Tesla Continues to Walk Back Full Self-Driving Claims

I think in a recent CES talk Mobileye CEO mentioned that Googles models for speech recognition hit a wall after a certain level of expontial improvement. It is not clear that the required robustness and accuracy can be solved with Deep Learning and cameras alone (Not this year or next anyways). I cannot find the exact quote but the talk in itself is very interesting. [1] He is very bullish on vision but says Tesla is making a huge gamble by betting the house on Deep Learning alone. They seem to be far ahead of Tesla in terms of developing a safe and viable FSD solution with the necessary redundancy.

1: https://www.youtube.com/watch?v=h4BXxPvHi60

Solving the 50% case of self driving is something any talented hobbyist can do. Solving for 70% is something that a few companies with billions of dollars will be able to do. Solving for 90% is something that a few companies with billions of dollars of investment may be able to do. Solving for 100% is something that at present no entity on the planet can do. Claiming that you can is disqualifying, it turns you from a 'possible' into a 'must fail' because you lack the realistic constraints required to participate.

Tesla - unfortunately - falls squarely in the latter category, they have set themselves up for failure.

Hacker News was full of people somehow assuming progress there would advance linearly, and plainly telling so ("look how much progress we've made in the last 5 years"). Or comparing the problem of autonomous self driving to much, much more solvable problems (plane autopilots: mostly flying straight lines in a featureless void).

Though to be fair, there were also always people in the comments who realized the immense complexity of the problem, suspecting that fully autonomous driving might need either massive infrastructure changes (bringing those cars closer to something akin of an autonomous train), or generalized AI.

> Go back a few years and many here were incredulous if anyone suggested we might be looking at decades to have full door-to-door self-driving.

As someone who has done this kind of push-back here, I can confirm - this place was highly incredulous of, if not downright hostile, to people with experience in AI who were explaining all these were false promises.

Anyone who actually did robotics knew just how much of a pipe dream this is, how slow progress is, and how solving the easy 50% is worth nothing because the real world gets superlinearly difficult. ML made a great jump, but it only solved parts of the problems we needed solving, and so far failed to introduce a bridge to those other parts. But you try telling people, and they won't listen, and now I understand the curse of Cassandra. :)

But I'm reserving special bile for those CEOs and Directors of Research who knew full well that we don't know how to make this stuff, and that most of it is an open research problem that's decades away from productizing - and yet they marketed it anyway, as if it was a done product that's ready to go, and took customers' and investors' money for something that experts knew was very, very, very clearly vaporware.

It sees difficult for anyone who was in the ML computer vision or self driving vehicle field to have made those predictions, which make it all the more problematic that Tesla still decided too, despite having all the information at hand.

Driving in Jerusalem is a nightmare for humans too, the first video was really impressive.

It’s a shame they didn’t add road sign detection to the overlay it’s harder to judge some of the situations especially around right of way since in Israel right of way is for traffic incoming from the right unless there is a sign that tells you to yield so some of the more dodgy situations involved drivers not yielding correctly.

Overall they do seem to be currently quietly leading the pack and most importantly they are delivering an evolutionary progression as far as assistive driving goes.

Also it’s quite interesting that Mobileye seems to be one of the only few if not the only one that is not relying on deep learning to brute force this problem.

Which is think would be the deciding factor here since it reduces the demand on the hardware and results in a subset of more efficient algorithms which in theory can be formally verified and individually tweaked.

Any sources on Tesla cashing in on revenue from FSD?

Edit. I've found it: https://seekingalpha.com/filing/5494291

> Deferred revenue activity related to the access to our Supercharger network, internet connectivity, Full Self Driving (“FSD”) features and over-the-air software updates on automotive sales with and without resale value guarantee amounted to $2.00 billion and $1.93 billion as of March 31, 2021 and December 31, 2020, respectively.

> Deferred revenue is equivalent to the total transaction price allocated to the performance obligations that are unsatisfied, or partially unsatisfied, as of the balance sheet date. Revenue recognized from the deferred revenue balance as of December 31, 2020 and 2019 was $79 million and $57 million for the three months ended March 31, 2021 and 2020, respectively. Of the total deferred revenue on automotive sales with and without resale value guarantees as of March 31, 2021, we expect to recognize $1.21 billion of revenue in the next 12 months. The remaining balance will be recognized over the performance period which is generally the expected ownership life of the vehicle or the eight-year life of the vehicle.

Oh well. The lawsuits will be interesting to watch. I'm sorry for the people who have been fooled into purchasing FSD.

IMO they have less than 50% chance that they will actually get a car that drives itself without babysitting before 2025.

Also, false advertising.

Harsh but I have to say is there a car company with the kind of hype and fanboyism of Tesla?

I understand 5 years ago when Tesla was doing things nobody else was doing but electric cars are mainstream now. If I had the money I'd be buying a Volvo Polestar.

>Your $10k got you the hardware needed to self-drive to some degree

Don't all new Teslas come with the hardware and the self-driving is a software upgrade? My impression was that the purchase basically locked in a price. I could be wrong on this.

I'll be all-in on self-driving when I can use it to go to the bar and come home, until then it's mostly just an interesting technological exercise.

I think a lot of people saw it like a very, very, very expensive Kickstarter. They paid $10k to help make it happen. As is often the case with Kickstarters, it didn't match the hype.

Not op but I’ll give my reply anyway fwiw.

$7k (x 2) at the time, reason was to help Tesla survive. Was not expecting anything else for my money in the next few years.

Nowadays Tesla’s survival seems more assured, so if I did it today, it would no longer be for that reason.

Some people pay for it thinking that it will be much more expensive later.

I question this thinking though, because once it’s possible, it will be a table stakes feature we take for granted on every car, which means it shouldn’t add a huge bump to the price. But maybe I’m thinking about it wrong.

I wanted long freeway rides and driving in stop and go traffic to be less painful. I wanted to be able to watch my $10k in February 2021 turn into something really cool within a year.

Honestly, I think I’m safer with auto steer on. I’ve had the car hit the brakes when someone turns across my lane and I felt it was too conservative, but then I started wondering if I was too aggressive.

I wanted to watch my Model Y get smarter as OTA updates came in.

If you were making a deposit on a Roadster you were probably already a millionaire (initial deposit was 250k). Also, in an index fund doing 7-8% interest you couldn’t even double that 250k in 5 years. It’s not really fair to make that statement based on actual performance. Your other claims are accurate, but do people care? Tesla fans have shown a lot of patience and they have always delivered eventually. On everything but FSD.

All true. But you could argue that deposits for roadster or semi are made by people who have a lot of excess cash or are able to make their own due diligence on the subject. Neither can be said about the average Joe who shelled out 10 grand last year in expectation that his car will drive itself in a couple months.

I think companies need to develop some products for intermediate stages between where we are now and FSD. Here are some ideas.

1. FSD in certain parking lots. Make deals with some major stores that tend to have big parking lots (such Walmart, Target, Home Depot) to map their parking lots and designate specific areas for rider drop off and for unattended parking. Special signage or road markings would be added to the parking lot to identify these areas and how to move between them.

You can drive to one of these lots, go to the drop off area, and get out of your car. The car then can drive itself to the designated parking area while you go in and shop. When you are done, you can summon it and it drives back to the drop off area to get you.

2. FSD on certain stretches of major freeways. E.g., I-10 between Santa Monica and near downtown Los Angeles, or I-5 from where 99 splits off south of Bakersfield up to where the 580 starts and then up that a ways.

3. A service where for a monthly subscription you get FSD or close to it but only on trips between specific points, such as between your home and office.

The way it would work is that you would drive the route several times and the car records it and uploads it to the company. They use that uploaded data to test their FSD software to see if it can handle that particular route, and tweak it as necessary to either handle places it was having trouble or to hard-code special rules.

If the car notices that something has changed and it isn't sure it can handle it, you have to go back to manual driving and the car uploads footage and sensor data and the company fixes it.

Each route would probably have a setup fee and a monthly maintenance fee.

Am I understanding correctly that Elon is confident in Tesla jumping from level 2 to level 5 in the next eight months?