Did Tesla lose focus by making the Model 3 an autonomous car instead of a great EV?

[shrspeedblade]

Some people have a very different definition of a "driver's car" than I do

My version:
-nimble handling and quick, predictable weight transfer
-strong, easily modulated brakes
-quick, accurate steering with good feedback
-a broad, robust powerband that quickens the pulse when you nail it
-good grip but progessive in its nature

Other version:
-steering wheel gauges on a car that has no tach, has no temp gauge, and uses no gas

Which is more a "driver's car" to you?

(FYI if completely my choice, I'd have a HUD 1st or a Model S layout 2nd, but I'm fine with neither, I'm not fine without the "my version" list).

 

[Wolverinegeoff]

I think people need to understand that AP2 is not version 1 of FSD. The functionality that you see in AP2 is not the same as what will be shown by EOY when they attempt a drive across the country. They share the same hardware and they share the machine learning from from AP2 and Shadow mode, but they are currently independent solutions. In part, this is because the relationship with Mobileye soured so quickly and they refused to allow parallel development once they figured out Tesla was working on FSD. That combined with the Mobileye solution failing to see a big rig passing infront of its sensor, resulting in a drivers death, destroying what was left of the relationship.

My understanding, strictly from rumor and speculation, is that there is testing and driving being done with full FSD and leveraging all the hardware, not just 2 cameras and the radar. We have seen the first video of this last year. To think that FSD has not progressed since then is crazy. The only thing we know is that they are not testing in California as the Video was the only time they drove the FSD Model X in CA.

I think people will be shocked to see what they come out before the end of the year. Elon is more then confident that this development is farther along then we understand. I know he is prone to hubris and over promising, but he seems to be learning as he goes.

Seems like they probably have 2 teams, one just trying to get the AP2.0 up to AP1.0 snuff, and another one working on fully integrating all the sensors, UI, FSD, etc. They can quickly fold in the coding from the first team into the 2nd I'd hope!

 

[Reciprocity]

Seems like they probably have 2 teams, one just trying to get the AP2.0 up to AP1.0 snuff, and another one working on fully integrating all the sensors, UI, FSD, etc. They can quickly fold in the coding from the first team into the 2nd I'd hope!

Bingo. Its not the best way to do it, but they had no choice. About a month ago Elon stated that AP2 would surpass AP1 and start to head towards FSD in 3-6 months. Once AP2 surpasses AP1 in terms of features and accuracy, then that team will move to FSD and AP2 for all intense and purposes will only improve as the system learns from being used. At some point thereafter, all 8 cameras will be enabled and FSD 1.0 will take over for AP2 and you will start to see new features that are more FSD like, but still not self driving. This is more then likely the timeline that Elon alludes to with 3-6 months, now 2-5 months. This system will be capable of a cross country trip by the end of the year. The limiting factor is probably not even the devs, but the data. They need enough machine learning in different situations to compensate for a Level 3+ autonomous trip from LA to NY.

This is of course a highly simplified description of what might happen, but going from TACC to Level 4+ Autonomous is very complicated.

 

[ikjadoon]

As someone whose first car was a Hyundai Excel, I'm reflecting on how incredibly far we've come in the 24 years I've had a license, and I'm just flabergasted at the "now now now...me me me....must must must" attitude so many on here are displaying.

When Model 3's are delivered, I fully expect Tesla owners to have an even worse image in the eyes of the general public because of attitudes like this.

We're all going right into the pile w/BMW and Benz drivers........

Back in my day, you didn't open your mouth as a billion-dollar CEO making promises you couldn't keep. Nobody--especially most Tesla customers--ever claimed full autonomy was coming in 2017. Just Elon.

Thus, re-direct your nostalgia to Elon's mouth; that's where it needs to go.

 

[ModelNforNerd]

Back in my day, you didn't open your mouth as a billion-dollar CEO making promises you couldn't keep. Nobody--especially most Tesla customers--ever claimed full autonomy was coming in 2017. Just Elon.

Thus, re-direct your nostalgia to Elon's mouth; that's where it needs to go.

Speaking of his mouth.....you're forgetting the disclaimers that come out of it. He covered himself by reminding us that FSD was dependent on jurisdictional regulatory approvals.

I'll let you get back to your fascination with Elon's mouth now.

 

[N5329K]

In the past, decisions made by Tesla engineers would have to percolate through the RDF: Mr Musk's Reality Distortion Field. Sometimes they didn't survive the encounter and new goals, features, approaches, even product dimensions would have to be engineered. Sometimes for the better, sometimes not, and not always for the most compelling of reasons (the length of a particular lady's skirt once played a prominent role in altering Roadster sill height...a big deal once "pencils were down").
So what seems to be fixed and what will ultimately show up are not necessarily the same thing, even now.
Robin

 

[Garlan Garner]

I'm thinking that there might be a lot that Elon hasn't said yet. Something beyond HUD. Hmmm

I certainly wouldn't mind. I think it would be just great.......

OH NO! I almost forgot before all of you remind me......nothing is going in the Model 3 that isn't in the MX or MS.....lol
Elon Musk talks Tesla Model 3 display again, shows confidence in ‘Tesla Vision’ or ‘spaceship-like steering controls’?

 

[Garlan Garner]

3 more days until my $1K is a year old. Well At least I got a webpage with a number on it so far.

Just kidding....keep the responses down.

 

[ikjadoon]

Speaking of his mouth.....you're forgetting the disclaimers that come out of it. He covered himself by reminding us that FSD was dependent on jurisdictional regulatory approvals.

I'll let you get back to your fascination with Elon's mouth now.

That's putting the cart before the horse. I think, before any regulator is going to be interested, current AP2 owners should like to see something.

Tesla claimed a "demonstration drive", too. I think we'd all be happy with that; the regulatory approval should and must come after FSD actually works.

 

[ModelNforNerd]

That's putting the cart before the horse. I think, before any regulator is going to be interested, current AP2 owners should like to see something.

Tesla claimed a "demonstration drive", too. I think we'd all be happy with that; the regulatory approval should and must come after FSD actually works.

the car isn't going to be in the hands of a majority of people who've reserved it for at least a year.
cart before the horse, indeed...........

 

[ikjadoon]

the car isn't going to be in the hands of a majority of people who've reserved it for at least a year.
cart before the horse, indeed...........

Obviously. The timeline of deliveries shouldn't have anything to do with Elon's promises for the car. With your logic, he might as well leave out the rear seats in the 2017 deliveries. "Oh, it won't affect a lot of deliveries. Don't worry; the majority of you will get rear seats." If he makes a statement, he needs to stand by it. Simple stuff. No need to repeatedly apologize for him. He clearly is the only person in the world making "insider" statements about the Model 3.

To the original point (which has gotten lost in your shifting of the goalposts): hopefully, if he thinks we don't need a binnacle at all, he is very confident about AP2 being used frequently and for long durations in the Model 3. Thus, the steering wheel should have limited importance soon, too.

 

[Bladerskb]

No. Elon promised a fully autonomous drive from LA to NYC this year. They've got eight months to go. Let's get to it. The goal is not "someday"--come on. If EM's going to make promises on his product's abilities, he needs to deliver.

They should design cars for the functions they can perform. AP2 development is troublingly behind; tomorrow or Wednesday's final 8.1 release better be pretty darn spectacular in matching AP1.

-----

It all has to work out right. The very first Model 3 delivered to a normal customer (not a Tesla employee) should be incredibly close to performing a fully autonomous drive.

I think people need to understand that AP2 is not version 1 of FSD. The functionality that you see in AP2 is not the same as what will be shown by EOY when they attempt a drive across the country. They share the same hardware and they share the machine learning from from AP2 and Shadow mode, but they are currently independent solutions. In part, this is because the relationship with Mobileye soured so quickly and they refused to allow parallel development once they figured out Tesla was working on FSD. That combined with the Mobileye solution failing to see a big rig passing infront of its sensor, resulting in a drivers death, destroying what was left of the relationship.

My understanding, strictly from rumor and speculation, is that there is testing and driving being done with full FSD and leveraging all the hardware, not just 2 cameras and the radar. We have seen the first video of this last year. To think that FSD has not progressed since then is crazy. The only thing we know is that they are not testing in California as the Video was the only time they drove the FSD Model X in CA.

I think people will be shocked to see what they come out before the end of the year. Elon is more then confident that this development is farther along then we understand. I know he is prone to hubris and over promising, but he seems to be learning as he goes.

Shocked??
incredibly close to performing a fully autonomous drive??

You two would be incredibly naive if you actually believe that. First of all, a drive from Tesla HQ to Time Square has just 2 miles of surface street so that trip elon has been hyping not only has been done multiply times before but its also completely worthless as a fully self driving indicator. 0.0006% of the drive only matters to self driving. That's 2 miles out of 2,943.

Secondly, Google mastered highway driving back in 2010 before moving onto city street driving and completed multiple 1,000 mile drives without human intervention. Google also
Eartheasy Blog » Google hits the road with a self-driving Prius

Audi in 2015 let a journalist sit behind the wheel of its highway L4 car as it drove 500 miles without a single touch. (Palo Alto to Las Vegas for CES)
I Rode 500 Miles in a Self-Driving Car and Saw the Future. It’s Delightfully Dull

Delphi actually completed a cross country 3,400 miles trip, SF to NY with 99% on automated mode
This Is Big: A Robo-Car Just Drove Across the Country

and before you claim they are using non production ready bulky sensors.
Here are the sensors on the audi and delphi cars.

So no your model 3 will not be anywhere close to being self driving just because elon manages to release PR on new years eve about a cross country drive that's already been done to death. Which is actually what they are claiming EAP is supposed to do. So what we should be saying is that we hope tesla can actually deliver on EAP before the end of 2017


AUDI

DELPHI

 

[ikjadoon]

Shocked??
incredibly close to performing a fully autonomous drive??

You two would be incredibly naive if you actually believe that.

You've forgotten a few points that make it somewhat more interesting:

>According to Musk, the trip would occur "without the need for a single touch" from the driver, including recharging the car’s battery.

What car is automatically getting gasoline from gas stations or charging the battery autonomously? Not to mention, I've yet to see one with 100% 'no driver needed'. Even that last one by Delphi: they avoided night driving and it was taken over still 1% of the time.

However, thank you for all the links. That was very informative.

 

[CAVU]

I've operated in the automation business for a long time. The reason the autopilot won't work for the M3 is because not every else is doing it. Automation works in the air with commercial planes because on the jet routes and the approaches everyone is doing the same thing. There are no stop signs or red lights or people texting. The commercial routes are predictable and predetermined, programmed into the flight computer before the engines are started.

 

[Bladerskb]

You've forgotten a few points that make it somewhat more interesting:

>According to Musk, the trip would occur "without the need for a single touch" from the driver, including recharging the car’s battery.

What car is automatically getting gasoline from gas stations or charging the battery autonomously? Not to mention, I've yet to see one with 100% 'no driver needed'. Even that last one by Delphi: they avoided night driving and it was taken over still 1% of the time.

However, thank you for all the links. That was very informative.

gasoline or electric has nothing to do with anything. it is mean-less when it comes to FSD tech.
Secondly, It was taken over when it was time to drive into city streets, hotels, grocery store, photo-ops, etc

Tesla will likewise do the same thing. They just won't tell you cause they will need multiple back up drivers (taking turns) and they would need to sleep, eat, drink, use bathroom, etc.

it won't be end to end.

They avoided night driving not because they couldn't do it, but because they wanted to collect data and you collect useful data in the day not at night..

Not to mention, I've yet to see one with 100% 'no driver needed'.

Which part of going cross country is mean-less because you are only driving 2 miles on surface street don't you understand?
Google has done 1,000 mile trips (with no touch), Audi has done it. Even delphi said it was meanless other than for collecting data.
Driving 2 miles on surface street out of 3,000 miles is utterly mean-less.
The point is everyone can do automated L3 Highway. Doing cross-country just to do it as a demo is a total waste of time.

 

[Skotty]

In my honest opinion, I think they should just ignore surface street autonomous drive completely until they achieve full no-driver-supervision-required autonomous drive on divided highways without stop signals. That would make the challenge a lot easier (much simpler rules and far fewer scenarios) and, once achieved, it would eliminate the worst part of driving cars.

I'm perhaps biased here. I don't mind driving city streets at all, but 95% of my driving right now is interstate, and it sucks major you know what. Would love to just set it on auto and watch a movie. Even with the best AP1, you still had to supervise, and I don't consider it a win until you can completely do something else.

This could be ideal for truck drivers too. They would still be needed for the city, refueling, and such, but could play with their packages (from the trailer of course) or sleep while rumbling down the interstate.

Trickiest part would be the transition from autonomous back to human driver at the exit points. Would need to be able to pull over if the driver doesn't respond to take over. Even that doesn't sound too bad. It could ask the driver to take over about a mile before exit point, and if they don't respond, pull over on shoulder. Avoids having to deal with the exit ramp, which could be more complicated.

Heck, I'd even be happy if I had to deal with all the intermediate interchanges between interstates or in downtown areas so long as it could do the snoozer 2 to 4 hour straight fall asleep at the wheel stretches.

 

[somnambule]

I can't believe how many people seem to want a HUD. Do you folks not wear sunglasses when you drive during the day? A HUD would be completely useless during the daytime (well, unless you use cheapo non-polarizing sunglasses).

God - the Phaeton!! The W140 S Class!!! The E39 5 series! The E36 3 series! The world will not see their likes again. But their time has passed and the nexus of what matters in a car has inexorably shifted.

IMO, this is the most profound statement in this thread. Automakers have been sweating that young people are losing interest in cars for quite some time now. Tesla is making a bet here that may alienate some of the "old school" buyers but I think keeping the big touch screen and dropping the instrument pod is the right call (assuming something had to go). If Tesla can get AP2 sorted out, between that and the UI a large display allows, they'll gain far more than they lose.

And if anyone has delusions about how much the car-buying public cares about driving for fun, just look at the % of vehicles that are FWD vs. RWD (and I'm willing to bet that the majority of RWD buyers do so on brand rather than handling/feel).

As for those of you alarmed about the early Model 3s not having dual-motor etc., keep in mind the Model S didn't even have parking sensors at launch.

 

[Qball]

I say YES. I am severely disappointed.

 

[ikjadoon]

gasoline or electric has nothing to do with anything. it is mean-less when it comes to FSD tech.
Secondly, It was taken over when it was time to drive into city streets, hotels, grocery store, photo-ops, etc

Tesla will likewise do the same thing. They just won't tell you cause they will need multiple back up drivers (taking turns) and they would need to sleep, eat, drink, use bathroom, etc.

it won't be end to end.

They avoided night driving not because they couldn't do it, but because they wanted to collect data and you collect useful data in the day not at night..



Which part of going cross country is mean-less because you are only driving 2 miles on surface street don't you understand?
Google has done 1,000 mile trips (with no touch), Audi has done it. Even delphi said it was meanless other than for collecting data.
Driving 2 miles on surface street out of 3,000 miles is utterly mean-less.
The point is everyone can do automated L3 Highway. Doing cross-country just to do it as a demo is a total waste of time.

Exactly. Elon's claims are 1) end-to-end, 2) with charging, and 3) "without the need of a single touch". What part of these claims you don't understand? Can you understand that? Charging? Not a single touch? Full autonomy? Re-read it. Slowly. Carefully. Now read it again one more time for good measure.

Obviously gasoline cars won't be doing this; I'd like to see a gasoline car pump its own gas.

They can't collect useful data at night? Maybe they just don't have the ability to capture & analyze that data; obviously, there is data at night that is unique for the sensors.

2 miles of surface street: sure. Obviously, we all want to see more than that, but I'll take self-charging and "not a single touch".

A waste of time? You have no idea of the public's perception of self-driving if you think demonstrations are still a waste of time in 2017.

I can't believe how many people seem to want a HUD. Do you folks not wear sunglasses when you drive during the day? A HUD would be completely useless during the daytime (well, unless you use cheapo non-polarizing sunglasses).

I can't believe someone at TMC doesn't follow ICE car news. Do you folks not Google simple questions before you post?

Head's up! Ford's Lincoln unit improves head-up displays

HUD w/ polarized lenses was figured out a while ago by Texas Instruments...and, look here, Lincoln is using that very technology.

 

[Reciprocity]

Secondly, Google mastered highway driving back in 2010 before moving onto city street driving and completed multiple 1,000 mile drives without human intervention. Google also
Eartheasy Blog » Google hits the road with a self-driving Prius

Audi in 2015 let a journalist sit behind the wheel of its highway L4 car as it drove 500 miles without a single touch. (Palo Alto to Las Vegas for CES)
I Rode 500 Miles in a Self-Driving Car and Saw the Future. It’s Delightfully Dull

Delphi actually completed a cross country 3,400 miles trip, SF to NY with 99% on automated mode
This Is Big: A Robo-Car Just Drove Across the Country

You know what none of those companies have. Cars in consumers hands with the same hardware they are testing. Even AP1 could do a large amount of what you are showing as proof that Tesla doesnt have anything.

You can say there is no value to LA to NY, but it will get a lot of press and sell a lot of cars.

All I am saying is that there is a lot of dev and testing going on that they are not making public. Enough to have Elon saying that he considers full autonomy solved. By which I believe he means that all they are waiting for is enough data to solve all the corner cases. When compared to Google, which by law has to make it public because they are testing in CA. You can also tell when Google or any of those other companies are testing, because they cars look like they came out of a Madmax movie with all the crap bolted to them. Any Model X you see outside of CA could be testing level 3-4 autonomous driving.