PSA: Please don't expect any FSD features this year, or even early next year, really

[OPRCE]

And Tesla is responsible for accidents the occur while the system is being tested. How fast would the DMV pull their permit if there were multiple major accidents?

I sincerely doubt that actually, as each employee owns the vehicle and remains legally responsible for where it goes on the road, just like any of the rest of us with AP. In effect they are operating as alpha testers to identify rough edges before early-access owners get the software, and AFAICT there is no permit to pull, save perhaps the individual's driving licence, if they end up wrecking someone while DUI or whatever.

 

[S4WRXTTCS]

I've mentioned this before- but there's 2 "easy" fixes for Tesla with the above that makes a lawsuit impossible.

Option 1) EAP is the NN that runs on HW2.5. Period. Even with a HW3 car if you only have the EAP option you get the 2.5 NN. People who paid for FSD get the HW3 NN which has more advanced (and additional, not included in EAP at all) features. No lawsuit basis. EAP people got what they paid for (EAP) since it's feature complete once they remove the nag for lane changes and maybe a slight summon improvement.

Option 2) New cars, once HW3 is in the factory, don't even get an EAP option. It's FSD or nothing. 2.x cars that pay for FSD still get the HW3 upgrade free of course. The HW3 NN as above does more than EAPs NN does. This is marginally cleaner, and allows them to move on from the 2.x codebase a bit sooner, but it also leaves more $ on the table from HW3-era car buyers who might spring for EAP but not FSD.


You can't sue for not having FSD HW in this scenario because if you paid for FSD then you do have FSD HW via the free upgrade, and if you don't then there's no harm/quantifiable damage to the customer to base a lawsuit on.

Option 1 makes the most business sense, and it's probably the most likely outcome. The problem I have with is it doesn't fit the pattern that Tesla has developed over the years. That pattern is including the safety elements of AP/EAP in the car even when the buyer doesn't pay for AP/EAP. So it doesn't make any sense to go with the less capable neural network long term.

Option 2 is the only option that makes sense to me. So I'm a strong advocate of this option, and on a different thread I tried to convince people that they had to stop thinking about EAP and FSD being separate things when it came to HW3. I did because I don't see any other way that Tesla can take an incremental approach to FSD. It's controversial because it's going to be a tough thing to get the pricing right. The danger is pricing FSD too high for people to get or too low to get necessary revenue.

Elon Musk or actually Elon Tusk mentioned on Twitter that there would be a Tesla announcement tomorrow at 2pm pacific. So I'm curious if it will be a HW3 announcement. It's probably a Smart Summons thing which would also be a cool.

 

[S4WRXTTCS]

Dammit, you may be right that this Option 1 is a fatal loophole in my plan for great justice!

OTOH Tesla could have an interest in upgrading the EAP 2.x owners to HW3 in any case, namely the ability to offer free FSD trials to encourage the upsell, but I acknowledge it is not unlikely to happen anytime soon, if ever.

My bet is that they'll try to incentive people to upgrade.

Especially when it comes to HW2.5/MCU2 vehicles because there are no other differences. So on those vehicles its an easy board Computer swap.

Having quarterly driving safety reports puts pressure on Tesla to upgrade people to a more capable system. To get as many miles on that system as possible to differentiate it from the previous system.

 

[ItsNotAboutTheMoney]

You must be new here.

 

[S4WRXTTCS]

5. I'd be shocked if the Model Y to be revealed in a month or so has either of those features, though I agree both are definitely needed for safe FSD. Musk has whimsically talked himself into an ever-deepening ego-trap on the latter and innate obduracy means he cannot admit he was wrong. The possible exits are through successful customer lawsuit or government regulation imposing the requisite robust/redundant solution while allowing him to bluster "I was always right but they forced me."

When it comes speculation about Tesla I'm always torn between what I think is the right approach, and what approach Tesla will take.

I made the decision to speculate on what I believe is the best course of action, and not on what Tesla will actually do.

I also don't need to be right because we already have a TMC member that claims he's always right about Tesla. So someone already has that job.

 

[voip-ninja]

Before I begin I'm going to acknowledge that I'm unable to take Elon seriously when it comes to anything relating to Autopilot or FSD. He lost all credibility in those specific topics. I do take him seriously when he talks about SpaceX though.

It's a strange dichotomy to say the least.

The way I see it is EAP is essentially a piecemeal approach so I don't find it that hard to believe FSD will be anything different. Especially if you remember the old days where they gradually raised the speed at which you could use EAP. Oh, and how they initially started with only one camera activated, and then two cameras.

It's also worth pointing out that there is precedent in the market place of having ADAS systems work at only traffic speed. Like you can drive a 2019 BMW X5 with its enhanced traffic assist package at up to 37mph without hands on the steering wheel. It has a driver facing camera (it's only a visual camera and not the fancy camera like the Supercruise system uses).

The Audi A8 in Germany has L3 capability at up to 37mph on the white listed roads (only divided highways). You can read a book, watch the scenery, etc at these low speeds. Audi hopes to reach 80mph in the near or midterm future.

Will Tesla go that route with FSD? I dunno, but they can.

There is also the question of at what point do EAP features end, and FSD features begin?

The easiest example is stop light detection. I certainly believe Tesla will have stop light detection in HW3 within 3-6 months after release. But, it likely won't be given control over the brakes. They'll probably go through a 3-6 month validation period to make sure the detection rate is accurate enough to enable it.

Tesla likes having lots of time to "validate" things. Like right now we're supposedly validating NoA lane change suggestions so that Tesla can unlock unconfirmed lane changes.

I find it really amazing that people who argue that EAP is almost feature completely so easily glance over two of the hardest things about the EAP. Those two things being Smart Summons, and Unconfirmed lane changes. In both situations the accuracy of the camera detection is paramount.

Even when EAP is feature complete I'd argue that it's still far away from what's really wanted due to weaknesses in the system.

NoA with EAP is essentially L3 ready if it works REALLY, REALLY well. The only difference at that point is transferring liability to the car from the driver for an approved stretch of freeway. You can't sleep as you have to be ready to take over. With L3 you're given a certain amount of time to take over, but it's not enough to take over from sleeping.

So what Tesla is going to do is add features to HW3 vehicles that HW2/HW2.5 vehicles don't get. These features won't specifically target EAP, but will profoundly impact how the vehicles operates under EAP that they might as well.

Things like debris detection, and possibly pothole detection (dunno how well HW3 can detect potholes using the forward facing cameras).

So while you might not see FSD features I will say that you will see some dramatic improvements in some areas.

When will FSD actually be a reality? It will likely start right around the time of EAP on HW3 is so good that it can easily be approved by whoever these regulator people are. In Germany it should be straight forwards as by then the Audi A8 will already be driving on L3 in Germany. In the states the situation is more difficult because it has to be approved state by state. California is the most important state to follow when it comes to Tesla, and I don't know how that will play out. California is such a nanny state that they have ridiculous cancer warnings on everything under the sun. They'd probably try to put a label on the sun if they could. So if L3 can't be enabled in California then I doubt any other state will have it. If it's allowed in California then I expect it to quickly spread.

I don't expect us end users seeing much in the way of surface street FSD except for things like roundabouts, etc. They'll probably sneak in roundabout capability into HW3 EAP, and not say anything about it. Not the crazy roundabouts, but the simple ones.

With L3 on freeways I think we'll hit the limits of what HW3 and the current sensor suite can do.

Usually hardware can only be in front of SW for a generation or two.

The software gets to a point where weaknesses are seen in the hardware, and then the whole pattern repeats. The next car will have rear corner radars, and an inexpensive Lidar. Tesla will simply buy a small Lidar company that has a cool product.

Doing FSD w/out LIDAR seems to be the hill Musk is willing to die on and if I had to guess I'd say that some of the engineering departures in FSD might have been because of this obstinate position.

Maybe they will do FSD w/out it or maybe not.... as LIDAR prices continue to drop it will get harder and harder for Tesla to justify not using the technology when all of their competitors are moving in that direction and it offers some real advantages in real world application.

I'm glad I didn't buy FSD for my Model 3. I don't think HW3 will do much for my personal use of the car and it will be a newer revision vehicle with newer/better sensor hardware that will get us over the line into real FSD in a few years time, at that time I will just change out the entire vehicle.

 

[OPRCE]

I'm guessing, once demand settles down, we will be able to schedule a service visit and get the upgrade installed for something around $2K.

If it lives up to the hype and (initially) provides an increasingly reliable and safe motorway-capable L3 (once nags are removed) at up to 90mph then I cannot see it going for less than $10k. Somebody has to cough for the considerable dev costs!

 

[voip-ninja]

If it lives up to the hype and (initially) provides an increasingly reliable and safe motorway-capable L3 (once nags are removed) at up to 90mph then I cannot see it going for less than $10k. Somebody has to cough for the considerable dev costs!

If it truly lived up to the hype I would gladly part with $10,000 for it but I imagine there are going to be major limitations with what it can do, either from a technology standpoint or from a regulatory one. Probably both and probably for many years.

 

[lolder]

A spinning lidar will always have less raw data than 8 multi-megapixel cameras. That's Musk's gamble. He's a software guy, but wait, he lands booster rockets on robot ships and he's making cars that nobody else can yet match.

 

[dave999z]

When I ordered my M3 in late June 2018, Tesla said it had all the hardware for FSD, and said the price of purchasing FSD after delivery would be $5k. If it is the case that Tesla never releases FSD to run on HW 2.5, then I think, once they do release FSD on HW3, I should be able to get FSD and HW3 for $5k.

Tesla's lack of clarity regarding which features will be part of EAP and which will be part of FSD is pretty annoying. Maybe they just don't know exactly how they will ultimately decide to sort those features, but I don't see what would be so difficult about that exercise. Though it's too late now for people who already purchased (or didn't purchase) EAP and/or FSD.

 

[Daniel in SD]

I sincerely doubt that actually, as each employee owns the vehicle and remains legally responsible for where it goes on the road, just like any of the rest of us with AP. In effect they are operating as alpha testers to identify rough edges before early-access owners get the software, and AFAICT there is no permit to pull, save perhaps the individual's driving licence, if they end up wrecking someone while DUI or whatever.

Tesla has a permit to test self driving cars in California. Trying to circumvent the liability rules of that permit by having employees test in their own vehicles and claiming that they're not actually testing self driving would be ill advised IMHO. I'm no lawyer though!
Tesla does claim that they haven't even started testing self driving on public roads yet: https://www.dmv.ca.gov/portal/wcm/c...596f2625a7f/TeslaMotors.pdf?MOD=AJPERES&CVID=

 

[Knightshade]

When I ordered my M3 in late June 2018, Tesla said it had all the hardware for FSD, and said the price of purchasing FSD after delivery would be $5k. If it is the case that Tesla never releases FSD to run on HW 2.5, then I think, once they do release FSD on HW3, I should be able to get FSD and HW3 for $5k.

I agree.

But I also believe they'd have every right to include a cutoff date, like "HW3 is now available as a free upgrade to all who bought FSD. If you have a 2.x car and have not bought it, you must do so within the next 90 days to purchase at the existing price, after that the post-purchase price for everyone will go up to X.

W
Tesla's lack of clarity regarding which features will be part of EAP and which will be part of FSD is pretty annoying.

And also pretty imaginary.

EAPs functions are very clearly described on the website, and have been for years.

Your Tesla will match speed to traffic conditions, keep within a lane, automatically change lanes without requiring driver input, transition from one freeway to another, exit the freeway when your destination is near, self-park when near a parking spot, and be summoned to and from your garage"

Other than removing the "confirm" requirement for auto lane changes, which they've said is coming very soon, EAP is feature complete already

You're not obligated as an EAP buyer to get anything additional from Tesla (though certainly I'd expect those things it already does to continue, for a while at least, to still be updated to do those things "better" than today)

Everything ELSE is FSD. (We don't know how small the "else" steps will be of course, but if it's not in the EAP description, which again we've known for years, it's not EAP).

 

[voip-ninja]

A spinning lidar will always have less raw data than 8 multi-megapixel cameras. That's Musk's gamble. He's a software guy, but wait, he lands booster rockets on robot ships and he's making cars that nobody else can yet match.

I always find that someone who speaks in such absolutes can rarely be taken seriously. 15 years ago experts were authoritatively declaring that digital sensors would never match the resolution and ISO capabilities of chemical photo film and, here we are today with CMOS sensors capable of 50+ megapixels that can shoot photos at ISO 100,000+.

Not to be too much of a smart-ass, but landing a booster rocket on a barge, in a 3D space where no obstacle avoidance is needed is an infinitely easier problem than having a vehicle moving on two axis deal with the thousands of intersectional problems that happen to drivers every single day on crowded streets and highways.

Musk's argument is that LIDAR is not needed because humans don't have LIDAR, they just have two eyes. So when the Model 3 has a human decision making capability then Musk's argument holds water, until then, augmenting with additional hardware sensors seems like a prudent choice and Tesla themselves are doing it via the use of forward radar and ultrasonic sensors, all of which augment the "eyes" that the car has.

 

[Daniel in SD]

If you want to get some idea of the progress other companies are making on FSD: Autonomous Vehicle Disengagement Reports 2018
Waymo's report is pretty amazing. It looks like they're going about 10,000 miles between disengagements.

 

[Knightshade]

Musk's argument is that LIDAR is not needed because humans don't have LIDAR, they just have two eyes. So when the Model 3 has a human decision making capability then Musk's argument holds water, until then, augmenting with additional hardware sensors seems like a prudent choice and Tesla themselves are doing it via the use of forward radar and ultrasonic sensors, all of which augment the "eyes" that the car has.

The other problem with his argument is he also says he wants FSD to be "much safer" than human drivers.

So maybe additional senses humans don't have could be useful there.... (otherwise why have radar? humans don't)

 

[voip-ninja]

If you want to get some idea of the progress other companies are making on FSD: Autonomous Vehicle Disengagement Reports 2018
Waymo's report is pretty amazing. It looks like they're going about 10,000 miles between disengagements.

Google’s Waymo invests in LIDAR technology, cuts costs by 90 percent

Even geniuses get it wrong sometimes and I think Elon will have gotten it wrong this time. Edison was brilliant but was wrong about DC being superior to AC for transmission over real distances.

Why Thomas Edison Was Both Right—and Wrong—about Direct Current Distribution Networks

 

[OPRCE]

Tesla has a permit to test self driving cars in California. Trying to circumvent the liability rules of that permit by having employees test in their own vehicles and claiming that they're not actually testing self driving would be ill advised IMHO. I'm no lawyer though!
Tesla does claim that they haven't even started testing self driving on public roads yet: https://www.dmv.ca.gov/portal/wcm/c...596f2625a7f/TeslaMotors.pdf?MOD=AJPERES&CVID=

Correct, they have a permit which they have not been using (and quite possibly never have).

The operative part of that disengagement declaration letter is:

“For Reporting Year 2018, Tesla did not test any vehicles on public roads in California in autonomous mode or operate any autonomous vehicles, as defined by California law. As such, the Company did not experience any autonomous mode disengagements as part of the Autonomous Vehicle Tester Program in California.” [emphasis added]

The relevant definitions in California law, linked to in my comment above, state that they only apply to vehicles aiming to implement >=L3 SAE.

If, as has been reported and I presume to be true, Tesla has been testing its pseudo-FSD on public roads, whether in the form of Musk's personal Model S with cutting-edge software & HW3 or, more recently, by several hundred employees with their own vehicles running something similar, then it can only legally be as an extension of the L2 AP platform, via which the new features can conveniently be tested while remaining under constant human supervision.

So, strictly speaking, they are not "actually testing self driving" vehicles, as defined by the law, but rather are testing individual features of a future AV system under the existing AP L2 ADAS regime with full nags and human responsibility.

Hence it is completely legal, or at least that seems to be an interpretation the regulators in California are currently happy to accept.

 

[Daniel in SD]

Correct, they have a permit which they have not been using (and quite possibly never have).

The operative part of that disengagement declaration letter is:

“For Reporting Year 2018, Tesla did not test any vehicles on public roads in California in autonomous mode or operate any autonomous vehicles, as defined by California law. As such, the Company did not experience any autonomous mode disengagements as part of the Autonomous Vehicle Tester Program in California.” [emphasis added]

The relevant definitions in California law, linked to in my comment above, state that they only apply to vehicles aiming to implement >=L3 SAE.

If, as has been reported and I presume to be true, Tesla has been testing its pseudo-FSD on public roads, whether in the form of Musk's personal Model S with cutting-edge software & HW3 or, more recently, by several hundred employees with their own vehicles running something similar, then it can only legally be as an extension of the L2 AP platform, via which the new features can conveniently be tested while remaining under constant human supervision.

So, strictly speaking, they are not "actually testing self driving" vehicles, as defined by the law, but rather are testing individual features of a future AV system under the existing AP L2 ADAS regime with full nags and human responsibility.

Hence it is completely legal, or at least that seems to be an interpretation the regulators in California are currently happy to accept.

Very Legal & Very Cool!
What's the difference between testing FSD and testing pseudo-FSD? You think that Waymo doesn't have full nags after what happened with Uber? haha. So I guess Tesla is arguing that they're not testing FSD because their FSD system isn't good enough to test yet? It all seems like a very shaky legal argument. Why are other companies bothering to report disengagements when they could just be testing pseudo-FSD like Tesla?

 

[lolder]

I always find that someone who speaks in such absolutes can rarely be taken seriously. 15 years ago experts were authoritatively declaring that digital sensors would never match the resolution and ISO capabilities of chemical photo film and, here we are today with CMOS sensors capable of 50+ megapixels that can shoot photos at ISO 100,000+.

Not to be too much of a smart-ass, but landing a booster rocket on a barge, in a 3D space where no obstacle avoidance is needed is an infinitely easier problem than having a vehicle moving on two axis deal with the thousands of intersectional problems that happen to drivers every single day on crowded streets and highways.

Musk's argument is that LIDAR is not needed because humans don't have LIDAR, they just have two eyes. So when the Model 3 has a human decision making capability then Musk's argument holds water, until then, augmenting with additional hardware sensors seems like a prudent choice and Tesla themselves are doing it via the use of forward radar and ultrasonic sensors, all of which augment the "eyes" that the car has.

My comment about the rocket and the cars which are hardware was in mock rebuttal to my statement that Musk was a software guy. I agree about the human decision making and so does Musk. He has said they're teaching the car to "See". He has eschewed the route mapping of others. He wants the car to function autonomously after being plunked down on a dirt road in an Asian "-stan" country. He recently said they could probably get the 2.x cars to drive autonomously but it was a more difficult software task than with the 2000 frame/sec 3.0. Have you seen these:

,

,

.
Time will tell.

 

[Knightshade]

Or Tesla could just be testing >=L3 versions outside of CA to avoid having to publicly report any results.