Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Firmware 9 in August will start rolling out full self-driving features!!!

This site may earn commission on affiliate links.
Ok here is my answer to your question @Bladerskb . It appears that Elon considers NOA, either now or the version that will go out with no stalk confirm, to be FSD on highways because it can do the functions required to navigate a highway without driver input (ie take a highway ramp, pass a car, take an exit). I do not agree that NOA now is FSD. So no, I cannot defend that statement from Elon.

No, Elon, the Navigate on Autopilot feature is not ‘full self-driving’

There's alot i can get into here but there's really no point really. But i'm glad you didn't try to find a way to defend it. Neither are others.

However, if NOA in a future update with no stalk confirm can do what Elon describes (ie drive on the highway, pass cars, take exits etc) reliably without any driver intervention or nags, then I would consider that to be highway self-driving.

Even that won't be self driving. Can't be self driving if a human is still monitoring the system's performance and the environment.
 
  • Informative
Reactions: Matias
I did not. HW3 upgrades will only happen once there is any feature whatsoever from FSD released to the public since they will mandate the use of HW3, and have been promised to anyone with HW2/2.5 who have paid for FSD.

I think arguments could be made either way.

I'm strongly of the belief that HW3 will be a significant improvement over HW2/HW2.5 that almost out of the gate we'll see that the HW3 cars do better at various tasks than HW2 because they can take advantage of a much more sophisticated neural network. Like less false ghosting on the side, and even ridiculous things like whether the rear view camera is actually working.

HW3 replaced A LOT of stuff. It's not just an the Neural Net Accelerator, but the entire SOC that was changed out. It's all the same cameras, but what they connect is entirely different.

Whatever was true for HW2/HW2.5 is no longer a guarantee.

I'm expecting to hold off until about 6 months after the first HW3 cars start rolling off the line until I upgrade if it's available. Just to let them work out some expected kinks.

At that point I expect to see a lot of pressure from FSD owners like myself for it. Especially if we see HW3 cars doing better at EAP tasks.

Especially if we can get Fred to complain.

I'm also curious about the accounting side. If Tesla upgrades a car to HW3 then can they take part or all of the money that was set aside for it?
 
I think arguments could be made either way.

I'm strongly of the belief that HW3 will be a significant improvement over HW2/HW2.5 that almost out of the gate we'll see that the HW3 cars do better at various tasks than HW2 because they can take advantage of a much more sophisticated neural network. Like less false ghosting on the side, and even ridiculous things like whether the rear view camera is actually working.

HW3 replaced A LOT of stuff. It's not just an the Neural Net Accelerator, but the entire SOC that was changed out. It's all the same cameras, but what they connect is entirely different.

Whatever was true for HW2/HW2.5 is no longer a guarantee.

I'm expecting to hold off until about 6 months after the first HW3 cars start rolling off the line until I upgrade if it's available. Just to let them work out some expected kinks.

At that point I expect to see a lot of pressure from FSD owners like myself for it. Especially if we see HW3 cars doing better at EAP tasks.

Especially if we can get Fred to complain.

I'm also curious about the accounting side. If Tesla upgrades a car to HW3 then can they take part or all of the money that was set aside for it?

Two answers to your post -

1) @verygreen, @DamianXVI , and @jimmy_d have already taken a look at a 2018.48 vintage firmware for a HW3 APE board and confirmed Tesla is running the same neural nets on that faster board as it is running on AP2/2.5. Obviously this will change over time, but currently the truth is that having a HW3 board today won't result in anything different until the NNs change and that isn't the case at the moment.

2) From the accounting standpoint, Tesla can, and already has "used the money" they took for FSD payments. They simply have a deferred liability setup on their balance sheet that they'll only be able to clear once they deliver FSD in some form. Your point about "if they deliver the new HW3 APE board now, but FSD later, does that allow them to reduce their deferred liability?" is a good one...I'm not sure what their accounting would do for that, but, they certainly have used, and can use "the money" they took for FSD now.
 
Hate to be the realist, but I’m sure HW3 in usual Tesla style will start like AP2.5 or MCU2. The hardware is ready but the software still isn’t quite ironed out. I wouldn’t be surprised to see more software instability with AP3 at first. They’ll surely get it ironed out over time, but it took months before MCU2 people had a way to two-finger-reset their hardware, and even longer for things like sleeping to work properly.
 
100% yes where attentive braking is concerned. Lane following and NOA are equivalent but braking while using EAP is very different between my 2.0 & 2.5 cars.

I think there's a good chance that the radar model is the difference and not the compute... but it could also be the "redundant" GPU they added in 2.5 being used. If it's the radar, then unless they also upgrade the radar as part of the HW3 upgrade it won't help.
 
However, if NOA in a future update with no stalk confirm can do what Elon describes (ie drive on the highway, pass cars, take exits etc) reliably without any driver intervention or nags, then I would consider that to be highway self-driving.

So in your view, a level 2 system -- i.e., one that requires the driver to be paying attention and ready to take over at any moment, which in practice means hands on the wheel and eyes on the road -- but which is "reliable" enough to rarely actually need intervention, is in fact "Full Self Driving"?
 
  • Funny
Reactions: Matias
So in your view, a level 2 system -- i.e., one that requires the driver to be paying attention and ready to take over at any moment, which in practice means hands on the wheel and eyes on the road -- but which is "reliable" enough to rarely actually need intervention, is in fact "Full Self Driving"?

No, I said it was not FSD. I was very careful to say it would be "highway self-driving" not, "highway full self-driving". It would be self-driving but not full self-driving.

What I find noteworthy from Elon's quote is that he does consider it to be FSD because he is defining FSD as being when the car can perform all the necessary driving tasks without driver intervention. And from his answer about traffic lights and stop signs, I would venture a guess that once they reach 99.999% reliability, as he put it, with intersections, traffic lights, stop signs and parking lots, I think Elon will release that under the "FSD" package and just like with NOA, he will declare it to be FSD. I think that is how Tesla will release FSD. They will argue that the car can drive from A to B without driver intervention and park itself, hence it fulfills the promises of FSD.
 
Last edited:
Even that won't be self driving. Can't be self driving if a human is still monitoring the system's performance and the environment.

At the risk of nitpicking, yes it would be self-driving, not full self-driving and still L2, but self-driving nonetheless. When a car is able to drive from A to B without human intervention, that is by definition self-driving. Again, it might be L2 self-driving, but it would still be self-driving. Just because the driver is supervising, does not take away from the fact that if the car is able to drive itself without human intervention, it is still self-driving. You are thinking of full self-driving where the driver does not need to supervise. That is different of course.
 
Last edited:
  • Informative
Reactions: pilotSteve
At the risk of nitpicking, yes it would be self-driving, not full self-driving and still L2, but self-driving nonetheless. When a car is able to drive from A to B without human intervention, that is by definition self-driving. Again, it might be L2 self-driving, but it would still be self-driving. Just because the driver is supervising, does not take away from the fact that if the car is able to drive itself without human intervention, it is still self-driving. You are thinking of full self-driving where the driver does not need to supervise. That is different of course.

The words "self driving" are not a fair use when describing EAP on the highway. Being that I HAVE to keep more torque on the steering wheel (than I did in my BMW to actually drive it) to keep it from nagging me means that it is and should be labeled lane assist. If I am in control, which I am by design as of now, it is not self driving on the highway.
 
  • Like
Reactions: am_dmd
No, I said it was not FSD. I was very careful to say it would be "highway self-driving" not, "highway full self-driving". It would be self-driving but not full self-driving.

What I find noteworthy from Elon's quote is that he does consider it to be FSD because he is defining FSD as being when the car can perform all the necessary driving tasks without driver intervention. And from his answer about traffic lights and stop signs, I would venture a guess that once they reach 99.999% reliability, as he put it, with intersections, traffic lights, stop signs and parking lots, I think Elon will release that under the "FSD" package and just like with NOA, he will declare it to be FSD. I think that is how Tesla will release FSD. They will argue that the car can drive from A to B without driver intervention and park itself, hence it fulfills the promises of FSD.

OK, so then if your car is going to participate in the Tesla Network as a "self-driving" ride share, you just have to hire a driver to sit in the driver's seat and (sometimes) do nothing (but remain alert and attentives at all times!). The money's just going to start rolling in!

Note that Tesla has never backed away from the Tesla Network promise. It was supposed to launch in 2017. And I think it was at last quarter's earnings call that Elon doubled down on it still being in Tesla's future... though perhaps he was talking about a later generation of cars with HW5...
 
I think Tesla is going to end up using the term “full self driving” as a marketing term, not its literal meaning. So full self driving will end up being whatever they define it as (probably more advanced Level 2 stuff where the driver still has to monitor).

I mean, sure they once said the car would drive itself unassisted and there would be a Tesla Network and all, but they have a history of just paving over past statements and pretending they never existed. They have gotten pretty good at that.
 
  • Like
Reactions: am_dmd
OK, so then if your car is going to participate in the Tesla Network as a "self-driving" ride share, you just have to hire a driver to sit in the driver's seat and (sometimes) do nothing (but remain alert and attentives at all times!). The money's just going to start rolling in!

Don't be daft. Of course not! In that scenario, Tesla would release FSD without the Tesla Network. Tesla can only launch the Tesla Network when FSD is so reliable that a person is not required to be in the driver seat at all. But they could and would release FSD before it reaches that point.

Note that Tesla has never backed away from the Tesla Network promise. It was supposed to launch in 2017. And I think it was at last quarter's earnings call that Elon doubled down on it still being in Tesla's future... though perhaps he was talking about a later generation of cars with HW5...

Well, you will note that Elon avoided answering the part of the question about the Tesla Network. He kept his entire answer to just talking about their work towards "FSD". I think Tesla's focus right now is just to deliver some form of "FSD" software update to customers who paid for the option. Basically, their goal right now is to finish the work on traffic lights, stop signs, and complex parking lots as Elon mentioned so that they can release a software update to cars with the "FSD" package that allows them to drive from A to B with minimal driver intervention at last at first. The Tesla Network will come later.
 
I agree that NOA in its current state is not FSD. But here is my question: IF in a future software update,Tesla removes the nags and NOA is able to do highway driving with no nags, no stalk confirmation and no driver intervention, how would that still not be some level of self-driving? Of course it would be!
 
  • Disagree
Reactions: rnortman