Live data from Hacker News

Andrej Karpathy leaves Tesla

twitter.com

181–190 of 441 posts

Re: Andrej Karpathy leaves Tesla

#181

Tesla has bet the company on robotaxis, but their vision only tech stack doesn’t seem capable of solving it, which is a problem because Tesla has repeatedly promised FSD is right around the corner, or less than a year away. It’s hard to believe Karpathy would step down if he felt they were close to solving the problem anytime soon. This announcement comes after a 4 month sabbatical where Karpathy said he wanted to ta…

Yeah, but Google's vision + lidar tech also doesn't seem any better at solving it either. They have been working on this problem the longest and they aren't even confident enough to produce a product with it. Google is probably the leader in AI and AI research. They are also the leader in data and mapping. They have billions of cash to play with. Yet it seem like they haven't gotten any closer at solving this problem…

Google/Alphabet is stuck and will make advances in specific territories but will never get there without a fundamental change in approach. Their approach relies on very detailed mapping/modeling of specific terrain, so they can make a usable case sooner, but outside of the map/model territory, they're literally lost. And maps/models change constantly and rapidly.

Tesla is taking a fundamentally more broad and deep approach - working with the fundamental fact that a pair of visual sensors and a compute engine (eyes & brain) can successfully figure out driving in strange areas in real time, ergo, it should be possible without a map/model or lidar. Once they get it solved, it will be solved once and for all. Bigger gamble, bigger payoff. Equipping the car with dozens of eyes is the easy part. The question is whether enough compute power can be brought to bear on solving the recognition problems, and the edge cases. They have obvious issues with failing to recognize large objects like trucks in unexpected orientations, left turns etc. Using millions of miles of live human driver data as a training set is great, except that the average driver is really bad, so it's entirely polluted with bad examples, ESPECIALLY around the edge cases that get people killed. There, examples from professionally trained drivers, who really understand the physics and limits of the car, adhesion, traffic dynamics, etc, are what you want to train on, but that isn't what they have. It is also possible that even if the set of training data would actually be sufficient, the big question will kill them - perhaps the solution requires orders of magnitude more compute power to approach human performance, and they just don't have the hardware to simulate human compute power. So, have they just hit the limits of what their compute power can do?

I think Tesla's approach is fundamentally the way to go, as it is a general solution, compared to everyone else's limited map/model approach.

But both may require either or both a more specifically programmed higher-level behaviors, and/or something much closer to AGI than exists, something that has actual understanding of the machine-learned objects and relationships, which does not yet exist (if one is known, pleas correct me - I'd love to know about it).

Re: Andrej Karpathy leaves Tesla

#182

Earlier quoted context omitted.

I remain convinced that "real" self driving (as in: go ahead and sleep in the backseat) will never happen without changes to road infrastructure and possibly some sort of segregation between robot-driven cars and people-driven cars. Things like traffic signals that actively communicate their status to nearby robot cars (more than just a red lamp that can be occluded by weather, other vehicles, or mud on the camera le…

> I remain convinced that "real" self driving (as in: go ahead and sleep in the backseat) will never happen without changes to road infrastructure and possibly some sort of segregation between robot-driven cars and people-driven cars. Yep, I talk to people working in traffic engineering, and their mindset is always building new road tech and road-side and cloud infra to support autonomous driving. They have no expect…

Which is funny because everything sort of comes back to trains.

The very marginal benefits of laying road vs track more or less disappear when automation in play.

Is is really worth maintaining the ability to go off road/track when you’re not even driving anymore?

Re: Andrej Karpathy leaves Tesla

#183
We haven’t seen Musk’s usual “full self driving is coming Xth of next month” tweet in a while now. I really hope it’s Andrej convince my him to stop, if so that is a great win in my book. After his leaving musk may feel the need to sell his company again to show everything is fine. I will not be surprised to see that tweet again in the next few months.

Re: Andrej Karpathy leaves Tesla

#184

Earlier quoted context omitted.

Isn't Tesla supposed to be producing Optimus, their human-like android, next year? Elon has been over-promising(i.e. flat out lying) about self-driving every year since.. 2014(there's a youtube video compilation of it)? It seems like his strategy is to just come up with increasingly grandiose promises every year when he fails to deliver on his past promises. He's trapped in his swirling vortex of bullshit. Very worry…

Elon in 2024: “by 2026 we’ll have actual, real teleportation” Elon in 2026: “by 2028 we’ll have FTL drives” Elon in 2028: “time machine!”

> Elon in 2028: “time machine!”

We might well have one - back to the stone age based on the potential to escalate ongoing conflicts...

Re: Andrej Karpathy leaves Tesla

#185
post #83

Earlier quoted context omitted.

> Things like traffic signals that actively communicate their status to nearby robot cars (more than just a red lamp that can be occluded by weather, other vehicles, or mud on the camera lens). Or lane markings that are more than just reflective paint, but can be sensed via RF. Rules around temporary construction that dictate the manner of signage and cone placement that the robot cars can understand. The cones might…

> The problem you then face is that any of those could be forged / faked without some kind of way of securely validating the message in some way. You could cause absolute chaos by driving down the road broadcasting false messages. Sounds like a job for...blockchain.

What value would blockchain bring to solving this problem?

Re: Andrej Karpathy leaves Tesla

#186

Tesla has bet the company on robotaxis, but their vision only tech stack doesn’t seem capable of solving it, which is a problem because Tesla has repeatedly promised FSD is right around the corner, or less than a year away. It’s hard to believe Karpathy would step down if he felt they were close to solving the problem anytime soon. This announcement comes after a 4 month sabbatical where Karpathy said he wanted to ta…

Uber bet the company on robotaxis and lost. Tesla is still building very good cars that happen to not be able to drive themselves. Just like every other car. If they could lose their obsession with self-driving and just focus on their incredible cars, they'd still make money.

Tesla is so overvalued only because of their claims of self driving. Musk utters a lot of bs but he's right that without this Tesla is doomed.

They have a big head start, but other car companies are now investing much more in battery tech etc and will quickly catch up. Not to mention Tesla's have terrible build quality, they have a lot of shady business practices like overcounting sales, reusing sold parts etc which came out in the recent leak.

Thing like the 4860 battery which were so hyped turn out to be not that much better. FSD is years away. Stop selling vaporware.

What they need to focus on is things they innovated on like OTA updates, integrated systems, no dealerships etc.

Re: Andrej Karpathy leaves Tesla

#187
post #83

Earlier quoted context omitted.

I remain convinced that "real" self driving (as in: go ahead and sleep in the backseat) will never happen without changes to road infrastructure and possibly some sort of segregation between robot-driven cars and people-driven cars. Things like traffic signals that actively communicate their status to nearby robot cars (more than just a red lamp that can be occluded by weather, other vehicles, or mud on the camera le…

> Things like traffic signals that actively communicate their status to nearby robot cars (more than just a red lamp that can be occluded by weather, other vehicles, or mud on the camera lens). Or lane markings that are more than just reflective paint, but can be sensed via RF. Rules around temporary construction that dictate the manner of signage and cone placement that the robot cars can understand. The cones might…

> You could cause absolute chaos by driving down the road broadcasting false messages.

I’ve personally tested creating a fake toad sign and my tesla reads it as a real sign just fine.

I wonder what happens to autopilot on a 70mph freeway when it encounters a 5mph limit sign…

Re: Andrej Karpathy leaves Tesla

#188

Earlier quoted context omitted.

The situation outside the US is even worse for Tesla. There's a whole swathe of great EVs these days from VW (ID.3, ID.4), Renault (Megane, E-2008), Opel (Corsa-e), Volvo (C40, XC40), Ford (Mustang Mach-E), Jaguar (I-PACE), Polestar, BMW (i4, iX), Audi (e-Tron), Mercedes (EQS), Skoda (Enyaq), etc. The range of models is also much wider in terms of affordability. In Europe, we've reached the point where an EV is just…

And still Teslas sell better both in EU & China than all those EVs you mentioned

Eh? Tesla is third behind VW Group and Stellantis in the EU (fourth if you count Renault-Nissan, but that's a marginal case), and third behind BYD and SAIC in China.

Now, it's true that they're the biggest single BEV _brand_ in Europe (and second-biggest in China), but that's... basically just a different marketing approach. VW Group has multiple constituent brands mostly selling largely the same thing (SEAT/SKODA/Scania/Cupra/VW/Audi); it's no skin off their nose if someone buys a VW id3 or a Cupra Born (these are the same thing).

Re: Andrej Karpathy leaves Tesla

#189
post #92

Tesla has bet the company on robotaxis, but their vision only tech stack doesn’t seem capable of solving it, which is a problem because Tesla has repeatedly promised FSD is right around the corner, or less than a year away. It’s hard to believe Karpathy would step down if he felt they were close to solving the problem anytime soon. This announcement comes after a 4 month sabbatical where Karpathy said he wanted to ta…

> This announcement comes after a 4 month sabbatical where Karpathy said he wanted to take some time off to “sharpen my technical edge,” which makes it sound like this is the result of frustration with the technical approach instead of burnout. I think Karpathy realized (probably way back) that cheap sensors + no HD maps + their (reckless) public testing feedback loop doesn't advance towards L5 self driving and is ba…

Maybe he was just updating himself on the latest DL trends like using transformers in vision and rethinking his original approach?

Re: Andrej Karpathy leaves Tesla

#190
post #12

Earlier quoted context omitted.

> doesn't seem any better at solving it I thought they had real self-driving taxis in Pheonix that you can order? Real ones, with no safety driver. That definitely sounds "better", even if it is heavily geo-fenced.

Waymo has superior performance based on their historical statistics. It makes sense, since their lidar sensors capture more of the environment, and directly in 3D. Their AI also seems better QA’ed. The Tesla AI Day[0] surprised me as it showed they only had a simple architecture for a very long time, simply feeding barely processed camera pixels to a DNN and hoping for the best with little more than supervised learni…

> After all, sure, we humans only take two video feeds as input…

No, we use both our perception and proprioception when we drive or walk for that matter. We have two accute visual sensors mounted in a housing with six degrees of freedom and a sensor feedback mechanism.

We also have fairly sensitive motion and momentum sensors all throughout our bodies. Additionally we have audio sensors feeding into our positional model.

All this data is fed into an advanced computing system that can trivially separate out sensor noise from useful signal. The computing system controls the vehicle, navigates, and even performs low priority background tasks like wondering if the front door of the house was locked.

We have dozens of integrated sensor inputs. It's just silly to assume we only use our eyes when driving. They're certainly an important sensor for driving but definitely not the only one.

Post reply on HN