Muted Auto Industry response to Tesla's Full Self Driving move - why?

[calisnow]

So it's been over a month since Tesla started putting in the AP 2.0 hardware. Yet the industry response has been dead silence. Is it possible that Tesla's lead in autonomous driving actually growing vs the industry catching up to them? I am a bit surprised there has been no flurry of marketing announcements by Benz/BMW/Audi in the last month saying they are accelerating their own efforts in autonomy.

It's been two years since AP 1.0 hardware shipped - and the closest competitor is the 2017 Mercedes system. Much ballyhooed by Benz last Spring prior to its release, it got panned and ridiculed by journalists this summer for its poor interface and poor performance in the task of lane keeping - despite superior hardware.
And as far as we know Benz has not announced any improvements to the 2017 cars will be released - even as dealer firmware updates - so as far as we know it will not improve.

Now Tesla leapfrogs itself by releasing a system with more than an order of magnitude more computing power - and Tesla will soon have the world's only fleet generating billions of miles of data on this new multicamera system.

If self driving is a harder problem than we realize - more strange corner cases, etc. - then maybe Tesla is building a moat.

Folks keep wringing their hands over the supposedly impending end of Tesla's competitive advantages in electric drivetrains, battery systems, charging networks and autonomy - yet the lead just seems to keep growing year after year.

One last example is the numerous German cars claiming 300+ miles range by 2018 on vapormobiles - by which time we conceivably could see a 120D with 400 miles plus.

 

[calisnow]

By the end of 2017 the Model S is likely to have:

• 340 miles range
• An upgraded interior which better competes with high end German cars
• Augmented reality windshield or HUD - "spaceship controls" as Elon put it
• Some form of full self driving turned on even if the disclaimer screens insist that you must pay attention

If the industry has no answer to Tesla now they *really* won't have any answers 10 months from now. I can't wait to see what the Model S/X sales charts look like in Q3/4 2017 vs the S/E/5/7/A6/A8 competition.

 

[calisnow]

An article Dec 1 in the Verge talks Delphi/Intel/Mobileye's off-the-shelf self driving kit - which they "hope" to have available for purchase by OEM's sometime in 2019 - and the piece goes on to point out that after the 18-24 months of OEM testing/validation this means 2021.

2021??? That's five years after 2016. For some perspective - five years ago the Model S did not even exist.

Five years from now Tesla will possibly be on their fourth generation hardware release - perhaps two generations beyond AP 2.0 today.

They will possibly have walked off with a huge chunk of the market.

The company with the data set and real-world fleet experience to sell a functioning self driving kit to OEM's in 2019 will not be Delphi - it will be Tesla.

Why Delphi and Mobileye think they have the secret sauce for self-driving cars

 

[182RG]

So it's been over a month since Tesla started putting in the AP 2.0 hardware. Yet the industry response has been dead silence. Is it possible that Tesla's lead in autonomous driving actually growing vs the industry catching up to them? I am a bit surprised there has been no flurry of marketing announcements by Benz/BMW/Audi in the last month saying they are accelerating their own efforts in autonomy.

Why? They are waiting for the software to see if it meets the level of hype around it. Hardware alone is meaningless.

 

[calisnow]

Why? They are waiting for the software to see if it meets the level of hype around it. Hardware alone is meaningless.

The software will meet the hype - it may take 12-24 months to train the networks - but if it gradually, yet continually improves as much as my AP 1.0 car has - it will meet the hype.

AP 1.0 is soberingly great - with a neural network running on hardware processing 12 trillion operations per second the next generation system is going to be insane.

But hey - you're a Cessna driver - pilots are famously skeptical of new technology.

 

[Mo City]

Many in the industry already announced earlier this year they will have full autonamy around 2021. There isn't anything left for them to say at this point, especially when Elon Musk said Tesla expects to have a car drive from LA to NY with no human assistance by the end of 2017.

Perhaps another reason is a survey of autonomy experts from competitors showed none of them believed AP2 hardware could achieve full autonomy. So most of them believe Elon Musk is promising what he cannot deliver. Time will tell.

 

[TLej]

If you don't have anything nice to say, don't say anything at all?

 

[182RG]

The software will meet the hype - it may take 12-24 months to train the networks - but if it gradually, yet continually improves as much as my AP 1.0 car has - it will meet the hype.

Remember too....no other car company has a customer base that is willing to buy and support features that are "in progress". You think MB could ship AP hardware only, with a promise of software delivery?

But hey - you're a Cessna driver - pilots are famously skeptical of new technology.

Love my steam gauges!

 

[calisnow]

Many in the industry already announced earlier this year they will have full autonamy around 2021. There isn't anything left for them to say at this point, especially when Elon Musk said Tesla expects to have a car drive from LA to NY with no human assistance by the end of 2017.

Perhaps another reason is a survey of autonomy experts from competitors showed none of them believed AP2 hardware could achieve full autonomy. So most of them believe Elon Musk is promising what he cannot deliver. Time will tell.

Musk under-delivered on AP 1.0 in some minor ways - the car can't/won't ever be able to safely change lanes without the driver checking behind the car first - it's visually blind to the sides and rear. But he way over delivered in other ways with the same hardware suite - my Tesla steers brakes and accelerates smoother than I am physically capable of doing myself. It is much better at faded lane markings than it used to be. It doesn't freak out if a car cuts it off. It anticipates lane changes with turn signals of cars ahead. If it notices a car ahead moving so much as a couple inches into my lane it will wait to see what that car is doing before it continues passing. It did not do that in October 2015 - the learning is real. It is freaking scary how much it has improved since they turned it on a year ago - and yet this system is working all this magic with 7 fewer cameras and 95% less computing power than the new system.

The "experts" who don't own Teslas with AP 1.0 and don't work for Tesla can say whatever they like but full self driving in at least limited, good weather is inevitable with this new system. I think even journalists who test drive and write about Teslas have less faith because they haven't driven one daily for months and thus been able to observe that the improvements/learning to the system are in fact real.

If you *do* daily drive an Autopilot 1.0 car it is much easier to logically extrapolate from your current experience watching the car get smarter over time that a system built by the same engineers, with the benefit of a lot of the experience of AP 1.0 - yet with 360 vision and a brain that is more than an order of magnitude more powerful - will indeed be soon capable of full self driving in good weather (legally maybe not - physically, yes).

I agree with you that this generation of hardware will not be able to drive in a blizzard, or fog, or have enough redundant systems to get signed off legally to remove the driver from the seat. But some form of limited full autonomy has already been demonstrated in public videos.

As for the competition - they said 2021 earlier this year - then Musk dropped the bomb and moved the time table up. It is going to be increasingly awkward for the sales staff at the competition a year from now. Can you imagine? "So - I test drove a Tesla which navigated intersections and stop signs. Why can't this Benz you want $120K for even stay in lanes on the 405 when the lines get fuzzy?"

Dude - five years is no longer viable. They will have to respond somehow.

 

[calisnow]

Remember too....no other car company has a customer base that is willing to buy and support features that are "in progress". You think MB could ship AP hardware only, with a promise of software delivery?

Love my steam gauges!

Yes, I do think MB could do the same thing and I think it likely that they soon will - along with other manufacturers who are watching Tesla walk away with the prize. Once customers are educated to understand how neural networks work they will understand that shipping hardware first, allowing large fleet of individual neural nets to drive around and iron out corner cases in shadow mode - and THEN turning on features for customer use is the only viable way to build this artificial intelligence.

What the press and so many enthusiasts don't understand is that they falsely conflate shipping half-baked product with the necessary process of giving neural nets time to learn and find the corner cases.

Mercedes is not shipping hardware with a promise of software delivery - they're shipping crap software with no hint it will ever improve unless you buy another Mercedes next year.

Claiming that their customers will not tolerate "in progress" features is just an easy excuse to avoid the hard work of building a fleet of cars that can learn - or if you just don't have an answer for your competition. And in fact it may be even a lamer excuse - the real reason may be that existing dealer networks don't want to get cut out of the process of improving cars.

 

[whitex]

Industry didn't respond because there is nothing to respond to. AP2.0 today is pure vapourware. AP1.0 did not live up to all its promises. I'm sure there was fine print in there somewhere or lack of detail that will let Tesla lawyers say they are covered, similar to the 691hp P85D they sold to people implying software upgrade is coming to get there, and of course we know that it ended with under 500hp, slightly more with a $5K retrofit of ludicrous. This time around there is an actual disclaimer that there are no guarantees whatsoever. Your car may be way past it's warranty turning into dust by EAP+FSD will work as advertised, if ever. So what is there to respond to?

 

[calisnow]

Industry didn't respond because there is nothing to respond to. AP2.0 today is pure vapourware. AP1.0 did not live up to all its promises. I'm sure there was fine print in there somewhere or lack of detail that will let Tesla lawyers say they are covered, similar to the 691hp P85D they sold to people implying software upgrade is coming to get there, and of course we know that it ended with under 500hp, slightly more with a $5K retrofit of ludicrous. This time around there is an actual disclaimer that there are no guarantees whatsoever. Your car may be way past it's warranty turning into dust by EAP+FSD will work as advertised, if ever. So what is there to respond to?

Nobody cares about the "horsepower" internet flame fight except a few P85D owners on a nerd forum. The story that matters when selling cars to customers is that the P85D is/was crazy stupid fast and the competition had no answer for it. You and I are fringe nut cases - we are irrelevant to the story that sells cars. That fact is borne out in Tesla's continually rising sales numbers. What matters for the company is when the doctor/lawyer/whoever goes and test drives the P85D, poops his pants and then writes a check after changing his underwear.

So - same goes for Autopilot 1.0 - nobody but some angry nerd somewhere remembers that Tesla mentioned some details about AP 1.0's capabilities that didn't come true. The angry nerd doesn't take the time to think about how Tesla overdelivered on the key features, that the competition has no answer for it yet, that Tesla's sales are increasing at a geometric rate, and that Tesla just leapfrogged itself with 2.0.

AP 2.0 is most certainly not vaporware - it's in everyone's cars - you, I and everyone else understands that it will soon get turned on and will be vastly more capable than 1.0.

Again - if you want to argue about the definition of full self driving, about regulations in different states, about weather conditions, redundancy, etc. - you are proving yourself to be the irrelevant nerd who doesn't understand business, marketing and the larger story at play.

You don't matter - you're not the target and you're not most customers. What matters is the test drive where Jim-the-dentist puts his hands in his lap and the Tesla salesperson says "navigate to the other side of town on surface streets" and then Jim is blown out of his mind as the Tesla drives itself through surface streets over to Starbucks.

If you or other forum members were in the back seat on that test drive and started going "Um, excuse me - if it were raining heavily right now this could not happen. I must point out as well that regulation 534.x21 in the vehicle code specifies, Mr. Dentist, that you may not leave the driving seat. Now may I give you a presentation about Level 3/45 autonomy and after that a history lesson of Mr. Musk."

This is the point at which guys like you and I get stuffed upside down in a trashcan like back in high school - and the dentist and salesmen laugh together - after which the dentist "rides" off in his new self driving car - and Benzo just lost one more customer.

That is the self driving that is now the bar for the industry in 2017 - not 2021. It matters dearly for car sales to the competition which already sees Tesla rushing up on it - and that is what the industry must respond to - their execs know this.

 

[larmor]

So it's been over a month since Tesla started putting in the AP 2.0 hardware. Yet the industry response has been dead silence. Is it possible that Tesla's lead in autonomous driving actually growing vs the industry catching up to them? I am a bit surprised there has been no flurry of marketing announcements by Benz/BMW/Audi in the last month saying they are accelerating their own efforts in autonomy.

It's been two years since AP 1.0 hardware shipped - and the closest competitor is the 2017 Mercedes system. Much ballyhooed by Benz last Spring prior to its release, it got panned and ridiculed by journalists this summer for its poor interface and poor performance in the task of lane keeping - despite superior hardware.
And as far as we know Benz has not announced any improvements to the 2017 cars will be released - even as dealer firmware updates - so as far as we know it will not improve.

Now Tesla leapfrogs itself by releasing a system with more than an order of magnitude more computing power - and Tesla will soon have the world's only fleet generating billions of miles of data on this new multicamera system.

If self driving is a harder problem than we realize - more strange corner cases, etc. - then maybe Tesla is building a moat.

Folks keep wringing their hands over the supposedly impending end of Tesla's competitive advantages in electric drivetrains, battery systems, charging networks and autonomy - yet the lead just seems to keep growing year after year.

One last example is the numerous German cars claiming 300+ miles range by 2018 on vapormobiles - by which time we conceivably could see a 120D with 400 miles plus.

You drive further on paper than on an actual road...

 

[RubberToe]

If you don't have anything nice to say, don't say anything at all?

Mom, is that you?

RT

 

[Canuck]

Muted Auto Industry response to Tesla's Full Self Driving move - why?

Because they have this attitude:

AP2.0 today is pure vapourware.

And they don't realize it's coming much sooner than they anticipate. As I posted in another thread:

“Ninety-four percent of crashes on U.S. roadways are caused by a human choice or error,” NHTSA Administrator Mark Rosekind said in a statement. “We are moving forward on the safe deployment of automated technologies because of the enormous promise they hold to address the overwhelming majority of crashes and save lives.”

The NHTSA is pushing automakers to remove humans from driving as much as possible because of this 94%.

And this:

U.S. regulators are eager to accelerate development of robotic driving systems to reduce traffic fatalities, which topped 35,000 in 2015.

35,000 death last year alone! Compare that to US casualties from other causes the vast majority of which we are not on the brink of preventing, or at least significantly reducing.

“In the 50 years of the U.S. department of transportation there has never been a moment like this, a moment where we can build a culture of safety as a new transportation technology emerges,” Transportation Secretary Anthony Foxx said at a Washington press conference on Tuesday. “Today we put forward the first federal policy on automated vehicles, the most comprehensive national automated vehicle policy that the world has ever seen. It is a first of its kind, taking us from the horseless carriage to the driverless car.”

The guidelines include a 15-point Safety Assessment governing the "safe design, development, testing and deployment of automated vehicles," the agency said late Monday.

That was 10 weeks ago. I say it's coming within months. It will start with a driver in the seat but the vehicle will do pretty much everything, as Tesla has shown in their videos. The NHTSA will not hold this back. They want it out and they want Tesla to lead the way (or anyone really but Tesla is the first to have full L5 hardware, including processors, on its vehicles) to show that we have the technology today to prevent a death that is happening right now -- due to human error! Motor vehicle crashes are the leading cause of death for U.S. teens. As a parent with teenage kids, it's my number one fear in life. That says a lot.

 

[RubberToe]

Stereo cameras and good enough software can navigate a vehicle through ANY situation. Every corner case imaginable happens thousands of times every day. It's solved by something called a human brain.

8 cameras and ultrasonic sensors are more than needed for level 5 autonomy.

All that is required is a bunch of very smart software. That's all. Best people in the world working on it, not just at Tesla. If you thing it is not going to happen, you are kidding yourself.

It's just code.

RT

 

[Spidy]

“Ninety-four percent of crashes on U.S. roadways are caused by a human choice or error,” NHTSA Administrator Mark Rosekind said in a statement. “We are moving forward on the safe deployment of automated technologies because of the enormous promise they hold to address the overwhelming majority of crashes and save lives.”

The NHTSA is pushing automakers to remove humans from driving as much as possible because of this 94%.

 

[Mo City]

some form of limited full autonomy has already been demonstrated in public videos.

As for the competition - they said 2021 earlier this year - then Musk dropped the bomb and moved the time table up. It is going to be increasingly awkward for the sales staff at the competition a year from now. Can you imagine? "So - I test drove a Tesla which navigated intersections and stop signs. Why can't this Benz you want $120K for even stay in lanes on the 405 when the lines get fuzzy?"

Dude - five years is no longer viable. They will have to respond somehow.

Agree completely. Full 100% Level 5 in all conditions won't happen with AP2. I'm not convinced truly "driverless" cars will exist in the next 20 years. But AP2 is poised to push autonomy to another level that will give Tesla an even bigger advantage than it has now.

After AP2 rolls out and improvements are noted publicly, sometime next year competitors will announce earlier timelines. But for now, there is nothing for them to say.

 

[Canuck]

From the slick Daimler powerpoint:

Our challenge: Not making errors would humans would make none.

That's hilarious. Tell me one area of driving where humans make no errors? But it gets even better:

Humans do much more right than they do wrong when driving.

We are now in bizarro land talk. If we did more wrong than right our cars would forever be in the body shop and us in the hospital or worse.

35,000 people killed a year on the roads and Daimler is praising us!

They don't even see what's coming. It reminds me of Blackberry saying people like physical keyboards.

I do, however, think drivers will be in the driver's seat for some time yet - perhaps a couple of years. That only makes sense as to how this technology will evolve and be proved safe to the public. And meanwhile Daimler is pumping out vehicles without even $50 backup cameras while producing these slick videos with points in their powerpoint presentation that are deserving of a comedy routine and not serious thought.

 

[Spidy]

And to say humans do more right than wrong when driving is bizarro land talk.

Did you have an accident this week?