Elon: "Feature complete for full self driving this year"

[diplomat33]

If what Musk was saying was really ”FSD no geofence feature complete”, he should have said that, instead of yes to ”Level 5 no geofence feature complete”. That would have allowed Tesla to re-define what it all (FSD) means. Referring to Design Studio then would have been much more reasonable. But he said Level 5 and that means a not just reliability but a specific set of features implemented.

And let’s not be naive, he knew what he was doing when he said yes to Level 5. He has been waving the Level 5 card since October 2016 when he announced AP2 as Level 5 capable hardware.

Elon talks about L5 because he believes that FSD will be L5. The ODD will be L5 and he believes that with all the features that Tesla is working on, that our cars will eventually be able to completely drive themselves on any road exactly like the SAE L5 definition that I quoted says.

SAE J3016 specifically allows for Level 4/5 prototypes that are Level 4/5 even without the reliability to be autonomous so that is no excuse.

So I guess when Tesla releases "feature complete L5" at the end of the year, our cars will be L5 even with a safety driver? I can't wait to drive my L5 Tesla. LOL.

Yes, obviously. That is what feature complete and Level 5 mean by their dictionary and SAE J3016 definitions respectively.

Well, I don't believe that Tesla will meet your interpretation of "feature complete L5" by end of this year. I believe Tesla will do traffic light and stop sign and intersections by end of this year. And I believe that Tesla will release those features to the public. But I don't believe that Tesla will complete all the hundreds of features necessary to have what you call "feature complete L5". Because by your interpretation, "feature complete L5" requires hundreds of features that Tesla has not done yet.

 

[EVNow]

I ahve a much better guess when some other companies will reach Level 4 though, since we know a few companies have operated driverless cars in limited fashion.

The problem with those companies is we don't know how fast they can expand to other cities.

I think the problem statement is similar.

Waymo can do L4 in one suburb. Tesla can do L2+ in US. When will they do L4 in top 200 US metros ?

 

[wcorey]

We absolutely should observe critically. But in situations like these where companies do not give the general public access to all the information there are only specific ways you can be critical.

Experts in the field seem to agree, Tesla is far ahead of everyone else because they adapted early in ways that competition has not event done yet, such as putting all the required equipment on production cars years before it had a consumer facing purpose. Collecting the data generated from that hardware to improve the software through Deep Learning.

My assumption is that we are all on the same page for the above.

How close Tesla is to solving that problem is the question. For the most part we are all just spit balling our opinions because the people who know how far Tesla is in the development process are all bound by very powerful NDAs. The things that those people have been able to say seem very optimistic, maybe not quite as optimistic as Musk, but seem to indicate that 1 year later than Musks timeline is likely a very safe bet. Some of these people's entire job is to survey the landscape of things like this and make multi-million dollar bets based on their assessment. Others are just fan-people who got the chance to ride in the car at the Autonomy day demo.

Pretty much anyone not in the group above is just making noise. I am willing to believe Musk in this case because he is looking at data and numbers that nobody else who has spoken publicly has seen. And even if he's wrong the outside analysts with any insight don't put him very far behind what he is saying.

So maybe FC FSD happens sometime next year. I would be disappointed but not surprised. I still think there's a good chance they nail it this year. But to the people saying its 10 years out with zero insight into the actual state of development. You do realize you are making wild baseless claims that someone else who has orders of magnitude more knowledge than you is making baseless claims? Do you not see the irony?

We absolutely should observe critically. But in situations like these where companies do not give the general public access to all the information there are only specific ways you can be critical.

Experts in the field seem to agree, Tesla is far ahead of everyone else because they adapted early in ways that competition has not event done yet, such as putting all the required equipment on production cars years before it had a consumer facing purpose. Collecting the data generated from that hardware to improve the software through Deep Learning.

My assumption is that we are all on the same page for the above.

How close Tesla is to solving that problem is the question. For the most part we are all just spit balling our opinions because the people who know how far Tesla is in the development process are all bound by very powerful NDAs. The things that those people have been able to say seem very optimistic, maybe not quite as optimistic as Musk, but seem to indicate that 1 year later than Musks timeline is likely a very safe bet. Some of these people's entire job is to survey the landscape of things like this and make multi-million dollar bets based on their assessment. Others are just fan-people who got the chance to ride in the car at the Autonomy day demo.

Pretty much anyone not in the group above is just making noise. I am willing to believe Musk in this case because he is looking at data and numbers that nobody else who has spoken publicly has seen. And even if he's wrong the outside analysts with any insight don't put him very far behind what he is saying.

So maybe FC FSD happens sometime next year. I would be disappointed but not surprised. I still think there's a good chance they nail it this year. But to the people saying its 10 years out with zero insight into the actual state of development. You do realize you are making wild baseless claims that someone else who has orders of magnitude more knowledge than you is making baseless claims? Do you not see the irony?

First of all, to @electronblue if by "appeal to authority" you were implying me. No, you are incorrect. I never claimed 'authority' over what happens at Tesla. I do give Musk, perhaps wrongly, the mantle of authority as he is the direct supervisor of the development group in question. Others with more knowledge on this subject than I have called him a micromanager which implies visibility into things he rightfully should leave to others tasked with supervision of that.

I've been very careful to limit my assertion to claiming only feature complete status by the end of the year yet claiming that is on L5 capability is completely coherent. It is not a logical falacy. If you were implying 'authority' to speak on software development life cycle, well...yes...I think in excess of 45 yrs, most recently very senior positions grants me some authority to cite what feature complete means rather than simply being a layman on the subject and resorting to Wikipedia for my technical information.
Completely coherent is distinctly different than is he factually honest. I completely expect Musk understands precisely the state of the development organization at Tesla.
As I previously said, that would be a whole nother kettle of fish if he were not factually correct.. I am not qualified to speak authoritatively on NNs or AI beyond where I have. I, further, am not qualified to speak on what's going on in the AI group at Tesla. I do trust Musk to be in a position to speak authoritatively on that specific subject which he has repeated in several venues. There is a consistency in what he is saying.

I am repeating myself as are others, which is why I unwatched this thread. I'll say this, since they are 'shipping' beta code, in 3 months we'll all have much greater clarity into what we can expect in the subsequent 3 months.

 

[electronblue]

First of all, to @electronblue if by "appeal to authority" you were implying me.

I was not implying anyone in this thread really, just responding to your musings about whom to believe and why. The logical fallacy of appeal to authority was just a counter example of the need to be vigilant with any source.

For example simply because Musk says something about topic X, and Musk knows a lot/more about topic X, does not mean Musk is right about topic X. He may be, but logically and realistically it does not automatically follow.

 

[rnortman]

@electronblue I admire your patience in continuing to participate in these sorts of discussions.

 

[electronblue]

Because by your interpretation, "feature complete L5" requires hundreds of features that Tesla has not done yet.

Bingo.

Level 5 no geofence feature complete is easy to say but hard to do.

Elon talks about L5 because he believes that FSD will be L5. The ODD will be L5 and he believes that with all the features that Tesla is working on, that our cars will eventually be able to completely drive themselves on any road exactly like the SAE L5 definition that I quoted says.

So I guess when Tesla releases "feature complete L5" at the end of the year, our cars will be L5 even with a safety driver? I can't wait to drive my L5 Tesla. LOL.

Your mental gymnastics in interpreting Musk in a way that fits your narrative are fairly amazing, even discussing ODD for the one SAE Level that does not even really have one.

If Tesla is (neither I or Musk ever claimed release) Level 5 feature complete at the end of the year, the car certainly will be Level 5 prototype if Tesla also declares that as its design intent.

If Musk meant Tesla won´t be Level 5 no geofence feature complete by end of 2019, he simply should not have made the claim it will be.

 

[electronblue]

The problem with those companies is we don't know how fast they can expand to other cities.

I think the problem statement is similar.

Waymo can do L4 in one suburb. Tesla can do L2+ in US. When will they do L4 in top 200 US metros ?

That is why I said Level 4. It has an ODD that can be limited. Level 5 mostly does not.

As for Waymo, certainly they operate Level 4 prototypes in more locales than that.

But I agree, I make no claims about who gets to production Level 5 first. I don´t think anyone knows really.

 

[EVNow]

That is why I said Level 4. It has an ODD that can be limited. Level 5 mostly does not.

As for Waymo, certainly they operate Level 4 prototypes in more locales than that.

But I agree, I make no claims about who gets to production Level 5 first. I don´t think anyone knows really.

L5 is not same as scaling up L4 to more than a very few select locations. When will Waymo cover my city - that could be L4, doesn't have to be L5.

 

[EVNow]

If Musk meant Tesla won´t be Level 5 no geofence feature complete by end of 2019, he simply should not have made the claim it will be.

He didn't make that claim. Its your interpretation.

He said Yes to a two part question. I've no doubt he was concentrating on geofencing and not L5.

Or it could be that basic freeway NOA + City NOA = L5 - with a lot of "edge" cases still to be handled. You are way over interpreting a general directional and imprecise answer.

It can happen even when you are auditioning for the most important job in the world.

Kamala Harris Backtracks on Support for Eliminating Private Insurance, Says She Heard Debate Question Wrong

 

[daniel]

All the talk about "feature complete" and Elon's latest tweets on the subject is kind of obscuring the fact that when I bought my Model 3, Tesla was saying that pretty soon if I paid for the FSD package, the car would be fully capable of driving itself pretty much anywhere without a driver, picking up kids from school or operating as a robotaxi, etc. They specified that regulatory approval was something they had no control over, but they said the car would be capable. I didn't pay for FSD because I didn't believe this would happen "pretty soon" or while I expected to own the car, and I didn't believe it could be done with the present sensors.

As we argue over what "feature complete" means, or what Elon intended in his tweets, lets not forget that anything less than "your car will drive itself with nobody in it" is a backtrack from the promise Tesla made until rather recently to people who paid an extra $5,000 (or whatever the amount was) for "FSD." And that buyers are within their rights to assume this means during the time frame that people normally keep a car.

And as always, I love my car and would hate to ever be without EAP again and would not sell it even for the price I paid when it was new. But I think that Tesla's promise is years further away than Tesla and Elon told us.

 

[EVNow]

And as always, I love my car and would hate to ever be without EAP again and would not sell it even for the price I paid when it was new. But I think that Tesla's promise is years further away than Tesla and Elon told us.

True - but that is not the topic of this thread.

 

[daniel]

True - but that is not the topic of this thread.

The topic of this thread is arguing about the definition of "feature complete" in Elon's recent tweet or interview comment, which people are interpreting to say that he's not promising that we'll have self-driving cars by the end of 2020, which, if that's the case, seems like an attempt to back out of his earlier more explicit promise that we'd be able to send our cars off without a driver in a couple of years from back then.

In all the arguing about the term "feature complete" people are forgetting that a couple of years ago Elon's promise was not couched in vague or technical language at all. Back then it wasn't about milestones in software development; it was about what we would actually be doing with our cars.

Nobody but a total software nerd or an employee in Tesla's software department cares about when the software will be "feature complete" in the narrow technical sense. We care about when self-driving without a human in the car will be ready to apply for regulatory approval for sale to the public. This thread is all about why Musk was not promising that by the end of 2020, when he made a comment that anyone who is not a software nerd would have thought meant he was promising just that.

In that context, the question of whether we'll actually have a million robo-taxis operating on U.S. streets, is indeed relevant to the topic.

 

[daniel]

True - but that is not the topic of this thread.

P.S., here's the OP:

He's at it again, this time on the ARK podcast: On the Road to Full Autonomy With Elon Musk — FYI Podcast

"I think we will be feature complete FSD this year, meaning the car will be able to find you in a parking lot, pick you up and take you all the way to your destination this year. I would say I'm certain of that. That is not a question mark."

He did then caveat that it would not 'work perfectly' and require observation. Even then - that would be an incredible accomplishment given where Tesla AP is today. They're either sitting on some incredible progress with HW3 or he is way off the mark - again.

The quibbling over what the term "feature complete" actually means comes later. In the quote Elon defines "feature complete" as meaning that "the car will be able to find you in a parking lot, pick you up and take you all the way to your destination this year." And he says he's certain of that. Only later he says it "may require observation."

The question of whether or not they can accomplish this is entirely on topic. Even if the thread has drifted off onto trying to explain that he didn't really mean the car would be able to do that because "feature complete" doesn't really mean what Elon said it did in the quote.

 

[EVNow]

This thread is all about why Musk was not promising that by the end of 2020, when he made a comment that anyone who is not a software nerd would have thought meant he was promising just that.

Well, just because someone who is not in the industry thinks 'x' means 'y' - that doesn't make 'y' the correct thing. If the Fed announces the funds rate as zero percent, a person who doesn't understand economics says the local bank should lend him money at zero percent, they are not going to do it.

At the same time, definitely Musk needs a lot of improving his communication skills. But then, Musk sold a lot of Tesla's without ads and mostly by getting a lot of press with his personality. We have to take the good with the bad.

 

[banned-66611]

Yes - the question is how many of these sub-features are "edge cases" according to Tesla. This is what I mean when I say the feature set is nebulous. I will post the query to Musk on twitter on the off chance he responds.

Let's face it, it's going to be jank. Tesla always release these things as beta before they are ready. It's just a question of how bad the failure modes are going to be.

Crash into a lane divider at 70 MPH? Stop in the middle of the highway? Get trapped in a car park? It could range from fatal to hilarious.

 

[rnortman]

That is why I said Level 4. It has an ODD that can be limited. Level 5 mostly does not.

As I like to say, L5 includes driving off-road (or through the interior of an abandoned shopping mall) while being pursued by a zombie horde with your passengers hanging out the windows firing shotguns to clear your path. If a human can in principle do it, an L5 car can do it.

(This is an exaggeration of course, but the concept of an ODD for an L5 car is ill-defined at best.)

This is why no serious contenders are aiming for true L5 in the next decade (give or take), only L4 with an expanding but still well-defined ODD.

 

[EVNow]

Let's face it, it's going to be jank. Tesla always release these things as beta before they are ready. It's just a question of how bad the failure modes are going to be.

Crash into a lane divider at 70 MPH? Stop in the middle of the highway? Get trapped in a car park? It could range from fatal to hilarious.

Lets face it, you are the glass is half empty kind

 

[EVNow]

As I like to say, L5 includes driving off-road (or through the interior of an abandoned shopping mall) while being pursued by a zombie horde with your passengers hanging out the windows firing shotguns to clear your path. If a human can in principle do it, an L5 car can do it.

Well, you don't have to go to zombie apocalypse. There will be a number of common scenarios in urban Asian roads that Waymo won't be able to handle in a long time.

Basically, L5 is an ill defined term. We should really quit talking about L5 and concentrate on getting to L4 that can work for 99% of the US population first. That is what I'm looking forward to, not some theoretical illusory L5.

So, the real practical question is :

Will Tesla get to L4 in top 200 US metros first or will Waymo or someone else.

L5 or for that matter geofencing etc are of limited use. Who cares if Waymo is geofenced, as long as that fence covers 99% of population ? Musk's only claim is the fastest way to get to this state is by developing a "non geofenced" solution from the beginning. We have to see if that is true or whether Waymo can actually scale up their geofenced solution to cover most of US.

 

[diplomat33]

Well, you don't have to go to zombie apocalypse. There will be a number of common scenarios in urban Asian roads that Waymo won't be able to handle in a long time.

Basically, L5 is an ill defined term. We should really quit talking about L5 and concentrate on getting to L4 that can work for 99% of the US population first. That is what I'm looking forward to, not some theoretical illusory L5.

I agree that L4 is a more appropriate and realistic autonomy level.

I think the reason Elon uses L5 is because he thinks it fits better with what he wants FSD to be. He intends FSD to be "self-driving everywhere with no geofencing". So L5 would seem to fit that better than L4. Plus, he probably thinks that if he says L4 that people will assume that FSD won't work in certain cities and that would turn off customers. It is also worth noting that contrary to Waymo which focuses on robotaxis in certain locations, Tesla is aiming to sell cars with FSD to customers across the US and even around the world. So Tesla wants to be able promote FSD that will work everywhere, hence the L5 label.

So, the real practical question is:

Will Tesla get to L4 in top 200 US metros first or will Waymo or someone else.

L5 or for that matter geofencing etc are of limited use. Who cares if Waymo is geofenced, as long as that fence covers 99% of population ? Musk's only claim is the fastest way to get to this state is by developing a "non geofenced" solution from the beginning. We have to see if that is true or whether Waymo can actually scale up their geofenced solution to cover most of US.

While Waymo is definitely ahead in terms of pure self-driving capability and technology, I think Tesla's general solution does give them an advantage in getting to L4 in the top 200 US metros. Waymo has L4 in certain cities but there are not even present in other cities. They have to deploy and test vehicles in each new city which takes time. Tesla may only have say 50% FSD but they have it in every city already, anywhere where Tesla cars are on the road. And as Tesla trains the neural net and improves FSD, and sends out an OTA update, their FSD will improve everywhere. Tesla's approach is like the rising sea levels. It will rise everywhere at the same time. FSD may be a bit better in some cities than others but overall, it will work generally everywhere. So maybe one city, FSD is only 99.99% and another city, it's 99.9999%. But as Tesla improves the neural net further, every city will improve at the same time. So when Tesla gets to true L4, they will get to true L4 in the top 200 US metros at the same time.

 

[EVNow]

Waymo has L4 in certain cities but there are not even present in other cities. They have to deploy and test vehicles in each new city which takes time.

But, they could train 20 cities at a time - and if it only takes 2 months per city, they can cover top 200 cities in a year. Afterall the differences between 2 US cities isn't that much (compared to say driving in Asia).

BTW, all the noise that Tesla is making in terms of FSD - is sure to put pressure on Waymo & Cruise to expand faster. They probably have better idea on exactly where Tesla is compared to publicly available information.