Major Phantom Braking Issues

[derotam]

Your persistence on seeing more data and calling this a situation-specific issue that requires case by case analysis sure makes it seems like you are trying to redirect the conversation from "consumers have a reasonable expectation of predictable behavior from their cars" to "what happened to you will not necessarily happen to the next person."

I would say you should do some research on my forum past and the official Tesla forum archives but I don't think you care about research.

There are lots of engineers and like minded people on these forums that like to understand how the car actually works, how it accelerates, how it decelerates, how it makes decisions, how it accounts for every milliwatt of power used and gained, etc etc etc... Just because we are around doesn't mean we work for Tesla and just the idea that a Tesla software engineer would be using these forums to conduct their job as you suggest is laughable.

Now, if you want to just whine and complain about something you don't like, real or not, then go ahead. There are lots of chorus's of users about all kinds of things on these forums that you can join in on.

I have a reasonable expectation that bluetooth should work flawlessly on every device in concert with the car...but it doesn't.....lots of chorus's about that in the past and there was lots of in depth technical discussion about the bluetooth standard and how it gets implemented in all sorts of devices.

Lots of people complain about PB, ok fine...but in an attempt to get a better understanding of the phenomenon people need data... First, it the issue fully random or not. If it is fully random it is a lot harder but not impossible to gather data about it. It is not random though as people have said that it happens all the time in repeatable places....but no one hits the dashcam to save the video!

Ahh PB events are not the same and they have to be looked at individually in order to start attempting to sort them into similar scenario buckets.

People go back and forth about the safety issue of it as well, and I agree there is a safety issue aspect to PB incidents, but again, there are lots of complaints, but not corresponding reports of ACTUAL crashes caused by PB issues. And everyone knows that the news would be ALL OVER Tesla if there were lots of people actually saying that PB CAUSED them to be rear ended. So where are all these news reports? Maybe I missed them.

 

[derotam]

Doesn't Tesla already have access to your vehicle data, and aren't they already aware of virtually all cases of phantom breaks (software reduces speed, person presses the gas pedal)?

Tesla doesn't collect all data from every vehicle all the time, that would be a massive amount of data. They can put in a flag for the car to send data based on a set of criteria. They do that all the time to get data to feed in to the NN training. Are they doing that for this issue, I don't know cause...(checking my work badge...), nope I still don't work for Tesla.

 

[Dewg]

I'd like to add the following to the discussion:

What are people doing from a diagnostic standpoint to resolve/mitigate the issue?

Have you opened a service ticket?
Have you reinstalled the software (via the service menu or have service do it for you)
Have you recalibrated the cameras?
Etc.

There have been people who've reported significant improvement after doing some diagnostic actions like those above.

 

[Far1]

I would say you should do some research on my forum past and the official Tesla forum archives but I don't think you care about research.

There are lots of engineers and like minded people on these forums that like to understand how the car actually works, how it accelerates, how it decelerates, how it makes decisions, how it accounts for every milliwatt of power used and gained, etc etc etc... Just because we are around doesn't mean we work for Tesla and just the idea that a Tesla software engineer would be using these forums to conduct their job as you suggest is laughable.

Now, if you want to just whine and complain about something you don't like, real or not, then go ahead. There are lots of chorus's of users about all kinds of things on these forums that you can join in on.

I have a reasonable expectation that bluetooth should work flawlessly on every device in concert with the car...but it doesn't.....lots of chorus's about that in the past and there was lots of in depth technical discussion about the bluetooth standard and how it gets implemented in all sorts of devices.

Lots of people complain about PB, ok fine...but in an attempt to get a better understanding of the phenomenon people need data... First, it the issue fully random or not. If it is fully random it is a lot harder but not impossible to gather data about it. It is not random though as people have said that it happens all the time in repeatable places....but no one hits the dashcam to save the video!

Ahh PB events are not the same and they have to be looked at individually in order to start attempting to sort them into similar scenario buckets.

People go back and forth about the safety issue of it as well, and I agree there is a safety issue aspect to PB incidents, but again, there are lots of complaints, but not corresponding reports of ACTUAL crashes caused by PB issues. And everyone knows that the news would be ALL OVER Tesla if there were lots of people actually saying that PB CAUSED them to be rear ended. So where are all these news reports? Maybe I missed them.

"I would say you should do some research...but I don't think you care about research."

"Now, if you want to just whine and complain about something you don't like, real or not,..."

"Lots of people complain about PB, ok fine..."

When you say things like the above, people stop taking you seriously. Good day to you kind sir.

 

[derotam]

"I would say you should do some research...but I don't think you care about research."

"Now, if you want to just whine and complain about something you don't like, real or not,..."

"Lots of people complain about PB, ok fine..."

When you say things like the above, people stop taking you seriously. Good day to you kind sir.

So you don't want to have a technical discussion about Phantom Braking then? You just want to take statements out of context to try and make someone look bad? Good day to you as well!

 

[stopcrazypp]

Your persistence on seeing more data and calling this a situation-specific issue that requires case by case analysis sure makes it seems like you are trying to redirect the conversation from "consumers have a reasonable expectation of predictable behavior from their cars" to "what happened to you will not necessarily happen to the next person."

Except it IS a case-by-case analysis given people's definition of phantom braking vary widely. It can range from a full on hard emergency stop (akin to AEB) to a minor 5mph slowdown. It can range from having no obvious triggers (like a clearly marked empty road in daylight with no mirages), to having triggers the driver themselves already know (shadows, overpasses, tunnels, mirages, merging traffic, oncoming traffic, semi trucks, speed limit changes, curve approaches, etc).

Phantom braking has existed even with radar active (and even in the earliest AP1), it also has existed with other brands (even if they don't call it that term). Some have seen improvements, some have seen it worse. It also is different if you use TACC, Autosteer, or FSD Beta. It also depends on location (people in CA Bay Area seem to have it the least).

Just screaming phantom braking doesn't really tell people the severity, the conditions it happens, or help track how it has evolved through updates.

As for the extent of Tesla's acknowledgement, they slapped a warning in the manual (like most manufacturers do):
"Traffic-Aware Cruise Control may occasionally cause Model 3 to brake when not required or when you are not expecting it. This can be caused by closely following a vehicle ahead, detecting vehicles or objects in adjacent lanes (especially on curves), etc."

Model 3 Owner's Manual | Tesla

Learn about Traffic-Aware Cruise Control and its specifications.

www.tesla.com

The thing that might get Tesla to change are the NHTSA reports (which you can also submit yourself) which might lead to a recall, but only if the incidents rise to a safety issue (like for example causing an actual accident), but so far this is rare or non-existent. NHTSA seems more concerned about the car NOT reacting (like for example to stopped emergency vehicles) and also if Tesla is doing enough nagging of the driver to pay attention. They are not at all concerned with the user experience (it can clearly be argued the increased nags throughout the years have made for worst user experience).

 

[stopcrazypp]

Doesn't Tesla already have access to your vehicle data, and aren't they already aware of virtually all cases of phantom breaks (software reduces speed, person presses the gas pedal)?

No, not necessarily (even assuming you opted in). The system many times would not know it would be considered phantom braking (especially if driver doesn't disengage) and there is no evidence that they are tracking it in this way. They are able to set triggers to send telemetry and NN data (and sometimes short snippets of video) to the cloud for certain things they are interested in, but there have not been suggestion that they do so for PB.

https://twitter.com/x/status/1096322821934993408

Tesla just released a video on their data collection and it's much more limited than you are suggesting:

Tesla Reveals Data Privacy Details: Which Data Is Used and What Remains Private

Tesla's data privacy measures set a high standard for the industry, with a proactive approach that goes beyond industry standards. Customers are empowered with

www.notateslaapp.com

 

[Far1]

Tesla just released a video on their data collection and it's much more limited than you are suggesting:

Tesla Reveals Data Privacy Details: Which Data Is Used and What Remains Private

Tesla's data privacy measures set a high standard for the industry, with a proactive approach that goes beyond industry standards. Customers are empowered with

www.notateslaapp.com

The following is not relevant to the topic of this thread...

But, in the video you linked, it says "you can request a copy of your data". Believe me as someone that tried, that is a lie. I did make a request, and it was promptly denied - Tesla said they only provide vehicle data if they are obligated to by law. See attached. So I wouldn't trust the rest of what it says either.

 

[Far1]

The thing that might get Tesla to change are the NHTSA reports (which you can also submit yourself) which might lead to a recall, but only if the incidents rise to a safety issue (like for example causing an actual accident), but so far this is rare or non-existent.

Why would anyone voluntarily go to a government agency and say "that rear-end accident, that was my fault, not the car behind me", and end up having their insurance rates increased?

 

[stopcrazypp]

Why would anyone voluntarily go to a government agency and say "that rear-end accident, that was my fault, not the car behind me", and end up having their insurance rates increased?

They would, if they believe it was the car's fault. Also people can report it even if the car behind was found at fault.

There have been successful recalls on AEB done via the NHTSA process (just some results from a quick search):

Through The Years Honda Has Had Its Problems, Now It Is Recalling 1.7M Vehicles For Auto Braking Problems

Through the years, Honda has had more than its share of problems. And, while they have remedied some of them, there are still others that have hit, like a 1.7 million vehicle recall of CR-Vs and Accords.

www.torquenews.com

Volkswagen recalls 246,000 Atlas SUVs due to unexpected braking

Volkswagen is recalling more than 246,000 SUVs in the US and Canada because faulty wiring harnesses can make them brake unexpectedly, sometimes while in traffic.

nypost.com

Automatic Emergency Braking Problems Lead to Mazda3 Recall Impacting 35,000 Vehicles - AboutLawsuits.com

A recall by Mazda warns that some vehicles' automatic braking system may engage unexpectedly, increasing the risk of an auto accident.

www.aboutlawsuits.com

Nissan was able to escape a full recall, only doing a TSB (even though they had known triggers of railroad tracks and overpasses):
Nissan Admits to More Than 1,400 Reports About Faulty Automatic Emergency Braking but Continues to Refuse Recall

This is very anecdotal, but the gist get I get is hard emergency braking down to a stop or close is very rare in Teslas, usually it's slowdowns that cars behind can still stop for (or driver can easily override with accelerator pedal). That's why just screaming "phantom braking" is not very helpful to describe the severity of the issue.

The only case I'm aware of was the most recent case on the Bay Bridge in San Francisco where driver claimed it was on FSD Beta, and even looking at the footage, the car seemed to take some time to come to a stop (which if the cars behind maintained proper following distance likely would have stopped in time).

 

[Far1]

They would, if they believe it was the car's fault.

Not very likely, because it would cost them money. Your car phantom breaks and the car behind you hits you, what incentive do you have to say it was a phantom break, therefore admit fault, get a ticket, put the claim on your own insurance, and raise your own rates?

Also people can report it even if the car behind was found at fault.

They can but they have a very strong financial incentive not to.

 

[stopcrazypp]

Not very likely, because it would cost them money. Your car phantom breaks and the car behind you hits you, what incentive do you have to say it was a phantom break, therefore admit fault, get a ticket, put the claim on your own insurance, and raise your own rates?

Well, enough people did it to get the linked recalls to go through. Also nothing says the report to the NHTSA has to be linked to your police report at all. You can do the report to the NHTSA regardless of what you say to the police. You also do not have to admit fault to do the NHTSA report. NHTSA also redacts the last 4 digits of your VIN, so the police can't tie a report directly to you either. I certainly have not heard of cases of police finding a person at fault for an accident due to a NHTSA report they made.

They can but they have a very strong financial incentive not to.

If the car behind was already found at fault (and presumably paid out already), what financial incentive would there be to not report? I wasn't able to find cases where recalls (even ones already issued before the accident) result in a denial of insurance payout, much less reversal of a previous pay out. In this case, there isn't even a recall yet, so not seeing how an insurance company can reverse a payout.

 

[avramd]

It is the following vehicles responsibility to keep a safe distance and be able to brake safely.

That's sort of a closet falsehood hidden inside another truth. Yes, it is technically true that it is the following driver's responsibility to avoid a collision. However, it is not acceptable to take an action that would likely result in a collision on the grounds that you have the right of way.

There is an underlying right-of-way law in most states that covers this: everyone is responsible for attempting to avoid all collisions regardless of who has right of way. If you try and fail, you have done your duty. But you can not take deliberate action, or deliberately avoid taking reasonable action, to cause a collision.

Given that the solution is astronomically simple, I think what it boils down to is that we are, by and large, taking responsibility on Tesla's behalf, either by not using TACC in the first place, or by adapting to it. I think all it would take is two major collisions in which somebody either sued Tesla, or was able to demonstrate to NHTSA that the car caused a collision that was otherwise trivially avoidable, before Tesla would "fix" it by giving us dumb cruise control.

 

[avramd]

Static cruise control is actually worse in every possible way, but we like it because we have adapted to using it, learning it quirks and cons and using it wisely.

All other things held equal, predictable is better than unpredictable in every possible way.

You are certainly entitled to your opinions on static cruise control, but they are just that - opinions.

The reality is much simpler: Static cruise control is a different thing than dynamic cruise control. Your statement is like saying "a wrench is better than a hammer in every possible way." No, it is not. It is better at what it is good at, and worse at what a hammer is good at.

 

[derotam]

Not very likely, because it would cost them money. Your car phantom breaks and the car behind you hits you, what incentive do you have to say it was a phantom break, therefore admit fault, get a ticket, put the claim on your own insurance, and raise your own rates?


They can but they have a very strong financial incentive not to.

The other thing is that in this extremely litigious country(U.S.) lots of people would want to try and sue the "golden goose"(Tesla) to try and get a huge payout...on both sides of the accident.

I will agree that on both sides of a rear end collision caused by PB there would be challenges to initiate some claim against Tesla but that wouldn't negate the minimal effort it would take to get the media involved.

The media(and others) who are rabidly against Tesla would also have a major incentive to look into this and go try to "ambulance chase" rear end collisions involving Tesla's to attempt to show(whether based in any reasonableness or not) that PB cause the collision.

 

[OncomingStorm]

on the insurance thing, i live in a no fault state, you better believe if PB caused me to get rear ended i would raise hell over it

 

[Far1]

The other thing is that in this extremely litigious country(U.S.) lots of people would want to try and sue the "golden goose"(Tesla) to try and get a huge payout...on both sides of the accident.

I will agree that on both sides of a rear end collision caused by PB there would be challenges to initiate some claim against Tesla but that wouldn't negate the minimal effort it would take to get the media involved.

The media(and others) who are rabidly against Tesla would also have a major incentive to look into this and go try to "ambulance chase" rear end collisions involving Tesla's to attempt to show(whether based in any reasonableness or not) that PB cause the collision.

I completely disagree.

You get into a rear-end, police comes, they find the rear vehicle at fault, their insurance pays, end of story.

Admitting fault in an accident is a huge liability and financial risk, and if you say your car acted erratically (which you'll need to for the media to get involved), you're admitting fault.

Admitting fault AND suing Tesla, a company with lawyers on payroll, is not something a regular person with limited time and resources would do - especially when there's a high likelihood that Tesla would say "we showed an alert to always be ready to take over." Add that Tesla has all your data and you only maybe have dashcam.

Plus you likely can't sue Tesla, because you likely didn't opt out of arbitration when you purchased your vehicle.

 

[KArnold]

You get into a rear-end, police comes, they find the rear vehicle at fault, their insurance pays, end of story.

Hypothetical. Suppose a Tesla get's rear ended and local LE is called. Now add the Tesla driver saying "I was using Tesla AutoPilot and it just slammed on the brakes for no reason", how would the officer respond?

1. Irrelevant. Following vehicle still was following too closely. OR
2. Maybe I need to tell my boss and let 2 billion other people comment on whether this is worthy of further consideration before writing this citation.

Maybe an easy case for an ambulance chaser given the big name? Then again, my experience is limited to late night TV.

 

[lowtek]

Hypothetical. Suppose a Tesla get's rear ended and local LE is called. Now add the Tesla driver saying "I was using Tesla AutoPilot and it just slammed on the brakes for no reason", how would the officer respond?

1. Irrelevant. Following vehicle still was following too closely. OR
2. Maybe I need to tell my boss and let 2 billion other people comment on whether this is worthy of further consideration before writing this citation.

Maybe an easy case for an ambulance chaser given the big name? Then again, my experience is limited to late night TV.

No. 1 for sure ... back the frak off, 1 car length per 10mph is the generally accepted rule of thumb. So if you're doing 70 be *at least* 7 car lengths away, plenty of time to react. Also, get your eyes off your phone and on the road.

As a Tesla driver, you agree that you are 100% responsible for the car's actions when using autopilot or fsd. If you don't want that responsibility, stop using those features.

It's the same with any car with traffic aware cruise control.

Just a footnote, I haven't had any phantom braking with my 2021 Model S and only twice with the previous Model S, and always with bridges crossing the highway. My wife has #4xx Model Y and hasn't had any phantom braking in a long time, so somewhere along the line they fixed it.

However, with the last FSD beta, before they pulled it, my wife's Y would get what I would categorize as a "brake bump" occasionally, nothing like what phantom braking used to be. It used to slam on the brakes decelerating from 70 to almost stopped in seconds, scary *sugar*.

 

[zoomer0056]

slammed on the brakes for no reason

There is always a reason.