Will Mercedes jump to level 3 before Tesla? Looks like it.

[Daniel in SD]

I can understand Mercedes take liability....perhaps the premiums are inbuilt in the price of the car...but it can’t tell you that you have no responsibility....the law is the law and the driver must be in control at all times

The law in California says otherwise and in many other US states too. In Europe, Germany has made L3 systems legal as well.

 

[momo3605]

But, why should Tesla bother with L3 certification? It would detract from the main FSD development and add liability to the company. For what purpose? To sell cars???

I agree with you. At this point, the competition is so far behind Tesla, they don't have any reason to take on liability. If and when competitors start to "fight" back with L3 certification, we might see Tesla step up.

One thing we've seen with Elon is he doesn't like competition beating him. The 0-60 numbers on the Teslas have to be better than the competition...when Lucid dropped pricing on their base model, Elon dropped prices on the model S. When Taycan Turbo S took performance EV king, Elon announced Plaid 2 weeks later...i would expect similar behavior if other manufacturers start releasing more competitive and L3 self-driving features.

 

[OncomingStorm]

per that chart even regular AP/NOA basically is L3 if we are honest. I "drove" for 3 hours yesterday and never took over once (all highway regular NOA). certification or not, thats functional L3.

 

[sleepydoc]

View attachment 784163wow this definition has evolved a bit from previous years . based on this and my experience with the latest FSD beta ( note my further caveat I am not in a city but in the suburbs ) FSD is at Level 3 when I use it today based on these descriptions. Recently I have only had to pop it out of FSD beta under very limited situations ( usually some other idiot driver) , maybe 2-3 more updates and I could honestly say level 3 at 99.9 percent of the time ( probably needs to be 4 to 5 9's) . I wonder what " test" or tests would be to take a claim and certify it as level 3. for me it does this now . and it is definitely not level 4.

I was going to say the same thing. Actually, AutoPilot with AutoSteer met that definition years ago.

 

[Daniel in SD]

per that chart even regular AP/NOA basically is L3 if we are honest. I "drove" for 3 hours yesterday and never took over once (all highway regular NOA). certification or not, thats functional L3.

Well Tesla claims it’s not and of course a L3 vehicle must give the driver adequate warning before asking them to take over.
There’s also the question of reliability. The consensus is that it needs to be safer than the average human driver. Tesla‘s statistics show about 1 collision per 2 million miles which is a high bar. Obviously there’s no way for any of us individually to determine if it’s that reliable.

 

[OncomingStorm]

Well Tesla claims it’s not and of course a L3 vehicle must give the driver adequate warning before asking them to take over.
There’s also the question of reliability. The consensus is that it needs to be safer than the average human driver. Tesla‘s statistics show about 1 collision per 2 million miles which is a high bar. Obviously there’s no way for any of us individually to determine if it’s that reliable.

It very much does warn you to take over, i'm not sure the SAE levels specify how much time the L3 car should give you, nor do i see any specific reliability requirement. The wheel nag and "attention" camera are the only things stopping it from being outright L3 and even that is debatable.

Really tho, IMHO, there is no up side for Tesla going after the L3 tag right now. All it would do is create needless hurdles for them. Most consumers aren't even aware of the "levels" and even if they were, people are focused on what the thing can do (in this context) rather than what its called. And considering tesla pretty much has 0 competition beyond marketing BS from other companies (i mean look at the production and sales numbers of EV's from anyone, let alone what other car makers call ADAS, its laughable). Plus given the fact they, clearly, have no need for marketing anything atm, it would be utterly pointless for them to pursue that and would only take resources away from more important tasks.

 

[mtbwalt]

Humans don’t drive based on vision only except when driving someplace completely new.

Much of our driving is based on memory and experience.

You didn’t consciously “see” that pothole after the turn by the Wendy’s, but you avoided it anyway because you have hit it in the past and swerved around it a hundred times or more.

IMO visual data (which must be prioritized because things can change and information can become stale) augmented by location-aware experience and habits of how the individual driver handles tricky areas would most readily simulate a human driver.

Add to that collective location-based habits of all Tesla drivers and you would have a vehicle that should exceed human capability, especially driving in areas where the human “driver” has never been.

If every time a Tesla encounters a confusing traffic pattern it has to “navigate it for the first time,” it is hard to see a scenario where it will perform anywhere near as nimbly as humans who can navigate it by habit and experience.

 

[Daniel in SD]

It very much does warn you to take over, i'm not sure the SAE levels specify how much time the L3 car should give you, nor do i see any specific reliability requirement. The wheel nag and "attention" camera are the only things stopping it from being outright L3 and even that is debatable.

Really tho, IMHO, there is no up side for Tesla going after the L3 tag right now. All it would do is create needless hurdles for them. Most consumers aren't even aware of the "levels" and even if they were, people are focused on what the thing can do (in this context) rather than what its called. And considering tesla pretty much has 0 competition beyond marketing BS from other companies (i mean look at the production and sales numbers of EV's from anyone, let alone what other car makers call ADAS, its laughable). Plus given the fact they, clearly, have no need for marketing anything atm, it would be utterly pointless for them to pursue that and would only take resources away from more important tasks.

Many consumers want a car where they can legally and safely not have to babysit the car. I don't think Tesla is anywhere close 1 collision (>12mph) per 2 million miles unsupervised so practically speaking I think they would be shut down pretty quickly by the DMV if they declared their system L3.

 

[sleepydoc]

Many consumers want a car where they can legally and safely not have to babysit the car. I don't think Tesla is anywhere close 1 collision (>12mph) per 2 million miles unsupervised so practically speaking I think they would be shut down pretty quickly by the DMV if they declared their system L3.

THere’s not much point in me posting this as you seem hellbent on slamming Tesla no matter what the evidence, but what’s the evidence of Mercedes making 1 collision per 2M miles unsupervised? By your own posts, Tesla met the criteria for L3 some time ago, there are no tests to certify L3 capability/safety and the Mercedes system is actually more limited than Tesla’s.

As far as liability goes, you can bet that Mercedes will have waiver language deflecting their liability for ‘improper use.’ If you don’t think they will then I have a few bridges to sell you.

 

[Bouba]

Humans don’t drive based on vision only except when driving someplace completely new.

Much of our driving is based on memory and experience.

You didn’t consciously “see” that pothole after the turn by the Wendy’s, but you avoided it anyway because you have hit it in the past and swerved around it a hundred times or more.

IMO visual data (which must be prioritized because things can change and information can become stale) augmented by location-aware experience and habits of how the individual driver handles tricky areas would most readily simulate a human driver.

Add to that collective location-based habits of all Tesla drivers and you would have a vehicle that should exceed human capability, especially driving in areas where the human “driver” has never been.

If every time a Tesla encounters a confusing traffic pattern it has to “navigate it for the first time,” it is hard to see a scenario where it will perform anywhere near as nimbly as humans who can navigate it by habit and experience.

That is a good way of describing the process and if Tesla can mimic this then that would be good in my opinion....the car can’t drive anywhere by itself that’s new but if you’ve already been there then the car will be L5 for those places only....I could live with that compromise

 

[Daniel in SD]

THere’s not much point in me posting this as you seem hellbent on slamming Tesla no matter what the evidence, but what’s the evidence of Mercedes making 1 collision per 2M miles unsupervised? By your own posts, Tesla met the criteria for L3 some time ago, there are no tests to certify L3 capability/safety and the Mercedes system is actually more limited than Tesla’s.

As far as liability goes, you can bet that Mercedes will have waiver language deflecting their liability for ‘improper use.’ If you don’t think they will then I have a few bridges to sell you.

By far the most important criteria for L3 is that the manufacturer states that it's L3. I'm very curious how well the Mercedes system will actually work. I suspect they have not tested it enough miles to actually prove statistically that it's safer than humans. They're probably only attempting to estimate the safety.
I can't see how a waiver would be legal since the law includes the SAE definitions and a L3 system has to be incapable of being used outside of its operational design domain. What type of improper use will Mercedes not allow?

Mercedes L3 system is much more limited than Autopilot however their vehicle also has a L2 system which works up to 130mph so it's actually less limited in some ways.
I don't see how it's slamming Tesla to express interest in a system that would drive me without supervision.

 

[Daniel in SD]

That is a good way of describing the process and if Tesla can mimic this then that would be good in my opinion....the car can’t drive anywhere by itself that’s new but if you’ve already been there then the car will be L5 for those places only....I could live with that compromise

There are only 2.7 million miles of paved road in the United States. FSD Beta users alone could cover that in a few hours...

 

[OncomingStorm]

Many consumers want a car where they can legally and safely not have to babysit the car. I don't think Tesla is anywhere close 1 collision (>12mph) per 2 million miles unsupervised so practically speaking I think they would be shut down pretty quickly by the DMV if they declared their system L3.

customers also want a car that can fly, but thats just as unreasonable right now haha. if telsa wanted to they could clearly certify AP/NOA as L3 and if there were other competitors out there i'm sure they would. as to safety, keep in mind AP has been around for a while if the colision rates were super high we'd know about it. hell it seems like everytime i read an article about someone who claims they were using AP during an accident, it's always something stupid like they were sleeping, in the backseat or are clearly lying casue they screwed up and dont want to admit it lol. (not that there haven't been legitimate accidents caused by AP, i'm sure there have been)

 

[Daniel in SD]

customers also want a car that can fly, but thats just as unreasonable right now haha. if telsa wanted to they could clearly certify AP/NOA as L3 and if there were other competitors out there i'm sure they would. as to safety, keep in mind AP has been around for a while if the colision rates were super high we'd know about it. hell it seems like everytime i read an article about someone who claims they were using AP during an accident, it's always something stupid like they were sleeping, in the backseat or are clearly lying casue they screwed up and dont want to admit it lol. (not that there haven't been legitimate accidents caused by AP, i'm sure there have been)

Seems like you misunderstand the difference between a supervised system and an unsupervised system. You can’t judge how well a system would work unsupervised by looking at the raw safety statistics of a supervised system. There are many collisions that are prevented by the human supervision!
I don’t see why it’s unreasonable for customers to want something that you claim is possible…

 

[Raurele]

Two things.

I see zero evidence that Tesla is even interested in L3. The only people that seem to have any interest in L3 is those of us in environments where traffic assist type L3 would be really beneficial. Like I could probably drive from I5 near me all the way past Olympia, WA without ever exceeding the speed limits of L3 systems posed to come onto the market. That is if I left my house between the hours of 3-6pm.

I don't have a lot of faith that the initial roll out of L3 systems in the US will go well. We're total complete idiots on the road. Maybe we should allow Japan and Germany to have a bit of time with L3 before we get our hopes up of it ever being released in the US. Let the safe drivers use it and see what happens.

FSDs goal is level 3. There is no arguing that.

 

[Daniel in SD]

FSDs goal is level 3. There is no arguing that.

You can’t have a L3 robotaxi so the goal of FSD is at least L4.

 

[Raurele]

You can’t have a L3 robotaxi so the goal of FSD is at least L4.

The goal right now is L3. L4 is long term. Can’t skip a step.

 

[BMW740iL]

There are only 2.7 million miles of paved road in the United States. FSD Beta users alone could cover that in a few hours...

I saw Chuck Cook being disappointed that FSD beta 10.11 can not drive around potholes

 

[Daniel in SD]

The goal right now is L3. L4 is long term. Can’t skip a step.

The levels are not steps. There is no evidence that Tesla has plans to make a L3 system.

 

[OncomingStorm]

Seems like you misunderstand the difference between a supervised system and an unsupervised system. You can’t judge how well a system would work unsupervised by looking at the raw safety statistics of a supervised system. There are many collisions that are prevented by the human supervision!
I don’t see why it’s unreasonable for customers to want something that you claim is possible…

L3 is technically supervised, it has the cavot that you must take over if requested to do so. and assuming that the low stats are due to the fact that people are hyper vigilant about avoiding an accident while AP is active seems like a much further reach than the system itself is actually pretty safe lol. both are guesses of course tho the latter has less asumptions.