Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla Vision ... The Next AP Hardware

This site may earn commission on affiliate links.
Here’s Tesla statement in full:

Recent statements made by MobilEye about Tesla Autopilot are inaccurate.

Here are the facts:
MobilEye had knowledge of and collaboration with Tesla on Autopilot functionality for the past 3 years.

Tesla has been developing its own vision capability in-house for some time with the goal of accelerating performance improvements. After learning that Tesla would be deploying this product, MobilEye attempted to force Tesla to discontinue this development, pay them more, and use their products in future hardware.

In late July when it became apparent to MobilEye that Tesla planned to use its own vision software in future Autopilot platforms, MobilEye made several demands of Tesla in exchange for continuing supply of first generation hardware, including:
  • Raising the price of their product retroactively
  • Demanding an agreement to extremely unfavorable terms of sale and
  • Demanding that Tesla not use data that was collected by its vehicles’ cameras for any purpose other than helping MobilEye develop its products
  • Requiring that Tesla collaborate on Tesla Vision and source future vision processing from them until at least level 4
When Tesla refused to cancel its own vision development activities and plans for deployment, MobilEye discontinued hardware support for future platforms and released public statements implying that this discontinuance was motivated by safety concerns.


While Tesla’s Autopilot has received more attention in the press as arguably one of the most refined such implementations on the road*, its feature set and hands-on requirements are very similar to other such systems, many of which use MobilEye vision products. In its current form, Autopilot is an advanced driver assistance system that maintains lane positioning while adjusting vehicle speed in the context of surrounding traffic. At no time has Tesla ever said or implied that Autopilot makes a car Autonomous or “self-driving” any more than autopilot on a plane, after which it is named, makes a plane self-flying. Any third-party descriptions to this effect are not accurate.

Since the release of Autopilot, we’ve continuously educated customers on the use of these features, reminding them that they’re responsible to keep their hands on the wheel and remain alert and present when using Autopilot. The Autosteer feature is off by default and must be enabled through the “Driver Assistance” tab in vehicle settings, where the user must agree to a dialogue box describing the system as “Beta” (to reduce complacency) and instructing them to pay attention to the road at all times. Every time the feature is subsequently activated, the user is reminded of this agreement with both visual and audible alerts. Failure to comply results in a rapidly-escalating sequence of audible and visual alerts, followed by deactivation of the Autosteer feature.

*Independent studies of Autopilot performance relative to similar Level 2 systems
  • Car & Driver: ~2x better than the next closest competitor (29 interventions vs. 56 for the BMW 7-Series)
  • Autofill: ~10x better than the next closest comparison (3 interventions on the Model S vs. 31 for the E-Class)
  • Motor Trend: 6-9x better than the next best system (12 interventions for the Model S vs. 113 for the S-Class)
  • The Drive: “Without a doubt, [Autopilot] is the best ADAS system on the market.”
 
Well, I guess we know more about AP 2.0 now...

Tesla will have to finish developing their new camera system before the next generation AP comes out. I'm guessing that means sometime next spring.

I don't think this tells us much about the timeline. All the press release says is that "Tesla has been developing its own vision capability in-house for some time".
 
I don't think this tells us much about the timeline. All the press release says is that "Tesla has been developing its own vision capability in-house for some time".

True. In another thread someone's making the argument that this response from Tesla must mean that Tesla is ready to ship their new cameras tomorrow, since they aren't worried if Mobileye were to refuse to sell any more cameras to Tesla.

That's why I use the word "guess". :) But we know that this summer they radically changed what AP 2.0 was going to be focused on (more radar focus, big online point cloud) and brought the whole camera system in house.

I'm betting even Tesla needs more time to pivot that far and make a cohesive product ready for the consumer. I could easily be wrong, but that's what the data says to me.
 
Here’s Tesla statement in full:

Recent statements made by MobilEye about Tesla Autopilot are inaccurate.

Here are the facts:
MobilEye had knowledge of and collaboration with Tesla on Autopilot functionality for the past 3 years.

Tesla has been developing its own vision capability in-house for some time with the goal of accelerating performance improvements. After learning that Tesla would be deploying this product, MobilEye attempted to force Tesla to discontinue this development, pay them more, and use their products in future hardware.

In late July when it became apparent to MobilEye that Tesla planned to use its own vision software in future Autopilot platforms, MobilEye made several demands of Tesla in exchange for continuing supply of first generation hardware, including:
  • Raising the price of their product retroactively
  • Demanding an agreement to extremely unfavorable terms of sale and
  • Demanding that Tesla not use data that was collected by its vehicles’ cameras for any purpose other than helping MobilEye develop its products
  • Requiring that Tesla collaborate on Tesla Vision and source future vision processing from them until at least level 4
When Tesla refused to cancel its own vision development activities and plans for deployment, MobilEye discontinued hardware support for future platforms and released public statements implying that this discontinuance was motivated by safety concerns.


While Tesla’s Autopilot has received more attention in the press as arguably one of the most refined such implementations on the road*, its feature set and hands-on requirements are very similar to other such systems, many of which use MobilEye vision products. In its current form, Autopilot is an advanced driver assistance system that maintains lane positioning while adjusting vehicle speed in the context of surrounding traffic. At no time has Tesla ever said or implied that Autopilot makes a car Autonomous or “self-driving” any more than autopilot on a plane, after which it is named, makes a plane self-flying. Any third-party descriptions to this effect are not accurate.

Since the release of Autopilot, we’ve continuously educated customers on the use of these features, reminding them that they’re responsible to keep their hands on the wheel and remain alert and present when using Autopilot. The Autosteer feature is off by default and must be enabled through the “Driver Assistance” tab in vehicle settings, where the user must agree to a dialogue box describing the system as “Beta” (to reduce complacency) and instructing them to pay attention to the road at all times. Every time the feature is subsequently activated, the user is reminded of this agreement with both visual and audible alerts. Failure to comply results in a rapidly-escalating sequence of audible and visual alerts, followed by deactivation of the Autosteer feature.

*Independent studies of Autopilot performance relative to similar Level 2 systems
  • Car & Driver: ~2x better than the next closest competitor (29 interventions vs. 56 for the BMW 7-Series)
  • Autofill: ~10x better than the next closest comparison (3 interventions on the Model S vs. 31 for the E-Class)
  • Motor Trend: 6-9x better than the next best system (12 interventions for the Model S vs. 113 for the S-Class)
  • The Drive: “Without a doubt, [Autopilot] is the best ADAS system on the market.”
Pivot-shift. Better to keep mobile eye on moving target....
 
Pivot-shift. Better to keep mobile eye on moving target....
This part seems important:
"When Tesla refused to cancel its own vision development activities and plans for deployment, MobilEye discontinued hardware support for future platforms and released public statements implying that this discontinuance was motivated by safety concerns."
Unless Tesla has a stockpile of cameras somewhere they are going to have to negotiate terms and a support agreement.
 
I think Mobile Eye has shot themselves in the foot and will pay a dear price for it. My guess is that they were trying to run roughshod over Tesla and they lost. These demands are not smart for building a long term relationship with the current leader in deployment of driver assist:

  • Raising the price of their product retroactively
  • Demanding an agreement to extremely unfavorable terms of sale and
  • Demanding that Tesla not use data that was collected by its vehicles’ cameras for any purpose other than helping MobilEye develop its products
  • Requiring that Tesla collaborate on Tesla Vision and source future vision processing from them until at least level 4
I think Mobile Eye blew it big time. Can you imagine Daimler agreeing to these terms?
 
  • Like
Reactions: Saghost
I think Mobile Eye has shot themselves in the foot and will pay a dear price for it. My guess is that they were trying to run roughshod over Tesla and they lost. These demands are not smart for building a long term relationship with the current leader in deployment of driver assist:

  • Raising the price of their product retroactively
  • Demanding an agreement to extremely unfavorable terms of sale and
  • Demanding that Tesla not use data that was collected by its vehicles’ cameras for any purpose other than helping MobilEye develop its products
  • Requiring that Tesla collaborate on Tesla Vision and source future vision processing from them until at least level 4
I think Mobile Eye blew it big time. Can you imagine Daimler agreeing to these terms?
How many car makers does Mobileye do business with? Tesla is a drop in a bucket for them..
 
If Mobileye asked Tesla to discontinue their Vision project it means that Tesla's version of autopilot is not ready and probably still in early stages. If it was almost ready to go there would have been no point asking for it to be discontinued.
 
So this scares me a bit for two reasons:

1. What if MobileEye gets to the point where they simply stop shipping product? What then? And semi-related to that...

2. The move to radar prominence for Auto-Pilot now seems like it was made out of necessity more than just a natural progression of the system. This makes sense when you consider that up until now it's been pretty apparent Tesla was prepping the cars for stereo cameras (Model X housing) and then suddenly drops the bomb that they think radar can be more accurate.

You don't have to read between the lines too much here. This looks like a move made out of necessity and that concerns me because unless Tesla has an alternative camera they are ready to bring to market pretty soon, they run the risk of not having anything to put into their cars.

I feel like for anyone waiting on AP 2.0 to arrive soon, these developments kind of take that off the shelf unless it's something that Tesla was working on independent of MobileEye all along.
 
  • Love
Reactions: Zybd1201
2. The move to radar prominence for Auto-Pilot now seems like it was made out of necessity more than just a natural progression of the system. This makes sense when you consider that up until now it's been pretty apparent Tesla was prepping the cars for stereo cameras (Model X housing) and then suddenly drops the bomb that they think radar can be more accurate.
.

Unfortunately, I agree with this so much I can't stand it. The truth is, nobody in the world thought this radar stuff was possible except that Elon says he felt like it was despite all of his engineers telling him otherwise. Then, all of this stuff happens and BAM, magic phone call and now they can use nothing but radar.

Except even in the autopilot 8.0 it seems to admit you simply can't use only radar and it will rely heavily on geolocation. Well, if you have ever used Tesla's navigation.... you'd be worried too.
 
I didn't reach nearly the same conclusions as the rest on this thread. To me, all things put together:
1. Radar is going to do more. The camera is not doing any less. It isn't like they're switching to radar instead of the camera, or doing any less with the camera.
2. I see nothing in any of the statements that suggests new hardware. There may very well be some, but they're just talking about using the same, existing, hardware with a more advanced software. As in, getting more information out of radar.
3. Even if there couldn't get the MobilEye chip - it doesn't imply "we're going to get 14 cameras and 300 sensors any minute now". At most it suggests they'd have to change that part.

I mean, I'm all for speculation, that's what this site is all about, but those are some serious leaps...