Is Tesla closer than we think?

[Microterf]

How feasible or likely is it that Tesla implements driving rules to change how FSD behaves and is able to get to L4 or L5 sooner than we all think?

Similar to how UPS trucks always route right turns unless taking a left is unavoidable. If you avoid the riskiest situations, you can definitely increase your safety by a huge margin.

We keep thinking about Tesla as trying to solve all of these situations and it's obvious that they are trying, but just like an 80 year old grandma, they could avoid certain dangerous situations and still get to the destination fine

 

[diplomat33]

How feasible or likely is it that Tesla implements driving rules to change how FSD behaves and is able to get to L4 or L5 sooner than we all think?

Similar to how UPS trucks always route right turns unless taking a left is unavoidable. If you avoid the riskiest situations, you can definitely increase your safety by a huge margin.

We keep thinking about Tesla as trying to solve all of these situations and it's obvious that they are trying, but just like an 80 year old grandma, they could avoid certain dangerous situations and still get to the destination fine

That does not seem like Elon's M.O. He seems committed to his "general FSD solution".

 

[Microterf]

That does not seem like Elon's M.O. He seems committed to his "general FSD solution".

Right. And that's why we are measuring based on that.

But for example the one thing I see it having problems with is left turns over major roads because of the pillar camera like in the intersection that Chuck Cook tests. If in that case you had a rule to go right, it would help tremendously.

 

[diplomat33]

Right. And that's why we are measuring based on that.

But for example the one thing I see it having problems with is left turns over major roads because of the pillar camera like in the intersection that Chuck Cook tests. If in that case you had a rule to go right, it would help tremendously.

Yes, creating restrictions like no left turns would help a lot. Elon just won't do it. But it's one reason why I predict Tesla will keep driver supervision for a long time to come because I don't think the current sensors can do L5 driverless. Elon will insist on a general solution and since it won't be able to do L5 reliably enough for driverless, Tesla will be forced to keep driver supervision.

 

[rxlawdude]

Elon will insist on a general solution and since it won't be able to do L5 reliably enough for driverless, Tesla will be forced to keep driver supervision. I don't think the current sensors can do L5 driverless.

L5 is "driverless" by definition. But I will be ecstatic for L2+ or L3- without geofencing. Something the competitors cannot do.

 

[diplomat33]

L5 is "driverless" by definition.

The reason I specify L5 is because L4 can also be driverless.

 

[rxlawdude]

The reason I specify L5 is because L4 can also be driverless.

But "driverless L5" is redundant.

 

[diplomat33]

But "driverless L5" is redundant.

No, I don't think it is. "Driverless" could be L4 or L5.

 

[techlogik]

Nowhere close. Since you have real autonomous dedicated companies trying to make self driving cars, and they still haven't perfected and made it possible in a mass manor, don't see how magically TSLA has the magic solution. Seeing v9 etc...FSD beta...wow, sketchy at best.

 

[rxlawdude]

No, I don't think it is. "Driverless" could be L4 or L5.

Driverless L4 is not redundant because a driver may be required. But "driverless L5" is redundant. That's all I'm trying to say.

 

[Cheburashka]

Agree with nowhere close. Even with L3, we need a ton of improvements to be able to safely hand the driving over to the driver. Another challenge in itself..

 

[rxlawdude]

Agree with nowhere close. Even with L3, we need a ton of improvements to be able to safely hand the driving over to the driver. Another challenge in itself..

Thus, expectations should be L2+ or L3-. Any capabilities beyond that on current hardware would be a gift.

 

[diplomat33]

Driverless L4 is not redundant because a driver may be required. But "driverless L5" is redundant. That's all I'm trying to say.

A driver is not required for L4.

 

[shrineofchance]

Similar to how UPS trucks always route right turns unless taking a left is unavoidable. If you avoid the riskiest situations, you can definitely increase your safety by a huge margin.

We keep thinking about Tesla as trying to solve all of these situations and it's obvious that they are trying, but just like an 80 year old grandma, they could avoid certain dangerous situations and still get to the destination fine

Karpathy says something similar at 23:06 in this video:

Personally, I don't know how close Tesla is to Level 4 in an absolute sense, but I think most people vastly underestimate Tesla's relative advantage to other players like Waymo and Cruise. Tesla has 1 million+ drivers generating labelled data every day. The data is automatically curated so that only the most novel, interesting, valuable examples get uploaded. This is a colossal advantage. It enables Tesla to do what other companies just fundamentally can't do.

Per Karpathy:

https://twitter.com/x/status/1363973271717171200

 

[DanCar]

...Tesla implements driving rules to change how FSD behaves and is able to get to L4 or L5 sooner than we all think?
...

My guestimate is Tesla will get to wide level release level 4 in 5+ years. Hopefully in 2 years we will get level 3 on the freeway. For very limited level 4, like limited to a gated old folks community, I still doubt they will do that with in the next 3 years. Good idea to limit left turns, I doubt Tesla will get there in the next 3.5 years.

... Tesla's relative advantage to other players like Waymo and Cruise. Tesla has 1 million+ drivers ...

1 million drivers is nice. I'm not sure it will help them that much. To me Tesla's big advantage is ability to take risk of harming people. I've talked to several at Waymo and Tesla and they are vastly different companies with different attitudes. My assessment is that Waymo has zero ability to take risk of harming people. Until that changes Waymo will be hugely unsuccessful as they are today, losing billions each year.

 

[diplomat33]

Personally, I don't know how close Tesla is to Level 4 in an absolute sense, but I think most people vastly underestimate Tesla's relative advantage to other players like Waymo and Cruise. Tesla has 1 million+ drivers generating labelled data every day. The data is automatically curated so that only the most novel, interesting, valuable examples get uploaded. This is a colossal advantage. It enables Tesla to do what other companies just fundamentally can't do.

Sorry but this is a tired argument that Tesla fans have been pushing for years that is total BS. If Tesla has such a colossal advantage, where is it? How come Tesla is not far ahead of Waymo and Cruise in autonomy? Yes, Tesla's data has allowed Tesla to improve features and deploy features to a large fleet. But those features are L2. Tesla's "data advantage" has not allowed Tesla to deploy any autonomy yet.

If Tesla were really labeling data from 1M drivers every day, for a couple years now, there should not be a single feature that Tesla has not labeled. Tesla should have solved vision already. Clearly that is not the case.

Put simply, Tesla supposedly has had this colossal "data advantage" for years now and they still only have L2 while other companies have L4? That does not make any sense.

 

[shrineofchance]

Yeah, I guess this whole “deep learning” thing is just BS. Never mind what the academics and engineers say! To hell with them!

Why use science to discover the truth when I can use my gut intuition?

 

[shrineofchance]

Put simply, Tesla supposedly has had this colossal "data advantage" for years now and they still only have L2 while other companies have L4? That does not make any sense.

Tesla has had Level 4 since at least 2016

Tesla also has a consumer autonomy product in 50 states, which is 5000% better than Waymo.

 

[diplomat33]

Yeah, I guess this whole “deep learning” thing is just BS.

No! I am not saying that deep learning is BS. Every AV company is using Deep Learning. What is BS (IMO) is that Tesla has some huge advantage in Deep Learning. Where is the evidence of that advantage? I am not seeing it.

 

[shrineofchance]

No! I am not saying that deep learning is BS. Every AV company is using Deep Learning. What is BS (IMO) is that Tesla has some huge advantage in Deep Learning. Where is the evidence of that advantage? I am not seeing it.

Many, many academic and industry sources which I have cited many times here on TMC and elsewhere.