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

Tesla Turns Off AEB In New Cars Produced Since July

This site may earn commission on affiliate links.
What would you rather have, six weeks without EAB, or your whole car's lifespan on older hardware? I'll take the Tesla approach, thanks.

Cars have long lives. Tesla changes the hardware very very often. Pretty much every Tesla on the road at any particular point is living with older hardware than the most current version. A delay in introducing new hardware doesn't make this dynamic better or worse.

Side Note: In light of this, it always amazes me when folks act like they are going to be hanging on to their current Tesla for decades because its relatively simple mechanics won't wear like an ICE. Sure, the car won't be worn out.... But by Tesla standards the systems will be totally obsolete.

Bonus side note: I suspect that the frequent hardware changes will make it relatively unlikely that older builds will find up getting the benefits of future software updates into the far future. Hard to make software work with varied hardware sets. This doesn't bode well for delivery of FSD features to existing cars if, as I suspect, actual level 5 software is a decade in the future.
 
  • Like
Reactions: Swift and u00mem9
AP 2.0 wasn't really an upgrade, it was "Mobileye is leaving us starting immediately and so we must switch over to our in-house project now, or you get nothing at all."

Hmmm... That's not how they marketed it. They seemed to claim that it just needed validation. And EAP would be validated in a couple months. And some of the non-AP1 features might need regulatory approval. Oh... And get ready for Tesla Network (but don't plan on using FSD with Uber)!

Your description sounds a lot more accurate...
 
AP 2.0 wasn't really an upgrade, it was "Mobileye is leaving us starting immediately and so we must switch over to our in-house project now, or you get nothing at all."

Exactly and this is fraud. It's not my problem that they lost in manufacturing partner. Then they made outlandish claims about what their auto pilot could do despite knowing the fact they couldn't possibly deliver on these claims. Pure and simple fraud.

The main reason I purchased the car was because I thought the auto pilot would be spectacular. If I knew auto pilot wasn't working up to standards I would've avoided the purchase entirely.
 
Regardless of everyone complaining about all of this, at least AP2’s hardware is much better than what MobilEye was able to equip the cars with regarding the time and place software was at.

And yes it’s probably going to be a wild ride and that’s what you try to avoid when getting a $70-$140k vehicle... but we are going to have to accept it, and just WAIT for it to be straightened out. I have faith they will right their wrongs. BUT I don’t have faith it will come in any time period supplied by Elon on his Twitter.

One upside to Tesla marketing the way they do, Autonomous vehicles are going to be a real competitive market within 5 years. I don’t think that leap would have came so quickly if it wasn’t for their outlandish claims and making the “Big” car companies nervous. Audi,VW, BMW, and Mercedes are truly getting somewhere they weren’t anticipating 5 or 6 years ago.
 
  • Like
Reactions: jeffro01
Regardless of everyone complaining about all of this, at least AP2’s hardware is much better than what MobilEye was able to equip the cars with regarding the time and place software was at.

No. Mobileye had additional products and next gen hardware that Tesla declined to apply. Tesla thought they could replicate the capability for the cost of some camera hardware. That has been proven naive.

THEN, the fraud started. Elon decided to misrepresent the reality of lost capability and rebrand it "next gen" hardware capable of full self driving and all the claims of enhanced AP. We now know that decision was made in spite of the experts in his company explaining to him it wasn't true.

A year ago, this forum would roast anyone who implied Elon lied. These days there is a group buy on the off-topic section for "Elon Lies" t-shirts. It's been a tough year.
 
I totally agree with you. Problems can't be avoided and if something doesn't work, you need to recall it and fix it.

IMO Tesla should disable AP completely and then try to really fix it. Shadow braking, following tar lines etc. that's not really acceptable in a feature that is used by untrained drivers. They can still use shadow mode to improve AP. But they need to make sure that it doesn't cause accidents.

And then add in the features that are tested and really work. If some fault is detected again, then take it back again. IMO they did the right thing with recalling AEB, now do the same thing for EAP.

The part in bold implies we are seeing a flood of accidents related to AP and related features. Are we? If so, I must be missing that news.

They’re also all a lot cheaper than a Tesla.

Are you suggesting poor people's lives matter less?
 
Exactly and this is fraud. It's not my problem that they lost in manufacturing partner. Then they made outlandish claims about what their auto pilot could do despite knowing the fact they couldn't possibly deliver on these claims. Pure and simple fraud.

The main reason I purchased the car was because I thought the auto pilot would be spectacular. If I knew auto pilot wasn't working up to standards I would've avoided the purchase entirely.

You are telling me that you would have bought from another manufacturer if Autopilot didn't exist? Okay.... you are one of the very few outliers. I really enjoy my car even without AP.
 
There's a big difference here. Tesla has a very established history of overpromising AP/EAP/FSD features that aren't actually completed, and then rushing the features into cars and using its consumers as beta testers/"validators". Then, when features don't actually work, are dangerous, or just plain aren't delivered, Tesla tinkers with them, disables them, or adds new disclaimers to its rules. Then Tesla (or at least its most ardent defenders on this site) acts like consumers should just be happy that they are receiving anything, since (I) Tesla is on the "bleeding edge," (ii) no one could anticipate that engineering problems would be so difficult to crack, (iii) Tesla is just a small company and can't afford to test things as carefully as the big guys, (iv) Tesla is so important to society that it has to get its products out as quickly as possible, (v) the survival/success of Tesla is so important to society that anything they do to succeed, even borderline fraud, is ultimately for the good of the world, (vi) established levels of testing and safety engineering are too "old school" and unnecessary, or (vii) ELON!!!

Other manufacturers don't do this. They release things that they think work, and don't assume that they can just fix safety-critical systems later if they wind up not working. Sure, sometimes features don't work right and there are recalls. And features (like infotainment systems) that aren't safety critical or related to the operation of the car often have frustratingly bad user interfaces and inexplicably missing functions. But that is very different from what Tesla has done and continues to do.

Note, BTW, that when Tesla has done the kind of recall that other manufacturers do (like when it had the problem with the seats), that didn't generate the level of angst that the AP/EAP/FSD dramas have.

Are you sure about the part in bold? Perhaps I'm cynical but I think every company gets away with what they can. Things like the VW diesel emissions scandal, GM ignition issues, Toyota engine sludge problems, Hyundai's over-stated fuel efficiency numbers etc. all reinforce my view. These were all issues the companies involved in all likelihood knew about for a long time but did not address or acknowledge until they had no other option. They all try to hide problems when they can for as long as they can. This is not unique to Tesla.

I also don't think Tesla is the only company that ships new technology before its perfected. Did you try Mercedes' Drive Pilot? That was much worse than AP/AP2. Do we know other companies' AEB implementations really work any better? We know about Tesla because they have the ability to disable/re-enable/improve features OTA. Other companies do not. There's no real way to know their AEB is any better in real-life situations as it's not something any of can can test. All we know is they all market features like AEB.

In essence, I don't disagree with you that some technology is shipped before its fully baked (as a techie, I would argue that shipping features when they are useful is the right approach, not much later when they are perfect, but that's a different discussion). But I do disagree that this is a Tesla-only issue. I've seen plenty of other examples. If the thought of potential bugs in new technology bothers you, you should stick with tried-and-true technology until the new technology matures. There's no rule that says everyone should adopt emerging technology. No one is forcing you to buy into any particular technology at any particular time.
 
Last edited:
I wonder why this was only implemented on new AP2.5 vehicles ... according to CR :cool:

This update applies only to Tesla vehicles built since Model 3 production began in late July, including all Model 3s, as well as newer Model S sedans and Model X SUVs. That means it will not affect the AEB systems on the vast majority of Teslas on the road today. CR learned about this move from a note posted by a Tesla owner on a Reddit message board; that owner said he received his notice on Saturday
My Model X received in March announced my AEB was no longer functional yesterday (9/12/17).
 
Are you sure about the part in bold? Perhaps I'm cynical but I think every company gets away with what they can. Things like the VW diesel emissions scandal, the GM ignition issues, Toyota engine sludge problems etc. all reinforce my view. They all try to hide problems when they can for as long as they can. This is not unique to Tesla.

I also don't think Tesla is the only company that ships new technology before its perfected. Did you try Mercedes' early Drive Pilot? That was much worse than AP/AP2. Do we know other companies' AEB implementations really work any better? We know about Tesla because they have the ability to disable/re-enable/improve features OTA. Other companies do not. There's no real way to know their AEB is any better in real-life situations as it's not something any of can can test. All we know is they all market features like AEB.

In essence, I don't disagree with you that some technology is shipped before its fully baked (as a techie, I would argue that shipping features when they are useful is the right approach, not much later when they are perfect, but that's a different discussion). But I do disagree that this is a Tesla-only issue. I've seen plenty of other examples. If the thought of potential bugs in new technology bothers you, you should stick with tried-and-true technology until the new technology matures. There's no rule that says everyone should adopt emerging technology. No one is forcing you to buy into any particular technology at any particular time.

Agreed. If companies shipped product before they were 100% ready for prime time with no major issues, we wouldn't have hardly any products. We certainly wouldn't have computers or software. Besides, even thorough testing cannot simply be done ahead of time. You can only hope to get most obvious issues fixed ahead of time. Once again, several people are not thinking logically. Do I want it tested as much as possible? Sure. But do I want to wait 100 years to get it? No. So do reasonable amounts of testing.
 
At ANY point of this cycle I can hit the accelerator and override it. False positives never get past Level 1 in my experience. But they aren't really 'false' they are potential. I've never seen a shadow, railroad Xing, etc, trip Level 1, only cars, bicycles, pedestrians, and plants. Works in reverse too.
]

Does AEB engage when the driver's foot is still on the throttle or only while the car is coasting or braking?

I've had construction cones that are too near the edges of the road trigger AEB in my Volt when driving with cruise control on.
 
-
Can there be better self driving? Does one of them drive with more panache? You can do it more efficiently, with less sensors, but you can't really achieve "better" self driving. Self driving either works, or it doesn't.
Really? I can think of a number of ways where the quality of the experience could vary well within the realm of safety.

As a simple example, with just ACC alone -- not even TACC -- I'd like the ability to control the "aggressiveness" of maintaining speed. A wrote about this years ago (on TMC)* and have yet to see a single car company deliver such functionality.

Between ACC and FSD there is a lot of room for such refinements to be significant enough to allow differentiation in the marketplace.

* Update: Link - Efficiency Miser Mode.
 
Last edited:
Oh man that was painful.

I started reading this thread and I had a couple of points to make about some early comments. Then I figured I'd better read the whole thing to make sure my points weren't already made or refuted or whatever. So for some reason I exhaustively read through this entire mess, despite having important work to do.

Now I'm here and I can't even remember what the points were I was going to make.

But make no mistake. They were good points. It would have shut you all down.

Why am I posting? Because having read through all that BS, I feel entitled. Entitlement seems to be a common theme here, so why not?
 
I totally agree with you. Problems can't be avoided and if something doesn't work, you need to recall it and fix it.

IMO Tesla should disable AP completely and then try to really fix it. Shadow braking, following tar lines etc. that's not really acceptable in a feature that is used by untrained drivers. They can still use shadow mode to improve AP. But they need to make sure that it doesn't cause accidents.

And then add in the features that are tested and really work. If some fault is detected again, then take it back again. IMO they did the right thing with recalling AEB, now do the same thing for EAP.
Actually if the standard is matching other manufacturers, then AP doesn't have any problem that needs a recall. It certainly doesn't qualify as a defect, as certain amounts of erratic behavior is expected in level 2 systems. AP1 set a high bar, but I think everyone is forgetting that the average bar for a level 2 system really isn't that high (given driver is expected to be alert entire time).

Reminder of how the state-of-the-art Mercedes system works (crossing into oncoming traffic and ping-ponging is acceptable behavior for a level 2 system). This is the "fully tested" system you are talking about:
It's important to read manuals to establish official specs and cautions.

However, it is also important to see how those specs perform in real life.

Hands off

"Unfortunately during our test drive there was little sign of an artificial intelligent chauffer to be seen. The E-Class is utterly dependent on clear parallel road markings and signs alike. In the eventual absence of one of the elements, the car required immediate input from the driver.

The same was experienced when the road became too twisty. The E-Class lost its track and wandered towards the side of the road, of worse still, towards the oncoming traffic, alarmingly without any sign of other safety systems interfering. Surprising, considering Mercedes have a reputation for being pioneers of automotive safety.

On stretches of road wider or narrower than standard, for example the fly on or off from a motorway, the system became unsetteled and required two helping hands. When the system did finally work, we were constantly reminded to take control of the steering.

The Mercedes system seems to be a bit like a child learning to ride a bike. The stabilizers are gone, it has the ability to ride itself, it just hasn’t yet developed the confidence. The results is a system that panics and without a helping hand, will eventually plant you in a ditch."

There are clearly yellow lane markings below which visually describes the above Mercedes "towards the oncoming traffic, alarmingly without any sign of other safety systems interfering":

obj.phpi



Here goes the theory of "works if there are zero, nada, zip lane markings"

obj.phpi


The above is not the only article.

The War For Autonomous Driving: 2017 Mercedes-Benz E-Class VS. 2017 Tesla Model S


" It oscillates between lane markings like a drunk driver. "

"It drove like a drunk ten year old, fighting for the wheel with a drunk fourteen year old. It was, in most conditions, dangerous."

Reacting to shadows and following tar lines is also acceptable limitation of a level 2 system, for example see Volvo manual for state-of-the-art Pilot Assist (also "fully tested"):
"In certain situations, it may be difficult for Pilot Assist to assist the driver correctly or to deactivate automatically. In such cases, it is advisable not to use Pilot Assist. Examples of such situations may be:
• the lane's side market lines are missing, badly faded or cross each other.
• the division of lanes is not clear, e.g., when a line divides or merges with another, at exits or if there are many road signs/markers.
there are edges or other lines on or near the lane, e.g., curbs, cracks, repaired areas, sharp shadows, etc.
• the lane is narrow or winding.
• the lane is at the top of a hill, on an uneven road surface or over a bump.
• bad weather conditions (rain, snow, fog, slush, poor visibility, backlighting, etc)."
https://volvornt.harte-hanks.com/manuals/2017/2017-Volvo-S90-Owner-Manual.pdf

Cadillac:
"LKA assistance and/or LDW alerts may occur due to tar marks, shadows, cracks in the road, temporary or construction lane markings, or other road imperfections. This is normal system operation; the vehicle does not need service. Turn LKA off if these conditions continue."
https://my.gm.com/content/dam/gmown...16/Cadillac/Escalade/2k16escalade1stPrint.pdf

Mercedes:
"DISTRONIC PLUS with Steering Assist and Stop&Go Pilot
...The support provided by the system can be impaired if:
...there are no, several or unclear lane markings for a lane, e.g. in areas with road construction work
...there are strong shadows cast on the road"
https://www.mbusa.com/vcm/MB/DigitalAssets/pdfmb/ownersmanual/MY17_S_Class_Sedan_Operator_Manual.pdf

I can draw more examples if necessary, but by your standard, almost all L2 systems on the market would need to be recalled and should not be released.
 
Last edited:
If companies shipped product before they were 100% ready for prime time with no major issues, we wouldn't have hardly any products.... You can only hope to get most obvious issues fixed ahead of time.

This statement is clearly not correct-- at least not with respect to vehicles and other dangerous heavy equipment. Most cars have at most two or three recalls over their lifetimes, and typically those have to do with failures that actually occur in only a tiny fraction of the cars out on the road (but all cars need to be fixed as a prophylactic measure). If cars shipped while there were still "major issues" left to be fixed later, the highways would be a total bloodbath.

This is the danger of importing a web design/software philosophy into a product line that demands traditional industrial/safety engineering and testing.

If your product can (indeed will) get involved in incidents that kill people, you've got to get it right the first time.
 
-
Really? I can think of a number of ways where the quality of the experience could vary well within the realm of safety.

As a simple example, with just ACC alone -- not even TACC -- I'd like the ability to control the "aggressiveness" of maintaining speed. A wrote about this years ago (on TMC)* and have yet to see a single car company deliver such functionality.

Between ACC and FSD there is a lot of room for such refinements to be significant enough to allow differentiation in the marketplace.

* Update: Link - Efficiency Miser Mode.

But would those require a different HW, or just SW adjustments?

Oh man that was painful.

I started reading this thread and I had a couple of points to make about some early comments. Then I figured I'd better read the whole thing to make sure my points weren't already made or refuted or whatever. So for some reason I exhaustively read through this entire mess, despite having important work to do.

Now I'm here and I can't even remember what the points were I was going to make.

But make no mistake. They were good points. It would have shut you all down.

Why am I posting? Because having read through all that BS, I feel entitled. Entitlement seems to be a common theme here, so why not?

Complaining, too. So you fit right in ;-)
 
Does AEB engage when the driver's foot is still on the throttle or only while the car is coasting or braking?

I've had construction cones that are too near the edges of the road trigger AEB in my Volt when driving with cruise control on.

Level 1 on the Volt is the red light HUD. FCW - Forward collision warning. True AEB comes on if you ignore the warning.