Perhaps Mobileye should have been given more credit , and less focus on autopilot vs EV

[S4WRXTTCS]

With the less than stellar performance thus far of Tesla's own AP2, one has to wonder if perhaps Mobileye was not credited sufficiently for the success of AP1. Certainly in hindsight Mobileye's product seems much better than what Tesla has developed thus far on their own.

I also wonder if perhaps autopilot is a distraction that will serve to delay the acceptance of EVs. After all, Tesla mission statement is to advance EVs, not to advance self driving cars. It seems like AP is consuming a tremendous amount of engineering and resources that could perhaps be better served developing the drivetrain and battery.

Thoughts?

There is a lot of false equivalencies.

The less than stellar performance isn't from what MobileEye really provided. There hasn't been any evidence that suggests that AP2 see's the lines less than AP1. It's simply observed that how it behaves isn't as good, but that isn't provided by MobileEye. MobileEye certainly didn't provide all the mapping information. TACC is also an area where AP2 has had some false detection issues, but I have to wonder if Tesla didn't add more cases where it would slow down for, and that's why we're seeing some false alarms.

There are things like AEB that MobileEye provided that are tried and tested, but the reality is that systems from Subaru and Volvo scored higher than MobileEye based systems. Like Subaru can stop at speeds differentials of around 30mph.

The reason AP2 is currently problematic (but getting better) is it was rushed without adequate time to work on things. Tesla didn't have the >6 months or longer that they took to release the autopilot element of AP1.

I don't think autopilot serves as a distraction that will delay the acceptance of EVs. The teams working on the two have very different competencies, and EV's lend themselves really well towards autonomous driving. Plus your quoting Tesla Motors mission statement, and not Tesla as it's now known. If you're an energy company it's in your best interest to have cars drive themselves to use lots of what you sell.

Now certainly AP serves as a distraction for owners where owners are so obsessed with it that they don't realize how enjoyable the car is to actually drive. As an AP1 owner I'm pretty satisfied with just TACC only. I find it rather amusing how so many people gush over AP1 when my own experience thinks of it as being a bigger hassle than it's worth. I guess I fail to see the excitement of a system where you have to hold the wheel. I guess that was cool when I was like 3 and my dad was driving.

There is also something really funny about growing up on Terminator movies, but then holding the wheel for Skynet.

 

[JRMW]

I also wonder if perhaps autopilot is a distraction that will serve to delay the acceptance of EVs

Possible.
On the one hand, AutoPilot is "cool". it's not enough for EVs to be "as good as" ICE vehicles. In order to make humans change, there must be something better. There were a lot of MP3 players on the scene. But Apple's iPod had iTunes, which was simply better. Dumb mobile phones didn't really kill the land line. Smart phones did. and so on.
Tesla is changing the world because Teslas are better than ICE vehicles in many, very visible, ways.

On the other hand: Tesla needs to be careful. They are juggling many spinning plates. In order to stay alive they must:
-get the Model 3 out in a reasonable time frame (a year late won't kill them, 2 will).
-improve Quality on all future Teslas. Tesla cannot get the Jaguar reputation
-increase SuperChargers. Number of stalls and speed of charging.
-Increase Service Centers. Number and quality.

If Autopilot efforts are in concert with the above, Tesla will survive.
If Autopilot comes at the expense of the above, they will fail.

Spinning plates.

 

[malcolm]

There is also something really funny about growing up on Terminator movies, but then holding the wheel for Skynet.

"Come with me if you want to drive"

Shame we're not at the "I see everything" stage.

Or maybe it's a good thing.

At what point does "Summon", summon the Demon?

 

[182RG]

Spinning plates.

Well said. I would also add Ludicrous and Insane mode to this as well. Useless distractions.

Focus on QA, and getting a cost effective, reliable M3 to market on time. The other, very real danger is the promise of a $35K M3. It will be a very long time, if ever, that one is delivered in "stripped" configuration.

 

[malcolm]

I would also add Ludicrous and Insane mode to this as well. Useless distractions.

True, but worth their weight in marketing.

 

[AnxietyRanger]

True, but worth their weight in marketing.

Possibly not if they cause major liability issues, though. Tesla has certainly lost a lot of goodwill due to the various P85D and P90D issues, from their best paying customers. And it has cost Tesla time and money to figure out, with not insignificant future warranty costs still quite likely.

That said, I'll address the flying elephant in the room, the greatest distraction of them all: the falcon wings. If Tesla had simply poured all that energy into ramping up conventional Model S and X, as well as willing the Model 3 into existence faster, the EV cause would very likely be much further along and there would be tons more Teslas on the road (the simple fact that Model X would have been expanding the market a year+ sooner and quite likely with conventional doors captured more market through familiarity and cheaper price).

These problematic pet-projects have quite likely had a major cost for that progress. Not fatal cost, but still a cost that is quite likely not paid by any marketing advantage.

And I say this as a guy who certainly personally likes the falcon wings and the Ludicrous mode. But in all honesty, I seriously don't think they have been a good thing for Tesla's business.

 

[PaulusdB]

Au contraire, @AnxietyRanger.

Those pet projects turning sour was a necessary step for Tesla to grow towards adulthood. The limits of their various capabilites were tested and confirmed within a confined scope. They learned from that and toke the anti-'hubris' medicin. For a better future of the company and products, now that the scope is growing.

(Some existing customers are left behind in varying degrees of dissatisfaction, due to those past mistakes. Hope we can see this providing new lessons.)

 

[stopcrazypp]

Au contraire, @AnxietyRanger.

Those pet projects turning sour was a necessary step for Tesla to grow towards adulthood. The limits of their various capabilites were tested and confirmed within a confined scope. They learned from that and toke the anti-'hubris' medicin. For a better future of the company and products, now that the scope is growing.

(Some existing customers are left behind in varying degrees of dissatisfaction, due to those past mistakes. Hope we can see this providing new lessons.)

Very good point. I remember the quote about hubris. It seemed to have made Elon super focused on making Model 3 easier to manufacture (even the "conventional" parts). Without the falcon wing doors, there probably would have been less focus on that part.

 

[AnxietyRanger]

@PaulusdB Sure, learning from one's mistakes is a silver lining.

However, I am not convinved Tesla needed to make these particular mistakes to learn, or that they are a net positive. Focusing on the mission is strategy 101, not some new discovery.

Willing the long-range EV into existence was a teachning moment in itself and that actually required new discovery. It is simply a weakness of the leadership they pushed it too far on the basics.

I mean, Tesla got it so right originally on their "no weirdmobiles" focus on normal car as EV. Their quest for falcon wings and Ludicrous clearly deviated from that focus - and some I guess could argue so did pushing for Autopilot a bit too soon to far for both AP1 and AP2... (I don't think having driver aids in general was a mistake, though, they are part of today's normal car.)

We all have weaknesses and make mistakes, of course. This is just pointing out where Tesla stepped away from the path they had so eloquently lined out.

Personally I think out of these deviations the falcon wings and monoposts must have been the most damaging one. Tapping into a more conventional (and huge) SUV market sooner and cheaper would probably have helped Tesla on demand quite a bit without need for discount quarters and other gimmicky demand levers.

 

[Bladerskb]

The reason AP2 is currently problematic (but getting better) is it was rushed without adequate time to work on things. Tesla didn't have the >6 months or longer that they took to release the autopilot element of AP1.

Ap2 started by atleast October. That's been more than 6 months.

 

[stevezzzz]

It was ten months from the time I took delivery of my P85D with AP1 hardware before Tesla deployed the first public firmware version that included Autosteer.

 

[FlatSix911]

Perhaps Tesla should have purchased Mobileye last year ... instead of going in house for development

Tesla’s ‘Vision’ and Autopilot chip efforts validated by Intel’s $15 billion acquisition of Mobileye

In the first generation of Tesla’s Autopilot, the company was using Mobileye’s computer vision technology powered by its EyeQ3 chip. This package, when combined with the third-party recommended sensors (cameras and radar antennas), constitutes the bulk Israeli company’s offering. Aftermarket products used on existing vehicles also represent about 20% of the company’s volume. Different configurations power safety features and driver assist features in different vehicles from OEMs. It can vary from active safety features, like automatic emergency braking, to convenience features, like active lane keeping.

Tesla had already been building features on top of Mobileye system with the first generation Autopilot, but the company has also long been working to discontinue the system. In 2015, Tesla CEO Elon Musk offered George Hotz a contract with a “multimillion-dollar bonus” for him to build a new Autopilot system in order for Tesla to discontinue Mobileye’s image processing system. With the introduction of the second generation Autopilot in October 2016, Tesla got rid of Mobileye with its own computer vision system called ‘Tesla Vision‘ and they powered it with Nvidia’s Drive PX2 onboard computer.

At the moment, Tesla’s own image processing is certainly not on par with Mobileye’s based on a simple comparison of the features it enables in Tesla’s second generation Autopilot versus the first generation. But a software update expected in the coming weeks could bring the system to parity.

 

[S4WRXTTCS]

Ap2 started by atleast October. That's been more than 6 months.

It took Tesla a greater part of a year just for Lane-Steering on AP1. Something where they had a starting point, and all they did was optimize it using the fleet learning.

There is absolutely no friggen way they would be able to develop AP2 to match AP1 within 6 months. To recreate everything from scratch? Sure it won't take as long as MobileEye took to originally develop it. They had the right idea with trying to use both Mobile Eye and NVidia where they slowly migrated off the one, and onto the other.

MobileEye wouldn't let them do that so something had to give, and in this case the buyer has to give. As to when it reaches parity is anyone's guess.

My guess it will never really exactly reach parity. Some elements will be better, and some elements might be worse.

Like AEB for example will likely brake for a greater amount of situations, and it will likely be more effective (>25mph speed reduction) but it likely will come with a few more false positives. The tradeoff will be likely worth it.

MobileEye was a pretty well baked solution that wasn't new and shiny, but it did what they promised it would. Tesla just took it too far.

 

[u00mem9]

"I view (autonomous driving) as a solved problem"
-Elon Musk, March 2015

 

[stopcrazypp]

"I view (autonomous driving) as a solved problem"
-Elon Musk, March 2015

Context is everything. Unfortunately the WSJ article is behind a paywall, but a long quote is: "I view it as a solved problem. We know exactly what to do and we will be there in a few years".

Essentially what he is saying is that the various companies have already figured out that self-driving cars are achievable and have a clear pathway to doing it, and that it is essentially inevitable (he made an analogy to elevator operators). The problem is "solved" in that context (the theoretical barriers are gone, so it's only a matter of when, not if). There are other problems that are not "solved" in this manner, a popular/similar example being flying cars (no clear path to making it a reality).

However, that doesn't mean there isn't any more development time needed (he explicitly said it will take a few years). In the same talk he mentioned the slow speed (under 10 mph) and highway speed is pretty well figured out. "Where it gets tricky is traveling at about 10 to 50 miles per hour in urban environments".
Tesla CEO Elon Musk and Nvidia CEO Jen-Hsun Huang declare self-driving cars “solved”

 

[AnxietyRanger]

It actually comes (or was similarly discussed) from the nVidia press conference, there is a video available.

 

[u00mem9]

Context is everything. Unfortunately the WSJ article is behind a paywall, but a long quote is: "I view it as a solved problem. We know exactly what to do and we will be there in a few years".

Essentially what he is saying is that the various companies have already figured out that self-driving cars are achievable and have a clear pathway to doing it, and that it is essentially inevitable (he made an analogy to elevator operators). The problem is "solved" in that context (the theoretical barriers are gone, so it's only a matter of when, not if). There are other problems that are not "solved" in this manner, a popular/similar example being flying cars (no clear path to making it a reality).

However, that doesn't mean there isn't any more development time needed (he explicitly said it will take a few years). In the same talk he mentioned the slow speed (under 10 mph) and highway speed is pretty well figured out. "Where it gets tricky is traveling at about 10 to 50 miles per hour in urban environments".
Tesla CEO Elon Musk and Nvidia CEO Jen-Hsun Huang declare self-driving cars “solved”

He also implied by making the statement that Tesla was the mastermind behind Autopilot's success.

Now we know that was smoke and turd-crusted mirrors. The way I view Elon shooting his mouth off at some programming conference, or South American space exploration sing-in has changed quite a lot over the last 18 months.

side note: Anyone remember when "shut up and take my money" was interweb speak for when can I buy it? Tesla has officially ended that phrase!

 

[AnxietyRanger]

He also implied by making the statement that Tesla was the mastermind behind Autopilot's success.

Now we know that was smoke and turd-crusted mirrors. The way I view Elon shooting his mouth off at some programming conference, or South American space exploration sing-in has changed quite a lot over the last 18 months.

side note: Anyone remember when "shut up and take my money" was interweb speak for when can I buy it? Tesla has officially ended that phrase!

Yeah, unfortunately Tesla/Musk have a lot of disillusioned former fans. The sad irony is, Tesla would be just fine - and in many ways better - without the hubris that makes the constantly overpromise and underdeliver.