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

Predictions - "Automatic driving on city streets."

This site may earn commission on affiliate links.
2) will also be subject to autonomous regulation when they are shipping autonomous products they declare autonomous.

Seriously? So if Tesla keeps calling it more and more advanced driver assist rather than fully autonomous, they will never be regulated? I actually agree here. I think this is the loophole that will forestall regulation, barring very bad publicity, worse safety than humans or other problems.

And define exactly what you mean by “autonomous products “.
 
Very simply put: When the manufacturer declares a production car responsible for the drive (and that declaration is approved by regulators) it will become autonomous within manufacturer’s declared design domain, not a moment before that.

I just think that opens up a huge loophole. What if the manufacturer, in order to avoid liability, never declares the production car autonomous even after the production car is capable of autonomous driving? Are we really going to say that the car that is capable of autonomous driving is still L2 just because the manufacturer says so? That would be bizarre. Which is why I think that regulators at some point need to step in if the system reaches a certain threshold of autonomy and say that the system is autonomous regardless of what the manufacturer wants.
 
Why would Automatic city driving suddenly mean autonomous if these previous two do not?

I have explained my answer but I am genuinely curious to hear your thinking.
The line has to be drawn somewhere and I think that is where regulators in California will draw the line. I think they can do so the way the rules are currently written the same way they did in the Uber case. I think it is very plausible that Tesla has autonomous production design intent for "automatic driving on city streets" since they say they do when you buy the feature on the website.
I think that a "Level 2" "automatic driving on city streets" feature if it's ever released will end up being very unsafe. Though paradoxically if it's released with similar quality to Smart Summon it might actually be perfectly safe since drivers will have to be super vigilant to even make it around the block. The problem arrises when it starts to work for thousands of miles without a problem, that's where Waymo and Cruise are now and why they've had to take more and more efforts to keep their safety drivers vigilant. I think in Tesla's case it's especially dangerous since it will be used on commutes and many people seem to believe that if they "test" the vehicle on a route and it works a few times that means it will work on the same route in the future.
 
I think everyone is missing the key feature, which is that AP and NAP have to be turned on, and second, they turn off automatically when the driver does not touch the wheel, and they turn off when they reach pre-set limits, in the case of NAP, when it leaves a highway.

Smart Summon is limited in distance and requires the owner to continue to hold the button.

These structural limits mean that no Tesla car can operate "autonomously" - so it does not matter what you call it.

Its a distinction between a true self driving car which has a driver assigned to it but is actually designed to operate autonomously, and not only "designed" to operate that way, but in fact does operate that way.

Due to this discussion, I think this is the key. It does not matter how advanced the capabilities of the car may be, as long as the capabilities cannot operate autonomously they are not autonomous.

The key to regulators is safety, not semantics. As long as the features are safe when supervised, there is no reason to believe regulators will treat them any differently than the existing features.

Now everyone can drift off to the follow-up question, which is whether drivers concentrate more or less.

I frankly think that is what the difference between "feature complete" and actual "FSD" will be. You can have all the features of self-driving but as long as the car is checking every couple of minutes whether the driver is there its not self-driving. And its not self driving if the "self-driving" feature has to be turned on or automatically turns off.
 
I think everyone is missing the key feature, which is that AP and NAP have to be turned on, and second, they turn off automatically when the driver does not touch the wheel, and they turn off when they reach pre-set limits, in the case of NAP, when it leaves a highway.

Autonomous driving systems have to be turned on too. Requiring a person to turn them on does not make them non-autonomous. And turning off because the car is leaving the highway also does not make the car non-autonomous. That is simply the system turning off because it leaving it's operational design domain. A L4 highway autonomous car would also turn off when it leaves the highway since it is only L4 on the highway.
 
Autonomous driving systems have to be turned on too. Requiring a person to turn them on does not make them non-autonomous. And turning off because the car is leaving the highway also does not make the car non-autonomous. That is simply the system turning off because it leaving it's operational design domain. A L4 highway autonomous car would also turn off when it leaves the highway since it is only L4 on the highway.

Right now AP is "autonomous" for a maximum of two minutes (or whatever the time is until you have to prove you are still driving), or until it decides there is too much traffic and turns itself off.

The regulators decided that that's not "autonomous" and I would agree. The main thing about city streets, as a first step, before you even get to right and left turns, is recognizing stop signs and stop lights. But its not just recognizing them, its how to allow the driver to intervene if it does not recognize them.

Sheer speculation would be that the system would be designed to indicate the intersection far enough in advance so that if the driver did not get the indication you take over. Or something like that.

But the point is that Tesla's entire developmental system is based on vehicles which are not autonomous running autonomous features.
 
  • Disagree
  • Like
Reactions: EinSV and mikes_fsd
Seriously? So if Tesla keeps calling it more and more advanced driver assist rather than fully autonomous, they will never be regulated? I actually agree here. I think this is the loophole that will forestall regulation, barring very bad publicity, worse safety than humans or other problems.

That is pretty much how the current rules are laid out. Now, I would never say they will never be regulated, as I’ve said they already are regulated for some forward-looking testing where clearly autonomous production design intent is being tested or demonstrated (Tesla does some smallish amount of autonomous miles)... and all this could change tomorrow anyway if there is, say, a major crash with NoA or Smart Summon and regulators so decide.

But yes, things being as they are, Tesla will be able to add more and more advanced driver assists and not be judged autonomous. Indeed I expect they will.
And define exactly what you mean by “autonomous products “.

In this context cars that are responsible for the drive (ie driver not responsible) at least some of the time in their production design intent. So basically when producing and/or testing SAE Level 3-5 cars.
 
  • Like
Reactions: OPRCE
The line has to be drawn somewhere and I think that is where regulators in California will draw the line. I think they can do so the way the rules are currently written the same way they did in the Uber case. I think it is very plausible that Tesla has autonomous production design intent for "automatic driving on city streets" since they say they do when you buy the feature on the website.
I think that a "Level 2" "automatic driving on city streets" feature if it's ever released will end up being very unsafe. Though paradoxically if it's released with similar quality to Smart Summon it might actually be perfectly safe since drivers will have to be super vigilant to even make it around the block. The problem arrises when it starts to work for thousands of miles without a problem, that's where Waymo and Cruise are now and why they've had to take more and more efforts to keep their safety drivers vigilant. I think in Tesla's case it's especially dangerous since it will be used on commutes and many people seem to believe that if they "test" the vehicle on a route and it works a few times that means it will work on the same route in the future.

I don’t really disagree with any of this, other than to note that the rules being laid out as they currently are does not stop Tesla from releasing a semi-autonomous Automatic city driving system, and not be regulated as autonomous. However, it is quite possible that would be unsafe and result in regulation afterwards for that or other reasons.
 
Maybe Elon is learning not to make predictions he can't keep?

upload_2019-10-11_21-25-49.png
 
I already use AP on almost all roads with a yellow line. I did have trouble with a few intersections in the past but I continue to see improvements in just the last month and rarely take over even in curved intersections. Love it!
AP and NoA are driver assist devices. Even as they rapidly require less frequent intervention, when they truly drive better and safer than us without interventions, it will require regulatory change before any manufacturer removes mandatory driver oversight and assumes liability.
I think city NoA could release this year but probably next year. I understand Tesla is now using their billions of miles of data to program neural networks for this. Should give them enough rare cases to NoA well in most cases with turns and lights, etc. Will be awesome! But of course we will initially need to take control somewhat frequently initially. Tesla will use that data to continually improve city NoA.
For those who don't appreciate Tesla's cutting edge early release style that I believe improves our safety even if not perfect initially, I say don't use it until it meets your standards. (I believe in the not so distant future there will be some who will be unwillingly forced to give up driving because humans will be so much less safe than our computer APs.)
For me, I have been watching this technology develop for many years and am very grateful to be able to use it today.
 
  • Like
Reactions: JohnnyEnglish
Maybe Elon is learning not to make predictions he can't keep?

View attachment 465089

Luckily for us we don’t have to wait for his predictions, as his company is already doing that for us.

We can go to the place where Tesla sells cars based on this information, the Design Studio, and see Automatic city driving is coming later this year (2019).

So 1.5 months maybe, 2.5 months definitely.

I mean they wouldn’t lie to customers buying cars? By now they must know if it comes out this year (or if there is risk it won’t and they need a retraction).
 
  • Love
Reactions: am_dmd and Matias
Luckily for us we don’t have to wait for his predictions, as his company is already doing that for us.

We can go to the place where Tesla sells cars based on this information, the Design Studio, and see Automatic city driving is coming later this year (2019).

So 1.5 months maybe, 2.5 months definitely.

I mean they wouldn’t lie to customers buying cars? By now they must know if it comes out this year (or if there is risk it won’t and they need a retraction).

Maybe you are rubbing off on me but Elon's "a few months" kinda makes me think that "automatic city driving" might get pushed back to say early 2020, I would say around Jan-March 2020. I mean, somehow I doubt "a few months" means in the next 2 months. And Elon's tweets sounds like Tesla wants to coordinate the release of the software to happen at the same time as the hardware upgrades. So my interpretation of Elon's tweet is that Tesla is planning to do the FSD computer upgrades en masse and release the FSD software at the same time around Jan-March 2020 if I had to guess.

And the Tesla website promoted Enhanced Summon as available now... really for months before it actually came out. So there is precedent for Tesla not issuing any retraction. Just sayin'. ;)
 
Maybe you are rubbing off on me but Elon's "a few months" kinda makes me think that "automatic city driving" might get pushed back to say early 2020, I would say around Jan-March 2020. I mean, somehow I doubt "a few months" means in the next 2 months. And Elon's tweets sounds like Tesla wants to coordinate the release of the software to happen at the same time as the hardware upgrades. So my interpretation of Elon's tweet is that Tesla is planning to do the FSD computer upgrades en masse and release the FSD software at the same time around Jan-March 2020 if I had to guess.

And the Tesla website promoted Enhanced Summon as available now... really for months before it actually came out. So there is precedent for Tesla not issuing any retraction. Just sayin'. ;)
@electronblue I mean how dare you all can accuse Tesla of lying! By now we should have been used to this autonomous features. AP1 came with on paper features of meet me at the curb, AP2 came with FSD video, All Tesla sold currently have FSD hardware since 2016 etc ¯\_(ツ)_/¯
 
Maybe you are rubbing off on me but Elon's "a few months" kinda makes me think that "automatic city driving" might get pushed back to say early 2020, I would say around Jan-March 2020. I mean, somehow I doubt "a few months" means in the next 2 months. And Elon's tweets sounds like Tesla wants to coordinate the release of the software to happen at the same time as the hardware upgrades. So my interpretation of Elon's tweet is that Tesla is planning to do the FSD computer upgrades en masse and release the FSD software at the same time around Jan-March 2020 if I had to guess.

And the Tesla website promoted Enhanced Summon as available now... really for months before it actually came out. So there is precedent for Tesla not issuing any retraction. Just sayin'. ;)

Yeah, if they’re not testing this with early access folks before December, I doubt it’s coming until at least February or so (since most companies try to avoid launching new stuff mid-December until mid-January or so)
 
  • Like
Reactions: diplomat33
Yeah, if they’re not testing this with early access folks before December, I doubt it’s coming until at least February or so (since most companies try to avoid launching new stuff mid-December until mid-January or so)

Good point. Usually, EA gets stuff a couple months before the wide release. And I am pretty sure if EA was testing "automatic city driving", we would have heard something about it. Something like that would be hard to keep secret.
 
Maybe you are rubbing off on me but Elon's "a few months" kinda makes me think that "automatic city driving" might get pushed back to say early 2020, I would say around Jan-March 2020. I mean, somehow I doubt "a few months" means in the next 2 months. And Elon's tweets sounds like Tesla wants to coordinate the release of the software to happen at the same time as the hardware upgrades. So my interpretation of Elon's tweet is that Tesla is planning to do the FSD computer upgrades en masse and release the FSD software at the same time around Jan-March 2020 if I had to guess.

And the Tesla website promoted Enhanced Summon as available now... really for months before it actually came out. So there is precedent for Tesla not issuing any retraction. Just sayin'. ;)
Time will tell... I’m going with June 2020.