the self-driving software in its vehicles. Elon Musk rather peremptorily called it "the best chip in the world... objectively." That might be a stretch, but it certainly should get the job done.
Called for now the "full self-driving computer," or FSD Computer, it is a high-performance, special-purpose chip built (by Samsung, in Texas) solely with autonomy and safety in mind. Whether and how it actually outperforms its competitors is not a simple question and we will have to wait for more data and closer analysis to say more.
Robotaxis also at TechCrunch.
(Score: 0) by Anonymous Coward on Tuesday April 23 2019, @06:19PM
> you’ll never get the same performance per unit area out of an ASIC as a full-custom IC
Not sure that the world is quite to that stage of development?
My guess--if Tesla has taken core (and/or bottleneck) portions of their self-driving software and implemented them in an ASIC with minimal overhead, they are going to run a whole bunch faster than on a general purpose processor, and probably faster than on any re-purposed AI engine or GPU.
The problem that I've heard about with the Tesla scheme is that the cameras just aren't very high resolution. The semi that crossed the Florida highway, T-boned by a Tesla, was barely a few pixels in the camera view, until the car was pretty close. Have forgotten details, someone posted images from different distances and human eyes are much better at that distance--an alert driver would have slowed down in advance, anticipating when the semi would finish crossing the highway.