Autonomous Car Progress

[Dan Detweiler]

When I bought my Model 3, Tesla said that my car had all the necessary hardware for the car to be able to drive itself without a human in it, and would be able to pick the kids up and bring them home from school, etc. Tesla said all that was needed was for the software to be completed and given regulatory approval.

This is why people who paid for FSD at that time were given the new computer for free: Because Tesla had given them a contractual promise that their payment would get them true and genuine FSD (not the watered down "feature complete" that Elon started talking about later).

When Tesla decides that more hardware is needed to fulfill their contractual obligation, they'll have to provide that for free also. Those legacy buyers of FSD were told they would get real driverless operation without paying any additional amount. If Tesla does accomplish FSD but needs more hardware to do it, it will have to upgrade or replace those cars. If it does not accomplish FSD within the normal expected lifetime of those cars, it will be in default, as it promised "If you pay now, your car will do thus-and-such." The risks of promising to deliver something that does not yet exist. If a farmer sells futures and his crop fails, he is legally obligated to pay the value of the crop he was unable to deliver. Tesla sold FSD futures.



When I'm driving I can see if a car is careening at me from the side or from behind. An autonomous car needs to be able to do the same. Camera data is inadequate to determine the speed of another vehicle, which is why Tesla uses radar for TACC. It needs radar (or lidar if they prefer, which Tesla doesn't) for side- and rear-approaching vehicles as well.

Sorry, calling BS on this one. I don't care how good your peripheral vision is or how big your rear view mirror is, you can not see some car coming at you from the side or behind and you certainly can't react in time to prevent an accident. My car has saved me more than once from cars coming at me in the next lane that I never would have seen without Autopilot, and that is with 2018 hardware and software.

Dan

 

[petit_bateau]

@Bladerskb you say,
- Doesn't handle construction.
- Doesn't read hand signals.
- Doesn't handle yield signs.
- Doesn't handle emergency vehicles.
- Can't reverse.

I could go on and on again : please do. I'm not saying you are correct, but I am interested in what is potentially on the "to-do" list to become feature complete.

@powertoold you say,
Not sure what you mean? Like features missing for it to be actually "fsd"? Tesla never really provided us with a list of expected fsd features. Elon mentions feature "complete" but doesn't go into detail on that either. FSD beta has yet to demonstrate
- three point turns and
- reversing on roads (in rare gridlock type situations).

Thank you both. I'm not trying to troll anybody, or ask for a FUD-storm, just trying to figure out how long the "to-do" list is, and what the potential level of difficulty is of each item. Then to mentally look for any progress in each of them being addressed, as progress is definitely being made. It is a separate matter in my mind as to how good the solution is initially for each one, and how rapid the improvement (or degradation) is - initially I am just looking for the qualitative signs that potential solutions are out in the beta-wild. So, seriously, what else is reasonably expected to be on the list so as to achieve "feature-complete" ?

(by the way, the beta videos I've seen appear to show "handling construction")

 

[Bladerskb]

I didn’t say disengagement. I said accident. But yes, it’s an ambitious prediction on my part (6-9 months from end of Oct).

safety related disengagement = accident.
Now you are trying to walk things back, but i won't let you. sorry.

 

[Bladerskb]

@Bladerskb you say,
- Doesn't handle construction.
- Doesn't read hand signals.
- Doesn't handle yield signs.
- Doesn't handle emergency vehicles.
- Can't reverse.

I could go on and on again : please do. I'm not saying you are correct, but I am interested in what is potentially on the "to-do" list to become feature complete.

@powertoold you say,
Not sure what you mean? Like features missing for it to be actually "fsd"? Tesla never really provided us with a list of expected fsd features. Elon mentions feature "complete" but doesn't go into detail on that either. FSD beta has yet to demonstrate
- three point turns and
- reversing on roads (in rare gridlock type situations).

Thank you both. I'm not trying to troll anybody, or ask for a FUD-storm, just trying to figure out how long the "to-do" list is, and what the potential level of difficulty is of each item. Then to mentally look for any progress in each of them being addressed, as progress is definitely being made. It is a separate matter in my mind as to how good the solution is initially for each one, and how rapid the improvement (or degradation) is - initially I am just looking for the qualitative signs that potential solutions are out in the beta-wild. So, seriously, what else is reasonably expected to be on the list so as to achieve "feature-complete" ?

(by the way, the beta videos I've seen appear to show "handling construction")

It handles construction the same way other ADAS systems handles it. which is following obvious/ non obstructing clear path.

But it doesn't actually handle constructions that isn't straight forward and obvious or detours or hand signals which goes along with construction.

 

[powertoold]

safety related disengagement = accident.
Now you are trying to walk things back, but i won't let you. sorry.

Here you go again. Although, I admit the context or detail of my statement / reponse wasn't clear, so we're going to disagree with my vs your interpretation.

 

[petit_bateau]

@Bladerskb
But it doesn't actually handle constructions that isn't straight forward and obvious or detours or hand signals which goes along with construction.
= fair point, thank you.

 

[daniel]

Testing new hardware for future vehicle revisions does not automatically mean that it somehow applies to the legacy fleet.
Put another way, a 2021 car with a newer radar does not mean that Tesla cannot deliver FSD as promised in 2016 or today: Autopilot
And we do not know if this is more hardware, this could be an updated radar unit for the current one.
Just like Model Y introduced the heated radar and older Model 3s did not have.

Of course they would not have to give legacy buyers more hardware than is needed to achieve true driverless operation. But they do have to give them any hardware that is needed for true driverless operation. If you are correct that the HW3 computer is the only thing those older cars need for true driverless operation, then Tesla doesn't have to upgrade their cars further.

But if they decide that a new radar is needed for true driverless operation, then they have to upgrade the legacy FSD cars for free. And if it turns out that other hardware is needed, such as surround radar, they will have to upgrade to that. Tesla has a contractual obligation to upgrade legacy FSD cars to whatever hardware is needed for true driverless operation, and to do it for free.

And I'd bet dollars to doughnuts that today's Teslas will need a significant amount of new hardware to achieve true driverless operation.

Sorry, calling BS on this one. I don't care how good your peripheral vision is or how big your rear view mirror is, you can not see some car coming at you from the side or behind and you certainly can't react in time to prevent an accident. My car has saved me more than once from cars coming at me in the next lane that I never would have seen without Autopilot, and that is with 2018 hardware and software.

Dan

I cannot see everything. But I can see out my side windows and I can see in my rear-view mirror. Tesla autopilot is far and away the best suite of driver-assist features available to consumers today. But it's still Level 2, and to go beyond that will require significant advances. I am of the opinion that some of those necessary advances will involve active surround sensors.

Nobody will be any happier than me if the next firmware update on today's cars achieves city and highway Level 4 (or even Level 3) operation. I'll immediately put in the order to upgrade or trade in my car.

My only point in all this is that Tesla was selling FSD several years ago and insisting that software that did not yet exist could operate on an arbitrary selection of hardware. You can't know what hardware you need until you actually have the software.

 

[Dan D.]

When I bought my Model 3, Tesla said that my car had all the necessary hardware for the car to be able to drive itself without a human in it, and would be able to pick the kids up and bring them home from school, etc. Tesla said all that was needed was for the software to be completed and given regulatory approval.

If you reduce this statement to the absurd, then it may be achievable. Imagine a road with no other cars or people, and imagine that the car is allowed to make the journey with some amount of body damage. Then it can probably achieve this today.

I would assume that most people read the statement as doing the journey safely, in full traffic.

Just saying, that in theory the statement can be read as true...

 

[hgmichna]

When Tesla decides that more hardware is needed to fulfill their contractual obligation, they'll have to provide that for free also.

Maybe they can fulfill their contractual obligation without added hardware. Maybe it just works better with added hardware, but still good enough without. Maybe it can just drive faster with the added hardware. Or maybe there is no contractual obligation.

 

[hgmichna]

I don't care how good your peripheral vision is or how big your rear view mirror is, you can not see some car coming at you from the side or behind and you certainly can't react in time to prevent an accident.

Why? I certainly can see cars coming from behind or from the side, provided I look for them properly, as every driver should. Drivers make mistakes though and overlook things because they didn't look. That's their fault though, not unavoidable reality.

The autopilot is certainly better and safer in this respect (except in Germany, because of unlimited speeds).

 

[daniel]

If you reduce this statement to the absurd, then it may be achievable. Imagine a road with no other cars or people, and imagine that the car is allowed to make the journey with some amount of body damage. Then it can probably achieve this today.

I would assume that most people read the statement as doing the journey safely, in full traffic.

Just saying, that in theory the statement can be read as true...

Except that Tesla cannot equip the car with driverless firmware until it gets regulatory authority to do so, and the promise was that you will be able to do thus-and-such.

Tesla made promises about what we would actually be able to do with our cars. Not promises about what the car could hypothetically do on an empty road. And they promised that at the time we bought our cars, the cars had all the necessary hardware for that purpose. That's why they upgraded the computers for free for FSD buyers: Because they acknowledged that they were mistaken and that the cars didn't have the needed hardware. The same will happen if they decide they need more hardware, such as different sensors. They'll have to replace those also for legacy FSD buyers.

Tesla has painted itself into a corner: By insisting that they could do the job with the then-existing hardware they put themselves into the position where they either need to make do with that hardware, or retro-fit a lot of cars. It was reckless and unnecessary! Selling something you don't have is risky. A farmer needs to sell futures to lock in a good price for his crop, but he also buys crop insurance to cover himself in the event of a crop failure. Tesla sold FSD futures before FSD existed. Tesla should have kept its options open to use whatever sensors turned out to be needed, once the software existed. In the end, they may have to use better sensors, and a lot of people who paid for FSD will have to be answered, either by telling them "Sorry, we lied" or by fixing or replacing a whole lot of cars.

I know a lot of people are insisting that they know what hardware will be needed. But we still don't have the software for true driverless autonomy, and until we do, nobody can know for sure what sensors will be needed.

 

[Dan D.]

Tesla has painted itself into a corner: By insisting that they could do the job with the then-existing hardware
...
In the end, they may have to use better sensors, and a lot of people who paid for FSD will have to be answered, either by telling them "Sorry, we lied" or by fixing or replacing a whole lot of cars.

There's another option if they find they can't do it, and that is to field a poorly implemented FSD using existing sensors and hardware.

 

[daniel]

Tesla has painted itself into a corner: ...

There's another option if they find they can't do it, and that is to field a poorly implemented FSD using existing sensors and hardware.

I actually don't think they'll do that. Tesla makes the safest cars you can buy and has put a focus on safety from the beginning. A poorly-implemented FSD crippled by inadequate sensors would be unsafe, and it would be entirely out of character for Tesla to sell such a car.

It will be full-on or nothing. No half-assed kludge.

 

[DanCar]

Doesn't handle construction.

I've seen plenty of videos where it does handle construction. Where have you seen that it doesn't?

Doesn't read hand signals.

Not a common case. Is this needed for launches a taxi service?

Doesn't handle yield signs.

Plenty of roundabouts with yields that it has handled. What do you mean?

Doesn't handle emergency vehicles.

How do you know this?

Can't reverse.

What makes you say that? With dumb summon it does reverse.

 

[stopcrazypp]

safety related disengagement = accident.
Now you are trying to walk things back, but i won't let you. sorry.

Here you go again. Although, I admit the context or detail of my statement / reponse wasn't clear, so we're going to disagree with my vs your interpretation.

Lol, I see Bladerskb is also trying to set up a strawman against you just like he did to me (3 years ago and trying to bring it up again with me as a "welcome" coming back to these forums 3 years later).
Any update for FSD wide release?

This (bold emphasis mine):

so you truly believe that tesla will release L5 in 2018?

Personally, I find it unlikely even if they get all the software done that there won't be a ton of regulatory red tape to get through, so 2018 doesn't seem likely. Regulators so far seem clueless about how to regulate these vehicles, so it's all a work in progress.

Even one simple thing can stop them, like for example the regulators requiring redundant non-visual sensors for the entire 360 degree view. Or competitors lobby to require Lidar for example, just to trip up Tesla; I would not be surprised if GM did it, given they have a history of lobbying for legislation specifically to trip up competitors (examples: lobby against Tesla direct sales, lobby to ban CHAdeMO in publicly funded chargers).

They might be able to release in one specific jurisdiction less strict though.

Gets twisted into a strawman like this:

oh look whose back. Welcome back.
Did you transition to an alt account here?

What do you think now about your previous comments before that Tesla will have L5 by the end of 2018...you were willing to bet on it before you disappeared?



I have been looking for you for 3 years!!!

Apparently "I find it unlikely" = "willing to bet on it" in his dictionary.

Note you can get a permalink by clicking the post number and also copy and past quote tags. It's pretty clear you were specifically referring to accidents every 150k miles.
FSD rewrite will go out on Oct 20 to limited beta

My estimate / guestimate is three years. What is your estimate / guestimate for no accidents on average every 150K miles?

It's crazy I even think this, but 6-9 months lol. We've seen essentially every human maneuver from this beta, even unintuitive or difficult to program ones like:
...

Accidents are completely different from disengagements and are classified differently by all the AV companies. For example there were only 105 reported AV accidents, but 8885 reported AV disengagements (6186 driver initiated, 2698 AV system initiated, 1 unspecified) in California for 2019, the latest available reported year.
Autonomous Vehicle Collision Reports - California DMV
Disengagement Reports - California DMV

I would be surprised if Bladerskb is not aware of the differences in magnitude given he seems to be following autonomous car news more than most people here. Regardless, the numbers above should make it quite obvious there is huge difference between the two.

 

[stopcrazypp]

Tesla has painted itself into a corner: By insisting that they could do the job with the then-existing hardware they put themselves into the position where they either need to make do with that hardware, or retro-fit a lot of cars. It was reckless and unnecessary! Selling something you don't have is risky. A farmer needs to sell futures to lock in a good price for his crop, but he also buys crop insurance to cover himself in the event of a crop failure. Tesla sold FSD futures before FSD existed. Tesla should have kept its options open to use whatever sensors turned out to be needed, once the software existed. In the end, they may have to use better sensors, and a lot of people who paid for FSD will have to be answered, either by telling them "Sorry, we lied" or by fixing or replacing a whole lot of cars.

Well it gave Tesla essentially interest free loans to further develop the technology. Kind of like all those crowdfunding websites (Kickstarter, Indiegogo, GoFundMe).

 

[Matias]

Is FSD a forward-only driving system? What about reversing?

It's a rainy day. You're reversing down your driveway into the street and some yahoo zips past you at 60mph. Or your driveway is on an angle so the yahoo approaches you at 60mph. If radar is a better way of detecting this, and your car doesn't have rear & side radar, then isn't that a problem?

Do we have tunnel vision and just say, well nobody would reverse onto a street where traffic is going 60mph, so that's that.

Seems like 360° radar is needed if you need it at all.

Many current cars have rear radar.

 

[Matias]

FSD isn't about perfection or dealing with every extreme case.

What does the “F” in the FSD mean?

 

[Dan D.]

Many current cars have rear radar.

True, however Tesla currently doesn't have rear or side long-range radar, nor lidar. The question is can vision-only sensors detect the speed of side and rear-approaching traffic in all weather and situations?

FSD seems somewhat unable to detect fast approaching side traffic - witness the difficulty Beta testers are having crossing traffic. Either the system is not yet aggressive enough to zip across the traffic, or perhaps it just isn't determining their speed in time. It's not clear which.

(also it seems to have some difficulty with forward approaching traffic too, several videos showing the car making mistakes crossing with traffic ahead, mainly failing to commit in time.)

 

[mikes_fsd]

What does the “F” in the FSD mean?

It means what Tesla defined it to be in 2016 during the reveal Autopilot