Tesla "full self driving" discussion
The chip has taped out and they have a prototype tile.
https://www.google.com/amp/s/electre...g-machine/amp/
They don't have a full cluster but the chip has been designed, verified, and in prototype form and taped out. It will take another year to build the actual supercomputer composed of thousands of tiles. It is not on paper only.
https://www.google.com/amp/s/electre...g-machine/amp/
They don't have a full cluster but the chip has been designed, verified, and in prototype form and taped out. It will take another year to build the actual supercomputer composed of thousands of tiles. It is not on paper only.
Great article as usual from Jalopnik, and I agree with most of it, and even then he gives Tesla too much credit. I especially agree with one of the thoughts - concentrating on an autonomous driving system designed specifically for stop and go traffic. Something that works only on highways, up to say 10mph. For anything else, FSD is a giant waste of time and a failed experiment.
https://jalopnik.com/teslas-ai-day-e...eyr-1847526454
https://jalopnik.com/teslas-ai-day-e...eyr-1847526454
Lexus Test Driver
Joined: Jan 2007
Posts: 7,460
Likes: 232
From: CA
Originally Posted by Och
Great article as usual from Jalopnik, and I agree with most of it, and even then he gives Tesla too much credit. I especially agree with one of the thoughts - concentrating on an autonomous driving system designed specifically for stop and go traffic. Something that works only on highways, up to say 10mph. For anything else, FSD is a giant waste of time and a failed experiment.
https://jalopnik.com/teslas-ai-day-e...eyr-1847526454
https://jalopnik.com/teslas-ai-day-e...eyr-1847526454
The article is completely missing the point. Tesla's aim isn't to aid human driving. It's to replace human driving completely. That's the long term goals and it has to start somewhere. You are literally criticizing their R&D showcase and recruiting event aimed at engineers and not the mass public. None of this will be productized anytime soon. It's a peek into the stuff they are working on and they need more engineers in this field.
but to say it's a dangerous and failed experiment ignores all the progress that HAS been made, even if it doesn't currently live up to the hype.
Also, i saw comments on youtube about how 'anyone over 60' is clueless about tech and won't want self-driving cars. That is an utterly clueless assessment. Everyone i speak to 60 and over CAN'T WAIT for self-driving cars, so they can visit the grand kids 50, 100, or 500 miles away without dreading the drive. CAN'T wait to know the car could get them to a hospital if they were in trouble. CAN'T wait to know the car can go get the curbside groceries and bring them back. It's all going to happen in the next decade and millions of boomers are very excited even if some luddites aren't.
Lexus Test Driver
Joined: Jan 2007
Posts: 7,460
Likes: 232
From: CA
I think full autonomous development is just beginning. We will be lucky to see a complete driverless system by 2030. I'm positive that there will be one when I hit my 60s though and it will be glorious. I won't have to rely on my kids or driver when I get too old to drive (hopefully in my 80s).
Lexus Fanatic
Joined: Sep 2010
Posts: 25,917
Likes: 4,274
From: Alberta
I didn't fully appreciate what Tesla is trying to do until A.I. day. Essentially Tesla is recreating reality via camera data then getting the car to identify and react to objects just like a person does. Of course I've always knew this was the goal, but when you actually see all that is involved what they are doing is incredibly ambitious. This goes way beyond autonomous driving the 3D space interaction can be applied to any type of machine including yes, a robot. This is just as much about object identification as it is about driving without hitting anything.
Then there is the predictive nature of the system, being able to guess where an occluded object will be after it becomes visible again has huge implications, expand on that and you should be able to predict a crash before it happens, predict if a person is going to run into the road and so many other events. Expand on that further and I can see it being possible to learn patterns in the broader sense meaning being able to predict based on the motion of say 20 cars when a multi car crash is imminent. There is so much potential.
Then there is the predictive nature of the system, being able to guess where an occluded object will be after it becomes visible again has huge implications, expand on that and you should be able to predict a crash before it happens, predict if a person is going to run into the road and so many other events. Expand on that further and I can see it being possible to learn patterns in the broader sense meaning being able to predict based on the motion of say 20 cars when a multi car crash is imminent. There is so much potential.
ClubLexus Stories
Celebrating Lexus & Toyota from Around the Globe
Top 10 Lexus & Toyotas to Drive Before You Die!
Joe Kucinski
Top 10 Lexus/Toyotas With The LEAST 5-Year Depreciation
Joe Kucinski
Lexus LC500 Convertible Auction: A Preview of Rising Values?
Brett Foote
GX 550 vs TX 550: Best 3-Row Luxury Lexus Family Hauler
Michael S. Palmer
9 Best Lexus Models You Can Buy for Half Price (And 1 You Shouldn't!)
Joe Kucinski
2026 Lexus NX Buyer's Guide: Models, Features, Prices & More!
Brett Foote
Top 10 Reasons to Buy a Lexus TX 550h+ (& 3 Reasons to AVOID!)
Michael S. Palmer
Top 10 Lexus & Toyota Models of the 1990s RANKED!
Joe Kucinski
10 Great Toyotas That Could Have Been Lexus Models
Joe Kucinski
The other huge win is using already known data. So the 'arguments' about how a car can navigate/autopilot when in snow becomes largely irrelevant because the car already knows where the roads go, what lane markings are there without snow, etc. of course there's always scenarios that can't be foreseen, like, a cow walks across the road in a blizzard and the car doesn't see it until 'too late'... well that would have happened with a person too and even then the car might do a better job of an evasive/safest maneuver.
Lexus Test Driver
Joined: Jan 2007
Posts: 7,460
Likes: 232
From: CA
One thing was clear though...Tesla isn't close to getting autonomous auto. They were trying to recruit at the event..haha. They are realizing it's way harder than they imagined at the beginning.
Lexus Fanatic
Joined: Nov 2011
Posts: 35,285
Likes: 311
From: ON/NY
Lexus Fanatic
Joined: Sep 2010
Posts: 25,917
Likes: 4,274
From: Alberta
Toyota saying they'd have 10 EVs by the early 2020, that was a blatant lie. GM said they'd have a bunch of EVs by now, they never intended to or had any way of doing so. They only wanted to appease investors. Are they doing the same thing now? Not sure.
i think you're very far off there. you know i don't approve of the hype and mischaracterization (deceit) of "fsd", but i DO think tesla will get where they're promising at some point. Their cars are basically data gatherers for their central skynet AI datacenters to improve the system, a little like how alexa/siri/google get smarter with feedback loops and data crunching. Those companies have used what they've learned to push more processing BACK to the end points (phones, speakers, etc.), which is great. Tesla cars WILL get smarter and better at what they do. At some point, it will be 'good enough' for the vast majority of driving. Will there still be crashes? Sure.
but to say it's a dangerous and failed experiment ignores all the progress that HAS been made, even if it doesn't currently live up to the hype.
but to say it's a dangerous and failed experiment ignores all the progress that HAS been made, even if it doesn't currently live up to the hype.
Pole Position
Joined: May 2021
Posts: 2,536
Likes: 1,358
From: IL
Och, you reflect all the understanding of the Tesla business philosophy that any reasonable person needs to have.
Lead Lap
Joined: May 2010
Posts: 3,904
Likes: 1,514
From: San Diego, CA
I'm trying to figure out whether it's better to be a righteous Tesla truth-seeker like Och, singlehandedly waging this just and noble crusade to expose this despicable company for the fraudsters that they undoubtedly are, or to be wrong about Tesla, like me, yet up huge amounts of money on their stock and sat on some quite extraordinary gains?
Can anyone enlighten me?
Can anyone enlighten me?