Phantom Braking

[drtimhill]

I‘m not so sure Tesla could fix PB overnight. Several months back I experimented with disabling all AEB and safety features for a week. PB events were decreased but definitely not eliminated.

You may not be able to disable it .. but of course Tesla could disable all non-driver brake applications very easily.

 

[hgmichna]

I remember that someone uttered a suspicion that a lower following distance setting leads to more phantom brakings.

Probably nonsense, but still … I have the distance set to the maximum of 7, and I have few phantom brakings.

• Does anybody who complains about frequent phantom brakings have it set to 7?
• Does anybody who experiences few phantom brakings have his setting at 2, 3, or 4?

These would be counter-examples invalidating the strange hypothesis. Just to make sure.

 

[DrGriz]

I remember that someone uttered a suspicion that a lower following distance setting leads to more phantom brakings.

Probably nonsense, but still … I have the distance set to the maximum of 7, and I have few phantom brakings.

• Does anybody who complains about frequent phantom brakings have it set to 7?
• Does anybody who experiences few phantom brakings have his setting at 2, 3, or 4?

These would be counter-examples invalidating the strange hypothesis. Just to make sure.

Interesting thought. We have ours set to 7, which is rational in our rural setting. PB is effectively gone for us. But we had it set to 7 when we were having frequent events earlier in the year as well.

 

[sleepydoc]

I remember that someone uttered a suspicion that a lower following distance setting leads to more phantom brakings.

Probably nonsense, but still … I have the distance set to the maximum of 7, and I have few phantom brakings.

• Does anybody who complains about frequent phantom brakings have it set to 7?
• Does anybody who experiences few phantom brakings have his setting at 2, 3, or 4?

These would be counter-examples invalidating the strange hypothesis. Just to make sure.

Except I would get phantom braking events on completely empty roads.

You may not be able to disable it .. but of course Tesla could disable all non-driver brake applications very easily.

true - I thought of that at the time. I can only disable what I have access to.

The problem is none of us knows what’s actually causing PB. Is it an overly sensitive AEB system? Bad GPS? Bad visual processing? Maybe just a bug in the code. To really figure it out you need access to what events the system is seeing and what’s triggering the braking; a level of detail none of us have.

 

[CyberGus]

It's pretty frustrating and definitely keeps the adrenal gland exercised.

Too bad they don't have a marketing department. "Tesla: We Get Your Heart Pumping"™

 

[hgmichna]

The problem is none of us knows what’s actually causing PB. Is it an overly sensitive AEB system? Bad GPS? Bad visual processing? Maybe just a bug in the code. To really figure it out you need access to what events the system is seeing and what’s triggering the braking; a level of detail none of us have.

My best guess has always been that TACC and the autopilot are overly cautious, according to Tesla's motto, "Do not crash!"

Consider that every car with automatic braking has phantom braking, just like every human. Recognition of objects and predicting their paths is never perfect, neither in computers nor in brains.

Some will say, Tesla has too many, and I'm sure we'll see gradual improvement, but they will never go away entirely.

Also it seems that a Tesla will not brake so hard that this causes a rear-end crash with the following car. I think Teslas try to minimize the total crash damage. I bet, if a crash between a car in front and a car behind is unavoidable, the Tesla will optimize the crash such that front and rear crash energies are about equal. This means that phantom braking is unlikely to cause a rear-end crash. The Tesla will just brake less when there is another car close behind. Evidence for this is the total absence of such crash reports.

 

[sleepydoc]

My best guess has always been that TACC and the autopilot are overly cautious, according to Tesla's motto, "Do not crash!"

Consider that every car with automatic braking has phantom braking, just like every human. Recognition of objects and predicting their paths is never perfect, neither in computers nor in brains.

Some will say, Tesla has too many, and I'm sure we'll see gradual improvement, but they will never go away entirely.

I've seen many people try to rationalize and justify Tesla's problem by pointing out that auto maker X has reports. First, this is the logical equivalent of "but Billy did it too!" More important is the rate of occurrences. I have yet to experience phantom braking in any other make of car. I have yet to talk to another non-Tesla owner who's experienced it. I have yet to talk to a Tesla owner who has not experienced it. @drtimhill and @CyberGus will say I'm taking a biased sample or jumping to conclusions, but the odds of of this being a fluke are incredibly small.

It's also hard to compare - most/all of the non-Tesla reports I've seen have been AEB activations (slamming on the brakes, alarms blaring) Tesla has those but then also has innumerable other, milder slowdowns on TACC.

Also it seems that a Tesla will not brake so hard that this causes a rear-end crash with the following car. I think Teslas try to minimize the total crash damage. I bet, if a crash between a car in front and a car behind is unavoidable, the Tesla will optimize the crash such that front and rear crash energies are about equal. This means that phantom braking is unlikely to cause a rear-end crash. The Tesla will just brake less when there is another car close behind. Evidence for this is the total absence of such crash reports.

I've seen some people reporting/hypothesizing this. No idea if it's true or not. If it is true then it begs the question - If the computer says it doesn't need to slow down as much because there's a car behind it, why was the car slowing down in the first place?

 

[Knightshade]

I've seen many people try to rationalize and justify Tesla's problem by pointing out that auto maker X has reports. First, this is the logical equivalent of "but Billy did it too!"

No, it's really not.

It's the actual equivalent of pointing out an inherent shortcoming in this type of system regardless of whose brand of car it's in.

More important is the rate of occurrences. I have yet to experience phantom braking in any other make of car. I have yet to talk to another non-Tesla owner who's experienced it.

[

And yet every time you tell this anecdote it's pointed out you've been linked to forums of every single brand you insist NEVER has the issue, with many many posts from owners of all those brands, reporting they have the issue.

Then you magically pretend you never saw it later and repeat the same nonsense claims.

 

[CyberGus]

@drtimhill and @CyberGus will say I'm taking a biased sample or jumping to conclusions, but the odds of of this being a fluke are incredibly small.

Nearly 4 out of 3 people fail to understand statistics

 

[clydeiii]

I remember that someone uttered a suspicion that a lower following distance setting leads to more phantom brakings.

Probably nonsense, but still … I have the distance set to the maximum of 7, and I have few phantom brakings.

• Does anybody who complains about frequent phantom brakings have it set to 7?
• Does anybody who experiences few phantom brakings have his setting at 2, 3, or 4?

These would be counter-examples invalidating the strange hypothesis. Just to make sure.

I'm at 7 (it's just the safest and who cares if people cut you off?) and hardly ever have PBing.

 

[Dewg]

I'm at 7 (it's just the safest and who cares if people cut you off?) and hardly ever have PBing.

Totally anecdotal, but I'm at 7 too with few PBs. We should test it and reduce our distance to 3 or 4 and see if PBs change in frequency.

 

[Prairie]

Phantom braking. Hwy 2 Nebraska. Two lanes, well marked sides and center line. Flat. No hills, diverging lanes, bridges, etc. straight road. 4 times in first 305 miles and 3 times in returning 305 miles in FSD beta (most recent update 2 days prior yo trip) and plain cruise control. Overall i like the new FSD ( combined stacks) but the phantom braking was a big problem. On straight ,clear, no oncoming traffic ot rear traffic went from 65 mph to 5 mph extremely rapidly.

 

[enemji]

Phantom braking. Hwy 2 Nebraska. Two lanes, well marked sides and center line. Flat. No hills, diverging lanes, bridges, etc. straight road. 4 times in first 305 miles and 3 times in returning 305 miles in FSD beta (most recent update 2 days prior yo trip) and plain cruise control. Overall i like the new FSD ( combined stacks) but the phantom braking was a big problem. On straight ,clear, no oncoming traffic ot rear traffic went from 65 mph to 5 mph extremely rapidly.

So it was a low to zero density traffic?

 

[Prairie]

Zero traffic except me in either direction straight flat 2 lane well marked

 

[enemji]

Zero traffic except me in either direction straight flat 2 lane well marked

That is helpful and kind of emphasizing the theory that the AP/TACC is designed for city traffic areas where there are plenty of vehicles to reference to as it moves ahead.

 

[Bad Horse]

Phantom braking. Hwy 2 Nebraska. Two lanes, well marked sides and center line. Flat. No hills, diverging lanes, bridges, etc. straight road. 4 times in first 305 miles and 3 times in returning 305 miles in FSD beta (most recent update 2 days prior yo trip) and plain cruise control. Overall i like the new FSD ( combined stacks) but the phantom braking was a big problem. On straight ,clear, no oncoming traffic ot rear traffic went from 65 mph to 5 mph extremely rapidly.

Happened to me more times than I can count this last trip down to Williston and back. It improves in the rain, cause it can't "see" too far ahead to try and apply the brakes ones it sees a car on the next hill over

 

[sleepydoc]

Flat

You already said you were in Nebraska - no need to repeat yourself!

@enemji ’s theory may be correct. I’ve noticed a dramatic improvement in PB but I drove mostly in the suburbs where there are more reference objects.

 

[DrGriz]

You already said you were in Nebraska - no need to repeat yourself!

@enemji ’s theory may be correct. I’ve noticed a dramatic improvement in PB but I drove mostly in the suburbs where there are more reference objects.

We live in a very rural place. 50 miles between gas stations once you get on the road. The extremely rare "PB" we experience is more of a hesitancy, lifting off of the gas pedal when a car is turning in front of us or we are cresting a hill. More like an overcautious new driver than phantom braking. The slam on the brakes events are a long ways back in the rear view mirror.

 

[sleepydoc]

I dont think many consider the slowing down the car does when a car in front of you starts to turn right/move over to the right turn only lane in front of you. Thats appropriate braking, but it is a bit excessively aggressive braking. Its like its not aware that the car that is slowing is actually going to turn away from you and as such, it initiates very aggressive braking. I wouldnt say thats phantom.
The phantom braking is what I get when Im driving on a interstate on a sunny day, no cars in front or behind or beside me, and I approach an area where trees are casting random shadows onto the roadway and the speed goes from 65mph in the car, down to 50mph rather quickly because FSD (still) is thinking the shadows are actual objects on the road.

I wonder if thats the same section of code that sometimes doesnt see the wall in my garage and attempts to put the car into drive vs reverse with the auto gear selection

I don’t consider that PB. When I talk about PB I mean randomly slowing anywhere from 5-50 MPH for no reason at all, generally while driving straight on a highway.

 

[DrGriz]

I dont think many consider the slowing down the car does when a car in front of you starts to turn right/move over to the right turn only lane in front of you. Thats appropriate braking, but it is a bit excessively aggressive braking. Its like its not aware that the car that is slowing is actually going to turn away from you and as such, it initiates very aggressive braking. I wouldnt say thats phantom.
The phantom braking is what I get when Im driving on a interstate on a sunny day, no cars in front or behind or beside me, and I approach an area where trees are casting random shadows onto the roadway and the speed goes from 65mph in the car, down to 50mph rather quickly because FSD (still) is thinking the shadows are actual objects on the road.

I wonder if thats the same section of code that sometimes doesnt see the wall in my garage and attempts to put the car into drive vs reverse with the auto gear selection

Oh, I agree. That's why I said that for me, the stuff (actual PB) you are experiencing is a year or more in the rear view mirror.

It is interesting that individual drivers/cars have such diverse experiences. One would think that they would be similar, given that they are supposed to be the same.