Has Tesla even applied for a permit to test driverless cars?

[aronth5]

I think regulators are being sensible and knowledgeable in the case of driverless car testing. This was a surprise to me, couple of years ago I would have expected them to drag their feet and be obstacle for companies like Tesla to make progress. Now it seems that Tesla has not even started the regulatory approval process that does exist. Instead they spread FUD like "regulatory approval is the big unknown" to point fingers to someone else (while the bottleneck is their own R&D)

On long term, the vision only path Tesla chose seems to be the right one: it looks to be on the path of being able to produce L4 with the lowest manufacturing cost. Thus Tesla's bet seems to be paying off. Eventually. Not sure how hard it will be for others to adapt to this and drop lidars.

Regardless of the above, Tesla is not a leader in autonomous cars. At least Cruise and Waymo are way ahead in actually getting real cars on real roads driving real people around. This far Tesla is all big vision and a very promising looking L2 perpetually coming to market in "two weeks".

Everyone is hung up on autonomous cars and I get that but there is a huge market for the "in-between". That is what many people are missing. I bet there are far more people over the next 5 years who still want/need a steering wheel but want to be able to look at their smart phone and/or watch a video even if it means once in a while they have to take over just not instantly. (William Smith in IRobot comes to mind )
And they will pay alot just for that. So it's not just the robo taxi market that is at play here.

 

[t3sl4drvr]

I bet there are far more people over the next 5 years who still want/need a steering wheel but want to be able to look at their smart phone and/or watch a video even if it means once in a while they have to take over just not instantly. (William Smith in IRobot comes to mind )

Why would L3 be significantly easier to build than L4?

I guess the difference would happen in situations where the car gets "stuck" and do not know how to drive itself out of the situation. Instead of just being there, it can ask human to help. Are there that many situations like that? I can imagine that some weird parking garage design and some other special cases like that, but nothing in traffic.

 

[qdeathstar]

Right now it is L2. But I would say it is aiming to be L4 since Tesla is trying to get it good enough to handle driving without a driver.

I don't think Tesla is really trying to do L3. L3 is kind of a special thing where the the car can do FSD but has to notify the driver in x seconds to take over. Currently, FSD Beta does not really notify the driver in advance when to take over. It is more "take over immediately". So I don't think FSD Beta is aiming to be L3.

Beta is not level 3 but I feel like Tesla is aiming for level three and will be satisfied with that. Past level three it begins to be more about liability than functionality.

 

[t3sl4drvr]

Beta is not level 3 but I feel like Tesla is aiming for level three and will be satisfied with that.

Elon Musk says Tesla's Full Self-Driving tech will have Level 5 autonomy by the end of 2021​

 

[Daniel in SD]

To be clear, that is your opinion. We have not heard anything official from Tesla about that.

I think it's extraordinarily unlikely that Tesla will not seek to get FSD approved in California and only deploy it states that don't require approval.
Obviously I was joking about the end of the year. It's funny t

Beta is not level 3 but I feel like Tesla is aiming for level three and will be satisfied with that. Past level three it begins to be more about liability than functionality.

Tesla would be liable for collisions caused by a Level 3 system too. At least that's the way it is for every proposed Level 3 regulatory framework that I've heard about. How can you hold someone liable when the system is driving the car?

Why would L3 be significantly easier to build than L4?

I guess the difference would happen in situations where the car gets "stuck" and do not know how to drive itself out of the situation. Instead of just being there, it can ask human to help. Are there that many situations like that? I can imagine that some weird parking garage design and some other special cases like that, but nothing in traffic.

Waymo and Cruise vehicles get stuck extremely often so they could be theoretically be L3 if they had a driver in the vehicle instead of using remote assistance.

 

[EVNow]

The levels are not a linear progression. You don't necessarily go from L2 to L3. You can go from L2 to L4. it all depends how the system is structured. So if Tesla chooses to create a system where FSD Beta gives the driver advance notice to take over, then it could be L3. But if Tesla chooses to just improve the system to handle things without a driver and the car pulls over when it can't handle something, then it could go straight to L4. It does not have to pass through L3.

The SAE is not specific on how long the advance notice must be but the L3 system that Mercedes is putting out is about 10 seconds warning. Certainly, the warning that FSD Beta gives now to put hands-on wheel is not L3 because it is way too short. It would need to be longer, like approx. 10 seconds advanced warning to be L3.

Its possible for legal reasons they will always advertise L2 and not assume liability until they get to L4.

But, it seems to me - even though cars don't have to go through L3 to L4, thats the most likely "de facto" path Tesla will take. They may never get to "officially you have "x" seconds before responding", but they will try to anticipate and give warnings earlier. Thats the good thing about not following wishy-washy "standards". Or for that matter well laid out standards that aren't elegant / user friendly (like the J-1772 which Tesla refused to use).

 

[t3sl4drvr]

I think the reason why they started insurance business is that they are preparing for FSD. This way they can cover all accidents, even if no insurance company would agree to insure self-driving vehicles.

 

[t3sl4drvr]

But, it seems to me - even though cars don't have to go through L3 to L4, thats the most likely "de facto" path Tesla will take. They may never get to "officially you have "x" seconds before responding", but they will try to anticipate and give warnings earlier. Thats the good thing about not following wishy-washy "standards". Or for that matter well laid out standards that aren't elegant / user friendly (like the J-1772 which Tesla refused to use).

L3 would allow to drop driver attention monitoring. That would be a nice feature for most drivers. Some more people might be willing to pay for FSD subscription if it would give L3 vs L2. Nice, but not a game changer.

L4 is game-changing. It enables robotaxis. This is a new business model that could add one Lyft to Uber worth of market cap to TSLA.

Also, shipping L5 would allow Tesla to recognize all of FSD purchases as revenue - giving them a huge uplift for a quarter.

 

[EVNow]

L4 is game-changing. It enables robotaxis. This is a new business model that could add one Lyft to Uber worth of market cap to TSLA.

Also, shipping L5 would allow Tesla to recognize all of FSD purchases as revenue - giving them a huge uplift for a quarter.

Yes, I'm aware of robotaxi potential

When they enable City NOA for all users (even in Beta), they can recognize vast majority of revenue. BTW, not going to be a big deal in future quarters as more factories come online and Tesla starts selling millions of cars. Might even be ignored by Wall St as a one-off item.

ps :

Currently Uber/Lyft charge about $2/mile and lose a ton of money (because they have to let the drivers keep most of the money - who are also not making all that much money). By charging just $1/mile, Tesla can reap $92k per robotaxi per year - taking into account all expenses like depreciation, tire, fuel, insurance, remote operator … even parking. The analysts are not off - what is unknown ofcourse is if and when Tesla can ever get to true robotaxi.

View attachment 426838

 

[qdeathstar]

Elon Musk says Tesla's Full Self-Driving tech will have Level 5 autonomy by the end of 2021​

Cool. He says a lot of things though.

 

[aronth5]

L3 would allow to drop driver attention monitoring. That would be a nice feature for most drivers. Some more people might be willing to pay for FSD subscription if it would give L3 vs L2. Nice, but not a game changer.

L4 is game-changing. It enables robotaxis. This is a new business model that could add one Lyft to Uber worth of market cap to TSLA.

Also, shipping L5 would allow Tesla to recognize all of FSD purchases as revenue - giving them a huge uplift for a quarter.

Its a game changer for many but obviously not for all. That was my point earlier about robotaxis. Yes that market is huge but I suspect FSD for the average driver at least in the next 5 years will generate much higher revenue. After 5 years who knows.

Also remember that as soon as snow/ice storms and harsh winter enter the picture having a driver for some rides will be required until Tesla can update the car to handle the worst winter conditions. Pre-mapped routes (i.e. GM Cruise/Waymo) will often be cast aside by the driver when drifting snow changes what a driver does to get to their destination.

FSD is about incremental progress and how it gets adapted along the way.

 

[Terminator857]

... Tesla would be liable for collisions caused by a Level 3 system too. ...

There is an inbetween. What happens when pings to keep your hands on the wheel increase in time period? And yes can be called sleezy L3 or L2.9+.
I suspect Tesla will take this route long before it releases real L3 or L4.

 

[Daniel in SD]

Its possible for legal reasons they will always advertise L2 and not assume liability until they get to L4.

I am extraordinarily skeptical that this plan would work. Elon has already said that there is a risk of complacency with current FSD Beta. What will happen when FSD Beta can go 10,000 miles without a safety disengagement? I think automation complacency will be a big problem.

There is an inbetween. What happens when pings to keep your hands on the wheel increase in time period? And yes can be called sleezy L3 or L2.9+.
I suspect Tesla will take this route long before it releases real L3 or L4.

L3 systems don't require your hands to be on the wheel or your eyes to be on the road. If Tesla were to market a L2 system that way I think they would have legal problems.

 

[EVNow]

L3 systems don't require your hands to be on the wheel or your eyes to be on the road. If Tesla were to market a L2 system that way I think they would have legal problems.

They will not market it to be L3. They will continue to say you have to put your hands on the wheel and eyes on the road - until they reach L4 or a competitor starts selling L3 systems that threatens their sales / image.

But - they can play around with nags to make them less frequent etc - to give us de facto L3 system.

 

[Microterf]

View attachment 716576

He said they're not working on it, he didn't say they aren't advertising it.

 

[aronth5]

They will not market it to be L3. They will continue to say you have to put your hands on the wheel and eyes on the road - until they reach L4 or a competitor starts selling L3 systems that threatens their sales / image.

But - they can play around with nags to make them less frequent etc - to give us de facto L3 system.

Remember Tesla now uses the internal camera to track the driver which has always been considered a short coming. Not sure how that will affect use of nags. Of course with potential FSD drivers trying to get a high safety factor number the last thing you want to do is see how well it works by looking at your phone or looking away. This will significantly reduce the criticism of FSD from many 3rd party reviewers (i,e, Consumer Reports)

 

[srs5694]

Elon Musk says Tesla's Full Self-Driving tech will have Level 5 autonomy by the end of 2021​

That article was published on January 27 of 2021. Given "Elon Time," and when the claim was published, I'd say mid-to-late 2023 is more likely.

 

[Terminator857]

... If Tesla were to market a L2 system that way I think they would have legal problems.

I'm thinking of an L3 system marketed as L2 because manufacturer isn't quite sure they want to accept liability yet.

 

[Daniel in SD]

I'm thinking of an L3 system marketed as L2 because manufacturer isn't quite sure they want to accept liability yet.

Haha. There is no way Elon Musk could resist saying something like "Tesla Full Self-Driving works at a safety level well above that of the average driver." after reaching that goal. He might even say it if it weren't true!

 

[Terminator857]

Haha. There is no way Elon Musk could resist saying something like "Tesla Full Self-Driving works at a safety level well above that of the average driver." after reaching that goal. He might even say it if it weren't true!

Elon can say alot of things but if Tesla isn't accepting liability then it is still an L2 system.