Is model 3 ready for future self driving?

[OPRCE]

I am guessing this is why Tesla removes it as an option because realistically no one can tell when it will be ready for public consumption.

I expect "FSD" features will be released piecemeal, legally still as a L2 supervised driver assistance system, until the ensemble has been validated by several billion on-road miles driven, before being declared out of beta sometime around 2022. At that stage it should be capable of L3 autonomy on all public roads, if having effective Driver Attention Monitoring (e.g. IR camera, like Cadillac SuperCruise) but otherwise still on current sensor suite.

i.e. the human crutch will be used to support gradual development, same as current ALC still requires confirmations for some time until reliability improves.

 

[S4WRXTTCS]

I wouldn't worry too much about it at this point in time.

The reason is we don't know what will be required for L3 self-driving or above. We can't really take Tesla's word for it because Tesla themselves don't have control over that. Tesla is well aware of this, and that's why they have the disclaimer that it's up to regulatory requirements.

Anything beyond L2 is going to depend on regulatory requirements.

There are some people like Blader (another user on TMC) that claims L3 will require a driver attention monitoring system, and the Model 3 doesn't have this. Who knows if the existing interior camera in the Model 3 can be used for this function. I highly doubt it, and the argument for requiring it is pretty solid so I can't see it working without it.

There are also those like myself who believe it will require a better sensor suite to achieve L3.

I have EAP+FSD, and even I don't expect to have full-self driving in my Model 3. You might ask why I got the FSD package in that case, and my answer is the journey towards self-driving is good enough for me. I'm interested in it, and it's one of the reasons I have the car.

A good example of a car company with a L3 capable car is Audi. Sure it's not capable of highway speeds, but it can be in L3 mode in slow traffic. This vehicle will be available with L3 capability in Germany in 2019, but won't be in the USA. The lack of a clear regulatory requirements across the entire country is cited as one of the reasons Audi decided not to allow it.

So even the right hardware, and software doesn't get someone to L3.

 

[Knightshade]

That Tesla quote does not claim that the car has FSD hardware but that self driving ‘capabilities’ will be offered in the future.

I'm not sure you read the quote.


Here it is again.

Model 3 comes standard with advanced hardware capable of providing Enhanced Autopilot features today, and full self-driving capabilities in the future.

So yes, they claim it comes...standard... with the hardware capable of providing self-driving capabilities in the future.

This is demonstrably false since they've already said HW3 is required for self-driving capabilities...therefore cars today do not come standard with HW capable of doing that.

 

[Vitold]

I'm not sure you read the quote.


Here it is again.




So yes, they claim it comes...standard... with the hardware capable of providing self-driving capabilities in the future.

This is demonstrably false since they've already said HW3 is required for self-driving capabilities...therefore cars today do not come standard with HW capable of doing that.

I see the problem - you are disregarding the comma. Tesla writes that car will have self driving capabilities but it does not imply anywhere that will do so with current hardware.

 

[aronth5]

I'm not sure you read the quote.

Here it is again.
So yes, they claim it comes...standard... with the hardware capable of providing self-driving capabilities in the future.

This is demonstrably false since they've already said HW3 is required for self-driving capabilities...therefore cars today do not come standard with HW capable of doing that.

What I find interesting in Tesla's language is "self driving capabilities in the future". By adding the word "capabilities" it implies FSD features will be rolled out over time which is precisely what I have assumed since I purchased by 3.
However, there are many on this forum that say FSD has to be all or nothing. Technically they may be correct but realistically that isn't how it will get rolled out.

 

[Knightshade]

I see the problem - you are disregarding the comma. Tesla writes that car will have self driving capabilities but it does not imply anywhere that will do so with current hardware.

No, you're not understanding the comma.

You can tell the self driving is related to the hardware because that's the thing it actually is talking about in the sentence

The entire rest of the sentence, before and after the comma, explain the things that hardware makes possible both now, and in the future.

There's no other way to read the sentence.

If self driving was not related to the hardware coming standard it would be in its own sentence, not the second half of a single one explaining what the standard hardware makes possible.

Or you could also look at the dozen other things Tesla said elsewhere (including their own website) in the past regarding the 2.0 (and 2.5) hardware, where, again, they repeatedly make it clear the hardware that comes standard enabled self driving... but that the software isn't ready yet.

The existence of HW3, and the fact it will be required for FSD features, tells you Teslas previous statements, and the one I just quoted that's still on their website, are factually incorrect.

What I find interesting in Tesla's language is "self driving capabilities in the future". By adding the word "capabilities" it implies FSD features will be rolled out over time which is precisely what I have assumed since I purchased by 3.
However, there are many on this forum that say FSD has to be all or nothing. Technically they may be correct but realistically that isn't how it will get rolled out.

Absolutely.... and it's what I've said all along.

Nobody goes from a good L2 system to a working L5 system in one step.

 

[N5329K]

Yep, right now Tesla employees are Beta-testing HW3 with the expanded neural nets, which reportedly handles roundabouts, stop signs, traffic lights, etc., very competently, so we should certainly expect to see some sort of incomplete and supervised "FSD" features released to the public fleet this year.

The problem comes in the substantial gap between the “full” part of FSD and “incomplete”. Perhaps it should have been sold as Incomplete Self Driving (ISD) or Pro Tem Self Driving (PTSD). Wait...
Still, be interested to see what part of “full” actually becomes usable near-term.
Robin

 

[Knightshade]

The problem comes in the substantial gap between the “full” part of FSD and “incomplete”. Perhaps it should have been sold as Incomplete Self Driving. Be interested to see what part of “full” actually becomes usable near-term.
Robin

I've seen 2 schools of thought on this...

One is the "EAP but better" path where FSD initially is pretty much a more reliable Level 3 version of EAP... (no more hands on wheel required, feel free to read a book or the like if state law allows-handles things like merges a lot better- and probably reads speed limit signs- but still very domain-restricted to divided highways.).... this seems more likely to me, as they're already mostly there (hell Audi already has a L3 system in production though the domain is both divided highway AND speeds below 37 mph), it's the least "risky" place to move to L3 since no pedestrians, etc.

The other is the "local roads" version where it's still level 2 but recognizes stops signs, traffic lights, can handle local intersections and turns without a car to follow, etc... this seems a lot less likely to me because it's a lot more complex problem....and more importantly "Self driving Tesla mows down baby carriage" is a not a good headline for an initial roll out feature.


Honestly if all FSD ever gets me is a dead reliable L3 or L4 version of EAP I'm more than happy with the $ I paid for it- especially given ~95% of my daily drive is divided highways.

 

[OPRCE]

The problem comes in the substantial gap between the “full” part of FSD and “incomplete”. Perhaps it should have been sold as Incomplete Self Driving (ISD) or Pro Tem Self Driving (PTSD). Wait...
Still, be interested to see what part of “full” actually becomes usable near-term.
Robin

LOL, yes, PTSD has already had a few votes, now we just have to nail down the form of words to make the acronym describe what we in fact get!

As a counterproposal to yours, how about Perennially Treacherous Self Driving?

 

[OPRCE]

Please don't ever say to Tesla that you want a

dead reliable

system, as they are highly susceptible to interpret that the very wrong way!

Come to think of it they have already delivered a "dead reliable" AP in the sense that it is currently guaranteed to kill you if on the motorway at 80mph a lead vehicle steps aside to reveal a traffic jam 200m ahead and you are distracted fiddling with the controls at precisely those wrong few seconds.

 

[MaryAnning3]

I am wondering: should I (can I) pay for FSD now so that I get the HW3 upgrade? I love EAP as it is with my model 3. Would my experience possibly be significantly better with HW3? Can anyone share thoughts on what HW3 is and how it might change the user experience, for example with regard to highway driving or other aspects?

 

[Mikeat00]

2. All new Teslas built after ~15 March 2019 will come with Driving Controller upgraded to HW3.

So where did this date come from?

 

[OPRCE]

So where did this date come from?

Mr Musk announced [by Tweet, I believe] in mid Sept 2018 that he expected HW3 in all new series production cars in 6 months. YMMV.

 

[Knightshade]

I am wondering: should I (can I) pay for FSD now so that I get the HW3 upgrade?

If you want the HW3 upgrade that's the only way to get it, so if you want it then yes.

I love EAP as it is with my model 3. Would my experience possibly be significantly better with HW3?

For just EAP? Probably not. Tesla has explicitly said HW3 isn't needed for EAP only cars.

Can anyone share thoughts on what HW3 is and how it might change the user experience, for example with regard to highway driving or other aspects?

HW3 will enable the first FSD-only features.

Nobody knows what those will be yet (outside of Tesla anyway). Many different paths of speculation there.

Mr Musk announced [by Tweet, I believe] in mid Sept 2018 that he expected HW3 in all new series production cars in 6 months. YMMV.

Oct 16 he said "~6 months" the ~ of course informally means "approximately", or "around", ... so that'd be ~April...or roughly mid-year as I've suggested.

 

[MaryAnning3]

For just EAP? Probably not. Tesla has explicitly said HW3 isn't needed for EAP only cars.

I am not sure what you mean here? What do you mean by "EAP only cars" or "just for EAP"? I thought HW3 was basically a new chip to enable additional enhancement of autopilot?

 

[Knightshade]

I am not sure what you mean here? What do you mean by "EAP only cars" or "just for EAP"? I thought HW3 was basically a new chip to enable additional enhancement of autopilot?

Nope.

HW3 is a new chip to enable any FSD features.

It's not needed for EAP at all.

That's why those who bought FSD get the upgrade free, those who bought EAP don't get it at all (unless they too buy FSD- or wait to buy their car until HW3 is the default chip in every car built, which seems silly since they'd risk missing even more tax incentive and STILL have to buy FSD to get any use out of HW3).

Elon Musk on Twitter

Those who order full self-driving get the upgrade at no cost. It isn’t needed just for enhanced Autopilot

 

[AlanSubie4Life]

I am wondering: should I (can I) pay for FSD now so that I get the HW3 upgrade? I love EAP as it is with my model 3.

If you want the HW3 upgrade that's the only way to get it, so if you want it then yes.

@MaryAnning3 Since you already have your Model 3, and you didn't buy the $3k FSD option, the argument to buy FSD now is pretty weak - except the price may go up from the current (unlisted) $5k (I assume they'll take your $5k right now even though FSD doesn't exist? (So I can't answer your "can I" question)). Personally I'd wait until it's released, quickly go check it out at the Tesla store on a test drive or from a friend who has it, then buy it if you like it & need it, before they raise the price. Likely they'll raise the price relatively quickly if it's awesome. If it isn't great, you don't have to worry about the price increase right away, but you also probably won't want it (win-win in that case!).

My guess is that FSD will start with speed limit recognition as the only feature that can be trusted without fear of death or destruction, and for sure everything will be in beta with no actual true level (3?) or 4 or 5 function (meaning you have to be ready to intervene at any time, possibly immediately), so it's possible you won't want to pay $5k for that. However, if the features truly start to come online subsequently, the price may ramp up. Or the price may come down, if there is relevant competition. Soon after HW3 comes out there will be talk of HW3.5 of course. Might want to wait for that, maybe in your next car!

 

[Knightshade]

@MaryAnning3 Since you already have your Model 3, and you didn't buy the $3k FSD option, the argument to buy FSD now is pretty weak - except the price may go up from the current (unlisted) $5k (I assume they'll take your $5k right now even though FSD doesn't exist? (So I can't answer your "can I" question)). Personally I'd wait until it's released, quickly go check it out at the Tesla store on a test drive or from a friend who has it, then buy it if you like it & need it, before they raise the price. Likely they'll raise the price relatively quickly if it's awesome. If it isn't great, you don't have to worry about the price increase right away, but you also probably won't want it (win-win in that case!).

My guess is that FSD will start with speed limit recognition as the only feature that can be trusted without fear of death or destruction, and for sure everything will be in beta with no actual true level (3?) or 4 or 5 function (meaning you have to be ready to intervene at any time, possibly immediately), so it's possible you won't want to pay $5k for that. However, if the features truly start to come online subsequently, the price may ramp up. Or the price may come down, if there is relevant competition. Soon after HW3 comes out there will be talk of HW3.5 of course. Might want to wait for that, maybe in your next car!

If they can get Tesla to honor a 4k price (which a lot of model 3 buyers had listed as their "post purchase" price at time of purchase, it might be worth doing rather than hoping the current 5k post-purchase price doesn't go even higher when HW3 and FSD features are released.

 

[MaryAnning3]

It's not needed for EAP at all.

I see what you mean. Thanks Knightshade and Alan.

I guess I am wondering, and this is pure speculation, if a faster chip might make EAP even safer? Like for example in an "occluded object situation" (wherein a lead vehicle moves aside due to an obstruction)? Perhaps the higher processing speed might help with repidly labeling and responding to the previously occluded object?

 

[Knightshade]

I see what you mean. Thanks Knightshade and Alan.

I guess I am wondering, and this is pure speculation, if a faster chip might make EAP even safer? Like for example in an "occluded object situation" (wherein a lead vehicle moves aside due to an obstruction)? Perhaps the higher processing speed might help with repidly labeling and responding to the previously occluded object?

It's absolutely possible the new chip could enable things like that.

It's unclear if it will actually do so as part of EAP however, since they seemed to repeatedly say the new chip is only going to be used for FSD-specific stuff.

(some folks question if this is to avoid them having any obligation to upgrade EAP-only cars for free or if they're genuinely reserving all HW3 improvements for FSD features- I don't think we'll know that until HW3 is rolled out though)