Public Split Between Tesla And Autopilot Chip Provider Mobileye Gets Messy

SEP 15 2016 BY JAY COLE 74

The split between Autopilot chip supplier Mobileye and Tesla has not only become quite public, but it seems to be getting a little bit messy too as both companies are starting to make some strong statements, not to mention that 3rd parties are also starting to wade in on the former relationship, and even Mobileye’s technology itself.

After a fatal Tesla Model S accident with Autopilot enabled last May, Mobileye reportedly decided to end a supply agreement with the electric car maker at the end of the current contract, citing disagreements on how the technology was deployed and being portrayed.

Mobileye Shield +

Mobileye Shield +

A key factor in deciding to cut ties with Tesla was due to how Mobileye might be publicly be viewed being associated with the high profile nature of Tesla’s Autopilot should more accidents occur, and how Tesla might handle them.

Mobileye currently has around 70% of the market share for collision detection systems, via 27 different automakers worldwide – the risk may simply have been too great for the return potential.

Well a recent war of words would seem to indicate that contract has now ended:

“It is not designed to cover all possible crash situations in a safe manner.  No matter how you spin it, (Autopilot) is not designed for that. It is a driver assistance system and not a driverless system,” said Amnon Shashua who serves as both Mobileye’s Chairman and Chief Technology Officer to Reuters in an interview.

Amnon Shashua, Mobileye Co-Founder, Chairman and CTO sites with Brian Krzanich, CEO of Intel (left) and Harald Krüger, Chairman BMW

Amnon Shashua, Mobileye Co-Founder, Chairman and CTO announced autonomous driving by 2021 with Brian Krzanich, CEO of Intel (left) and Harald Krüger, Chairman BMW

Tesla was quick to issue a statement of their own in response to Mr. Shashua

“Since the release of Autopilot, we’ve continuously educated customers on the use of the features, reminding them that they’re responsible to keep their hands on the wheel and remain alert and present when using Autopilot.  Drivers must be prepared to take control at all times.”

These comments come after Tesla said in a statement on the breakup that Mobileye, that the company couldn’t keep up the pace with Tesla’s product changes, with CEO Elon Musk adding “our parting ways was inevitable” at a press conference in July.

The Mobileye Chairman noted his company had reservations about their relationship with Tesla after watching how the company responded to the fatal Florida crash, saying Tesla was sending mixed messages – talking up Autopilots capabilities while also continuing to caution drivers to always keep their hands on the wheel.

“Long term this is going to hurt the interests of the company and hurt the interests of an entire industry, if a company of our reputation will continue to be associated with this type of pushing the envelope in terms of safety,” Shashua said.

Separately this week,  George Hotz of who just introduced his $999 autonomous driving add-on – the Comma One from TechCrunch Disrupt SF, and who almost took over part of Tesla’s Autopilot development program from Mobileye (until reportedly Elon Musk “called him up on his birthday and changed the deal“) had some choice words about Mobileye.

Check the video above from the 7:20 mark to hear Hotz’s comments (below) on Tesla and Mobileye and his involvement in the split.

Comma One promises abilities "about on par" with Tesla's Autopilot for $999

Comma One promises abilities “about on par” with Tesla’s Autopilot for $999

“They (Tesla) were using Mobileye’s chip, and Mobileye makes this very low power chip that has a lot of hand-coded rules about driving, and it really isn’t very good, because it’s not designed to be a self-driving car at all.  

Mobileye wants to work with regulators to lower the safety ratings to put their (makes air quotes) “safety features” in right, but it’s not a self driving chip.  So I said that I could build something way, way better.  I went to Elon Musk and I told him this, and he said “Ok, we’ll work out a contract.”

Mr Hotz, says he holds no ill-will with Tesla (as it prompted him to move forward with, and he infers that he started the rift between the two companies, or at least fueled the fire by stating:

“I regret any trouble that I caused him (Musk) in his relationship with Mobileye, who like I said, is a failing company…”  Not done, the Comma One creator added, “If you are looking to Mobileye to build a self-driving future – don’t.”

So is this the end of the back-and-forth between the two companies? We imagine not.

Reuters, Hat tip to sven!

Categories: Tesla

Tags: ,

Leave a Reply

74 Comments on "Public Split Between Tesla And Autopilot Chip Provider Mobileye Gets Messy"

newest oldest most voted

Clearly, Tesla’s implementation of driver assistance using a Mobileye EyeQ3 chip is far from a reference implementation. Further, Tesla specifically shielded Mobileye from criticism. It is Mobileye that seems to want to try to position themselves post break-up for future product by disparaging Tesla.

If Tesla’s implementation is so disturbing, why is Mobileye still taking Tesla’s money and still shipping EyeQ3 chips to them?

It is GeoHotz’s comments that are most cutting, and the fact that Tesla was even shopping around to replace them. Mobileye’s development arc appears to be both slower and less capable by their own curious admission.

How was Amnon Shashua’s quote disparaging? Genuinely asking, as it appears to me to be factual, and no different than how Musk himself describes AutoPilot. Is there another quote I’m missing, or just something I’m not understanding?

Did you not read the Reuter’s article?

I read the Reuters article, and I don’t see any quote there which the above summary didn’t repeat.

I’d like to know what you’re talking about, too. Yeah, some of the quotes about Tesla were negative, but I didn’t see anything which appeared to be distorting the facts, or presenting a biased viewpoint.

He’s implying that his system isn’t designed to see semi trailers in front of cars.

The Mobileye system failed spectacularly to do basic level 2 collision avoidance – let alone any sort of autonomy. So saying it’s not okay to do level 3 stuff is fairly damning.

His quote implies Mobileye is the one misrepresenting their product by saying Tesla is specifically misusing it.

Tesla did not shield Mobileye from criticism, explicitly calling out the failure of the camera in the second Autopilot death. This, despite the fact that the system was never designed to handle this specific case, and Tesla knew it.

Tesla was just stating the facts… and Elon Musk tweeted, “Btw, want to thank both Bosch and MobilEye for their help and support in making Autopilot better. Please direct all criticism at Tesla.”

Four Electrics posted more FUD:

“Tesla did not shield Mobileye from criticism, explicitly calling out the failure of the camera in the second Autopilot death.”

That’s factually incorrect… just like all your Tesla bashing posts.

Here is the actual quote from Tesla, which does not contain either the word “camera” or “Mobileye”:

“Neither Autopilot nor the driver noticed the white side of the tractor trailer against a brightly lit sky”

But of course, the fact that it’s completely untrue won’t stop you from repeating this B.S., now will it “Four Electrics”?

This is just another example of you childishly picking on the commenter at any opportunity rather than discussing the issue.

I’m not that familiar with the intimate details (other than some collisions), but I would give ‘four electrics’ the benefit of the doubt since he has owned 4 electric cars.

Also who appointed you to critique everyone who has ten times the knowledge, and 100 times the experience that you do?

Oh and by the way, he, I, and others truly say we’re sorry on the rare times we’re mistaken.

YOU, on the other hand, just give a perfunctory, “WELL, Hell, I don’t know anything about anything technical so it doesn’t matter what I’ve said in the past”. (!!!!)

Not designed for what? Mobileye says ‘not designed for autonomous’ implying Tesla misused their product.

But what their product failed was a very basic ‘see big semi in road’ auto-braking scenario. That the car was using lane-following and adaptive cruise control has little to do with the failure.

“If Tesla’s implementation is so disturbing, why is Mobileye still taking Tesla’s money and still shipping EyeQ3 chips to them?”

Because they have a contract.

Contracts have exit clauses

And maybe using the available clauses is not in their best interest…

Tesla’s AUTOPILOT should be called ENHENCED CRUISE CONTROL… But “Autopilot” is more “show off” for marketing and proud and happy drivers. Happy Tesla’s driver are so proud of their Tesla that they are ready to risk their life anyway. This my understanding.

View Bloomberg’s interview of Elon Musk at the launch of Tesla’s Autopilot. He makes the distinction between autonomous driving and autopilot quite clearly.

Further, the name really doesn’t matter. People will do what people do, and find a way to deflect blame.


Here’s the thing. You can’t take your hands off the wheel, and its now no good for parking either, except perhaps in an empty parking lot with non-white walls or guard rails.

And ‘risk your life’ is an understatement.

If something goes wrong, you are held liable in that it is adjudicated that you were too stupid to use it.

So, what is it good for?

You confirm by your comment that you are not only an indefectible fan of GM (mostly all of your comments), but also exaggerating the flaws of the Tesla. Autopilot saved many lives, and there is only ONE documented failure of the auto-park feature. How many dings have been avoided? You don’t know.

There is zero evidence that Tesla Autopilot has saved even one life. It has however been involved in the loss of at least two lives.

Really the Harry Potter movie watcher guy who thought it was autonomous driving car and who is the other one?

So says the hater that I already called out for claiming a la Trump that:

“I am frustrated by their lack of participation in the general transition to sustainable transportation.”

See here at bottom of comments:

Two? I’m only aware of one?

But it MAY have saved many lives. Of course there’s no evidence, because it’s impossible to prove!

Rexxsee, sorry, you are just another armchair quarterback that has no experience with any of this. I criticize GM more than I criticize Tesla, however lately I’ve been also complementing GM alot. Not today however since I took my ELR to the dealer at their suggestion, to fix a less than trivial problem, and no the car wont start so they loaned me an ATS but I was planning to go to the Ithaca Drive electric event, and unfortunately the VOlt I have is spoken for that day.

The first time I heard a Tesla owner proudly talking about “Autopilot”, my understanding was like : “you enter a destination in your Tesla Google maps and the car drives you there completely automaticly.

Autopilot: a device that steers a ship, aircraft, or spacecraft in place of a person.

While a pilot stays alert and ready to take the controls at all times. There, I completed your sentence for you.

It’s not comparable because there is plenty of time to regain situational awareness when disengaging an aircraft autopilot.

Not truth an air collision happened faster than a car accident.

“Not truth an air collision happened faster than a car accident.”


1. Air warning timing is much longer than driving due to the space involved.

2. Air traffic is much more sparse than car.

3. Air travel has 3D avoidance capabilities where car is mostly 2D only.

4. Speed of collison is faster but that doesn’t equal to the less time to avoid since the air system gives you much more time for reaction. That is why they also have a Co-Pilot.

The problem with these semi autonomous steering systems is they cross an “engagement” threshold, which runs against basic human characteristics.

Every driver aid up till now still requires a sufficient degree of driver engagement such that some concentration on driving the car is required.

These systems promise autonomous driving without drivers doing anything, but you have to be ready to take over at any moment. It is an unnatural state because humans can’t concentrate on doing nothing, so invitably end up concentrating on something else, or taking a nap. One ends up being lulled into a sense it really is a fully autonomous driving car, and then when an event occurs the time required to re-gain awareness and re-gage often means it’s too late to recover.

And let’s be honest, the marketing hype really does incorrectly imply full autonomy, regardless of the fine print. Add in a dose of Silicon Valley hyperactive kids with short attention spans (see the interview) and it is a recipe for real safety problems. I think I will keep my hands on the wheel for now.

Yes, it’s a well-known human factors issue. Safety-critical systems are designed to minimize “attention deficit” issues. Tesla’s implementation of Autopilot tends to maximize attention deficit.


Exactly. This is the same reason that humans are really bad at things like sentry duty, which requires being on full alert for many hours straight, watching the surroundings carefully while 99.5% of the time nothing actually happens.

NASA agrees with you, and this article pretty much sums up what you’re saying.

Dominic, you obviously don’t own a Tesla…I do. It’s not a risk if you leave your hands on the steering wheel as required when you agree before driving the car. The people who get into accidents are the people who don’t follow the instructions and stay alert.
The car will keep you in the center of your lane at a safe distance from the cars in front of you and you don’t have to really steer;
you just lay your hands on the wheel.
You don’t have to press the accelerator or the brake and it will keep you perfectly in the middle of your lane.
When you’re on the highway, it’s perfect. When you’re stuck in traffic, it’s great.
When they announce the vehicle is autonomous, then it will be autonomous.
You appear to be in the group of people who would mistake autopilot enabled for an autonomous car.

Dominic Matte said:

” ‘Autopilot’ is more ‘show off’ for marketing and proud and happy drivers.”

I thought so too, originally, but I’ve learned better since. The problem is that most people have an exaggerated idea of how an airplane’s autopilot is used. Actual pilots know that they can’t just set the autopilot and then leave their seat, or take their attention off the sky. But most of us aren’t trained pilots.

Of course, it’s reasonable to argue that this very misperception by the general public is sufficient reason for Tesla to be asked change the label.

No, ‘autopilot’ describes very well what Tesla’s product does – certainly in the context of the usual use of the word ‘autopilot’ ie in the context of its use in aircraft. If you knew what an aircraft autopilot can and can’t do, then you would know this as well. Is it tesla’s fault that there are so many lazy and/or ignorant people in the world who can’t be bothered to spend 6 seconds doing an internet search to look something up? Look, I’ll even save you the trouble…

Autopilot on airplanes doesn’t stop them from flying into mountains.

So where did anyone get the idea that it was completely hands free?

Hat tip to Sven. So take it with a grain of salt.

This split was inevitable however. And if Mobileye thinks they can supply same staff to all car makers, they should think again.

This is not some seat belts or airbags you know. It is going to define the car itself, each car. And each car maker will try to get exclusivity because will need it.

So is Tesla exclusively using now?

no it sounds like he was rebuffed by musk and is just doing the comma one AP aftermarket system now:

“Separately this week, George Hotz of who just introduced his $999 autonomous driving add-on – the Comma One from TechCrunch Disrupt SF, and who almost took over part of Tesla’s Autopilot development program from Mobileye (until reportedly Elon Musk “called him up on his birthday and changed the deal“) had some choice words about Mobileye.”

After a second cup of coffee, I could see the word “almost”.

So who does Tesla use now for AP?

Did they make up? What about this comment from Hotz:
I went to Elon Musk and I told him this, and he said “Ok, we’ll work out a contract.”

Minus 10 points for not reading the article.

I read the article. Late night, coffee was required.

Tesla please focus on making practical, affordable electric vehicles. Of all the great features that Tesla offers, for me, Autopilot is the ‘least’ compelling reason to buy a Tesla car.

Autopilot is the reason I bought my Tesla

I love autopilot and it works great, just pay attention. It has saved more lives than it has cost. Do people realize driving a car is dangerous?? 38,300 people were killed on US roads last year.

AutoSteer, Lane Change, and Traffic Aware Cruise Control are in the top 3 of my favorite things about the Tesla car I own. It is really a dream for highway driving.

Well, I understand your desire for good, cheap EVs but again Ap is at least 20% of the reason I would buy another Tesla.

We know Tesla is using the same sensors in their system since 2014-yes? and that Tesla has supposedly improved the system in its last rev. It was my understanding that they are just processing the data from this same mobile eye sensor with more sophistication.

Sounds like Mobileye and Tesla disagreed on this additional processing for some reason–perhaps mobile eye thought Tesla was squeezing more out of the sensor than it was good for —or perhaps mobileye thought it was an infringement on their domain.

great sleuthing Sven and Jay!!

I love what George says, at 15:30 in the video, where he compares current autopilot systems to early on new Teenage Drivers, where they (Both) are still learning, ‘getting their feet wet’, making mistakes, being corrected, and learning to drive better! Of course, it us one thing to learn how to drive up the road safely, or follow a car safely up the road, and a total other thing to learn how to anticipate what elements around you in traffic, could lead to an accident, unless you: a) Speed up, b) Slow Down, c) Change Lanes, or ‘D’) – pick 2 of the above (Speedup and Change Lanes; or, Slow Down and Change Lanes)! Some times a quick speed change is needed to fit in a small hole in traffic beside you, so you can safely change lanes, and better drivers have learned that; average drivers – not so much, but some have; and really bad drivers just hit what’s in front of them, never learning how to manage obstacles! When we can hear of that type of obstacle avoidance from Tesla Autopilot, we will know that it truly HAS BEEN learning how to drive, and Drive Better than Average!… Read more »

Self-driving cars need active scanners, such as lidar, or radar which can detect moving objects out to 100 yards or so. (The radars in current Tesla cars are far too short-ranged.) Cameras simply are not adequate.


But the active scanners can’t tell what’s irrelevant, but in concert with the cameras… They could tell things apart.

Which is how Google’s system works.

I think Tesla bit off more then it could chew with working with the self driving car system before it could build the model 3. In that now with the fatal accident and it raising a ton of questions.

This could cause Tesla to have to spend it’s limited resources on this auto driving mess. Instead of the model 3.

It is spending its resources on the Model 3. Note the big spend on the Gigafactory. Note the $1.27 billion in expansion at the Fremont factory.

Further, Tesla is by far not the only company working on driver assistance and autonomous driving vehicles. Nissan, Mercedes, and so forth are also doing it. They just haven’t managed to develop it as far. And there are no real questions. Both drivers were not using Autopilot correctly – they both were not paying attention at all.

That street sweeper situation in China would not be permitted in most of the developed world.

Every day in the developed world, cars break down in the left hand lane.

And most roads have a shoulder. The ones that don’t have a higher chance of accidents. This kind of accident happens w/o driver assistance technologies.

A street sweeper wouldn’t be left without warning in a left lane; there would be a train of hazard lights and vehicles around it.

That’s how it works here in the US, when sweeping the left lane or shoulder.

Elon, we need to see the data. Tesla is saying that they are collecting data and using it to hive-train all cars. Show us the data.

How many accidents did AP prevent?
How many times does the driver have to intervene?
How many drivers are using AP in an unsanctioned/unsafe manner?

It is impossible to say, for any given incident where autopilot intervened, because you’d have to know what would have happened had it not been engaged (the control case), and you can’t rerun the incident… It’s even worse, because various drivers would have reacted very different in the exact same situation.
So there’s no way of saying specifically “AP saved a life in Incident X”.
The only thing one can do is statistics — comparing the probability of an accident when AP is and isn’t — but even then, there are so many outside factors one would need to control for, I doubt there’s enough real data.

Now here is someone who understands the concept that there are limits to knowledge; that it’s impossible to know everything.

Thank you for explaining that, wavelet.

Guy sounds very arrogant, and given he has a dog in this fight (pushing his product), isn’t worth quoting. He’s no expert on the subject.

I know some of people involved in MobilEye, and I trust their extensive experience much more.

IMO it should be illegal to install any such system for use on public roads without many years of regulatory testing (which is the careful way all carmakers appear to approach this, except Tesla). There many factors here, and it’s a really complex problem to solve.

It is all just a demonstration in how real progress is made in the real world, its made incrementally 99% of the time.

If it wasn’t for disrupter par excellance, Elon Musk pushing this through Tesla then the progressing of this technology would have remained at a very slow pace as the laggard OEMs continued the trend of glacially slow innovation.

Musk through Tesla represents that ONE percent of fast-paced innovation that pushed the boundaries of the existing tech which in the case of the existing Mobileye tech has apparently been proven to be inadequate and this will lead to further improvements and innovation which we will all benefit from.

SparkEV-Fiat500-Leased - M3 Reserved - Bolt- TBD

So Tesla divorcing MobileEye and Comma is out. Who’s up from the other 25% market share to take on the task.

Where is AP now with this divorce is the primary question. Who owns what is key.

Reading between the lines, it looks like Mobileye was afraid of being names in lawsuits following accidents with semi-autonomous cars. While that’s understandable, it’s rather cowardly of them to not own up to their real motive, but rather try to shift blame to their business partners, and say things like:

“Long term this is going to hurt the interests of the company and hurt the interests of an entire industry, if a company of our reputation will continue to be associated with this type of pushing the envelope in terms of safety”

They should blame our litigious society, not their business partners!

I wonder what their product is for, considering it failed a level 2 collision avoidance that they said it was for and not the level 3 that Tesla was trying.