Tesla autopilot HW3

[diplomat33]

Finally, Tesla was marketing coast to coast FSD ride by the end of 2017. Back then we had no real information on how Tesla’s AP2 program was faring internally. We had the October 2016 video that was by that day’s standards no better or worse than the demos of April 22nd. Did we really have a realistic way of knowing nothing would be out by May 2019?

In retrospect yes. The key is that we had no real info on Tesla's AP2 internal program as you said. And we could see the sad state of AP2. In fact, a lot of you were arguing that there was no way FSD was just months away given the poor state of AP2. When AP2 was still struggling to get to parity with AP1 and Navigate on AP was a year away, that was a clear clue that FSD was a long ways off.

But this is why I think things are different now. Before we just had marketing hype but AP2 was poor. Now, We have good AP2. We have good NOA now. We have traffic light detection running in shadow mode now. And we have AP3 now in some cars. And we had an Autonomy Investor Day event that gave us specifics about Tesla's FSD development as well as test drives that were positive about Tesla's alpha FSD. So we have many concrete things now that indicate that we are closer to FSD.

And I am on record many times on this forum, saying that I think it was a huge mistake how Tesla and Elon handled the marketing of FSD. They should have just marketed the extra cameras of AP2 as just "adding hardware for future autopilot improvements" and not mentioned FSD at all. And, they should have waited until they were ready to do an Autonomy Investor Day before marketing or selling anything related to FSD.

 

[electronblue]

In retrospect yes. The key is that we had no real info on Tesla's AP2 internal program as you said. And we could see the sad state of AP2. In fact, a lot of you were arguing that there was no way FSD was just months away given the poor state of AP2. When AP2 was still struggling to get to parity with AP1 and Navigate on AP was a year away, that was a clear clue that FSD was a long ways off.

But this is why I think things are different now. Before we just had marketing hype but AP2 was poor. Now, We have good AP2. We have good NOA now. We have traffic light detection running in shadow mode now. And we have AP3 now in some cars. And we had an Autonomy Investor Day event that gave us specifics about Tesla's FSD development as well as test drives that were positive about Tesla's alpha FSD. So we have many concrete things now that indicate that we are closer to FSD.

And I am on record many times on this forum, saying that I think it was a huge mistake how Tesla and Elon handled the marketing of FSD. They should have just marketed the extra cameras of AP2 as just "adding hardware for future autopilot improvements" and not mentioned FSD at all. And, they should have waited until they were ready to do an Autonomy Investor Day before marketing or selling anything related to FSD.

Just to correct one thing: Tesla launched Autopilot 2 on October 19th with an Investor Conference Call where they called if ”Level 5 capable hardware” and followed up with the infamous October 2016 video. We did not see how AP2 fared until late December 2016. Many of us could not see the state of AP2 when we were ordering those AP2 cars. I certainly did not.

I agree with you it was a huge mistake. Well, not mistake, obviously it was intentional on their part.

 

[chillaban]

Just to correct one thing: Tesla launched Autopilot 2 on October 19th with an Investor Conference Call where they called if ”Level 5 capable hardware” and followed up with the infamous October 2016 video. We did not see how AP2 fared until late December 2016. Many of us could not see the state of AP2 when we were ordering those AP2 cars. I certainly did not.

I agree with you it was a huge mistake. Well, not mistake, obviously it was intentional on their part.

Even “intentional” is hard to define. I wouldn’t put it past Elon to read a DRIVE PX brochure and decide that their hardware and SDK would just magically give you AP1 functionality in a matter of a month or two. Seems like textbook Elon to gloss over real world complexity and consider things easy to do

 

[CK_Stuggi]

As far as we know today, IIRC, they were planning to use Mobileyes software on AP2. When Mobileye (naturally) said no, they had to start from scratch. While what they did since Karpathy came on board is certainly impressive, it's nowhere near where it should be given the claims they've made.
And until they have speed sign recognition, no matter how much better AP2 holds a lane, they still don't have AP1 parity. Plus, in Germany, model 3 has NoA, S and X don't. There's the progress for you.

Another aspect; they only showed FSD progress to investors, so to people who have a financial interest in FSD being a success story. Why wasn't anyone allowed to show anything or say anything substantial beyond advanced summon and NoA?

 

[Snuffysasa]

Basically making any 90 degree turn like at intersections but also auto lane changes without confirmation on local roads to follow navigation. Like we saw a bit of in the FSD demo, Tesla is planning to use NOA on city streets too where we get the 1 blue line on city streets and the car follows your navigation direction, making turns to stay on the right road.

Yea I understand all of that.

But I'm not sure that 90 degree turns would be included in "auto driving on city streets"... they could have NOA, but when there is a turn, it would just say unsupported maneuver like it does on highway.

Also, I'm not sure why you have added concern for lane changes on local roads any more than lane changes on highways.

 

[electronblue]

As far as we know today, IIRC, they were planning to use Mobileyes software on AP2. When Mobileye (naturally) said no, they had to start from scratch.

To be clear though Tesla knew all that in October 2016 when they announced AP2. The MobilEye break-up had happened and gone public already over the summer.

 

[diplomat33]

Yea I understand all of that.

But I'm not sure that 90 degree turns would be included in "auto driving on city streets"... they could have NOA, but when there is a turn, it would just say unsupported maneuver like it does on highway.

Because 90 degree turns at intersection is a maneuver that would be required for "automatic city driving". How else are you going to do self-driving A to B if the car won't turn at intersections?

Also, I'm not sure why you have added concern for lane changes on local roads any more than lane changes on highways.

Right now, AP does not do auto lane changes without confirmation on non-highways. Obviously, it is something that would need to be implemented, through NOA on city streets, in order to fulfill "automatic city driving".

 

[tufa4311]

We are kinda in uncharted waters, I will admit, since nobody knows exactly how Tesla will roll out these features. However, Tesla did separate the two features on the website. So my take is that the stop and go at traffic lights feature will work like Autopilot works now on city streets except that AP will be able to also stop on its own at a red light. So yes, as you said, it will just go down one street, stopping and starting but not make turns. At first. This will give us the advantage of being able to use AP on city streets with a bit more convenience, that we won't need to disengage when coming up on a red light. Then later, Tesla will add the ability to also make turns at intersections and change lanes to get to a destination as part of the "automatic city driving" feature. That's how envision it working.

Silly me, I guess I just need to wait another 24 months or so and will be just fine with the FSD seeing as there won't even be a steering wheel...

Tesla unveils car design without a steering wheel, coming within 2 years, says Elon Musk

 

[diplomat33]

Silly me, I guess I just need to wait another 24 months or so and will be just fine with the FSD seeing as there won't even be a steering wheel...

Tesla unveils car design without a steering wheel, coming within 2 years, says Elon Musk

According to Elon, that is the plan, yes.

 

[rnortman]

According to Elon, that is the plan, yes.

And this doesn't make you slightly uncomfortable with believing him? With using his tweets and this "investor autonomy day" as evidence on which to base your optimism? This should cause a little cognitive dissonance I think.

 

[diplomat33]

And this doesn't make you slightly uncomfortable with believing him? With using his tweets and this "investor autonomy day" as evidence on which to base your optimism? This should cause a little cognitive dissonance I think.

Sure, I am uncomfortable with Elon promising "no steering wheel" in 2 years although there is a part of me that really admires his boldness.

I am optimistic though about Tesla's FSD in general. I am just skeptical that Tesla will actually achieve "no steering wheel" level in just 2 years. I think it will take them longer than 2 years.

Frankly, as a Tesla owner, I am not personally interested in a Model 3 with no steering wheel. I just love getting new autopilot features that make my car more self-driving over time. That is of practical value to me.

 

[turnem]

Sure, I am uncomfortable with Elon promising "no steering wheel" in 2 years although there is a part of me that really admires his boldness.

I am optimistic though about Tesla's FSD in general. I am just skeptical that Tesla will actually achieve "no steering wheel" level in just 2 years. I think it will take them longer than 2 years.

Frankly, as a Tesla owner, I am not personally interested in a Model 3 with no steering wheel. I just love getting new autopilot features that make my car more self-driving over time. That is of practical value to me.

You've articulated my feelings pretty well here. I think Elon's statements are pretty aggressive but I also think he knows this and he's being very intentional here.

I just like having a Tesla to experience what it's capable of today and seeing how it evolves over time.

 

[tufa4311]

You've articulated my feelings pretty well here. I think Elon's statements are pretty aggressive but I also think he knows this and he's being very intentional here.

I just like having a Tesla to experience what it's capable of today and seeing how it evolves over time.

I agree they are aggressive. I understand why some don't like how aggressive they are, that many times the goals are missed, and some even feel they are outright lies and he "knows" he won't be able to do it. Like I said, I understand, but...rightly so or not, I am of the school to aim high, and even though you'll likely not hit your target, you'll get higher than if you didn't go for that height to begin with.

For me, FSD goes beyond simple coolness. I have epilepsy, I may have a seizure once or twice a year, and those may even occur while I'm asleep, and I legally have a license as I meet the medical requirements of my state (seizure free for prior 3 months), and I've driven for over 10 years, and I've chosen not to drive for the last 4 years, and, and, and...and I'd like a car that could do much of the driving for me...I'm waiting...

 

[turnem]

I agree they are aggressive. I understand why some don't like how aggressive they are, that many times the goals are missed, and some even feel they are outright lies and he "knows" he won't be able to do it. Like I said, I understand, but...rightly so or not, I am of the school to aim high, and even though you'll likely not hit your target, you'll get higher than if you didn't go for that height to begin with.

For me, FSD goes beyond simple coolness. I have epilepsy, I may have a seizure once or twice a year, and those may even occur while I'm asleep, and I legally have a license as I meet the medical requirements of my state (seizure free for prior 3 months), and I've driven for over 10 years, and I've chosen not to drive for the last 4 years, and, and, and...and I'd like a car that could do much of the driving for me...I'm waiting...

Agree. I think Elon is pushing folks to expand the way they think about autonomous driving.

I have no idea if they'll hit their goals but I hope they do for many reasons. I'd never even really considered your scenario if I'm being honest. I hope for your sake that they can give you some of your independence back.

 

[tufa4311]

Agree. I think Elon is pushing folks to expand the way they think about autonomous driving.

I have no idea if they'll hit their goals but I hope they do for many reasons. I'd never even really considered your scenario if I'm being honest. I hope for your sake that they can give you some of your independence back.

Thanks!

That recent video Musk showed with him zipping the city, not touching anything, is awesome, not sure what Level that would be. I'm looking for highway/city streets car handles most driving activities (stop sign/light as well as stop.start behind other vehicles) but human must be available to take over at any point. Even if it's within a bordered geographic area.

 

[Snuffysasa]

Because 90 degree turns at intersection is a maneuver that would be required for "automatic city driving". How else are you going to do self-driving A to B if the car won't turn at intersections?



Right now, AP does not do auto lane changes without confirmation on non-highways. Obviously, it is something that would need to be implemented, through NOA on city streets, in order to fulfill "automatic city driving".

It doesn't say "all functions of automatic city driving" it doesn't say "feature complete auto city driving"

So I'd bet that it does not include 90 degree turns... or at least that is not on the road map for 2019.


Right now NoA does "Automatic highway driving" .... but it doesn't do several types of double merges, cloverleafs, pulling over for emergency vehicles, filter lights entering highways, and many other things.


And I know that auto lane changes has not been implemented on non- highways..... I was asking... if it was added... why is there any more of a concern of auto lane changes on city streets than highways?

Furthermore, I'm not even convinced that auto lane changes is included in "Automatic driving on city streets". or at least is not on their roadmap for what's coming this year. But it could be I'm not sure.

 

[CK_Stuggi]

According to Elon, that is the plan, yes.

According to Elon at this point in time.
In October 2016, feature parity was to come end of December 2016. Didn't happen.
In March 2017, FSD features were 3 months maybe, 6 months definitely. Didn't happen.
End of 2017, coast to coast was a few months away. Didn't happen.
Now it's feature complete FSD end of 2019.

I've heard "end of next year" one too many times from Elon.

Don't get me wrong. I admire the guy and what he has achieved (and keeps achieving), I simply won't get back on the hype train after I was thrown off it in a costly manner.

 

[electronblue]

@CK_Stuggi

According to Elon at this point in time.
In October 2016, feature parity was to come end of December 2016. Didn't happen.
In March 2017, FSD features were 3 months maybe, 6 months definitely. Didn't happen.
End of 2017, coast to coast was a few months away. Didn't happen.
Now it's feature complete FSD end of 2019.

I've heard "end of next year" one too many times from Elon.

Don't get me wrong. I admire the guy and what he has achieved (and keeps achieving), I simply won't get back on the hype train after I was thrown off it in a costly manner.

Minor corrections, in October 2016 Tesla actually in the Design Studio sold EAP as a single update expected in December 2016, that would include NoA and Smart Summon. In December 2016. That’s the way it was worded in the Design Studio.

Elon’s FSD features in 3 to 6 months was in January 2017, so two months earlier. Didn’t of course happen. Coast to coast was also by end of 2017 originally. It got post-poned a couple of times and then quietly forgotten about as usual.

I agree now it is FSD feature complete end of 2019. Given the fact that Elon has missed so many times with AP2 he must know better by now than to mislead again (or alternatively the lie would be massive).

 

[verygreen]

It got post-poned a couple of times and then quietly forgotten about as usual.

It was not forgotten, it was declared (sour grapes style) as something not interesting because everybody could do/have done curve fitting or some such.

 

[CarlK]

According to Elon at this point in time.
In October 2016, feature parity was to come end of December 2016. Didn't happen.
In March 2017, FSD features were 3 months maybe, 6 months definitely. Didn't happen.
End of 2017, coast to coast was a few months away. Didn't happen.
Now it's feature complete FSD end of 2019.

I've heard "end of next year" one too many times from Elon.

Don't get me wrong. I admire the guy and what he has achieved (and keeps achieving), I simply won't get back on the hype train after I was thrown off it in a costly manner.

Really? Because he missed the (tough) schedule even that he has delivered pretty much everything he promised and LIKELY is the only one who got it right this time around too? What about those Lidar/mapping people if it turned out they indeed have waste their time on what they did?

@Snuffysasa Every action you questioned from 90 degree turn to city driving are already there in the prototype program. There are demo videos and interviews of people who took the test drive. You can search them on the internet. They are all there.

In retrospect yes. The key is that we had no real info on Tesla's AP2 internal program as you said. And we could see the sad state of AP2. In fact, a lot of you were arguing that there was no way FSD was just months away given the poor state of AP2. When AP2 was still struggling to get to parity with AP1 and Navigate on AP was a year away, that was a clear clue that FSD was a long ways off.

But this is why I think things are different now. Before we just had marketing hype but AP2 was poor. Now, We have good AP2. We have good NOA now. We have traffic light detection running in shadow mode now. And we have AP3 now in some cars. And we had an Autonomy Investor Day event that gave us specifics about Tesla's FSD development as well as test drives that were positive about Tesla's alpha FSD. So we have many concrete things now that indicate that we are closer to FSD.

And I am on record many times on this forum, saying that I think it was a huge mistake how Tesla and Elon handled the marketing of FSD. They should have just marketed the extra cameras of AP2 as just "adding hardware for future autopilot improvements" and not mentioned FSD at all. And, they should have waited until they were ready to do an Autonomy Investor Day before marketing or selling anything related to FSD.

I agree that it's better off for Tesla if it did not mention FSD or put it as an option until now. It did not do too much good for Tesla anyway as I doubt there were that many people who have ordered it. But that's just how Elon does things. Those promises will push him and his team to move as fast as they can. End result will still be better even that the process can be a little painful.

If Apple is to do this (they are doing this right?) it will keep it secret until it's done. Or still keep it secret if they could not come put with something. It sure is better for Apple but Is this better for you if you're relying on Apple for this product? i do not think so.