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

FSD - 10.69.25.2 - Phantom Breaking Problems

This site may earn commission on affiliate links.
Hi,

I have a 2022 Model S Long Range with FSD 10.69.25.2. I jsut spent 3 days driving from Boulder CO down to Austin TX for this weeks Investor Meeting. Things went well the first day - drove from Boulder to Santa Fe NM - really strong headwinds 70mph+ so kept the speed down maxing at 70mph and had zero FSD disengagments.

On day 2 drove from Santa Fe NM to Abilene TX - aprox 500 miles I must have had 8-9 Phantom Breaking incidents - which was not good. It was mostly duel lane highway - no freeways. Most of the time I had a 75mph speed limit and I was setting the FSD speed to 79-80mph. All the phantom breaking incidents occured at around that speed - and were severe enough to cause the tryes to squeel. In all cases there was no one close behind - but in once case I had just overtaken a large truck - so was fortunate to be far enough in front to avoid causing him to break.

After that drive - I did some googling and saw one suggestion which was to tunr off "emergency breaking" in the Autopilot settings. I did that for the final drive today from Abilene down to Austin - and only had 1 incident where the car slowed by more than 15mph. Interestingly I did have a few cases when I was driving over 75mph - where the car did slow down for no aparent reason - but it was only 3-5mph slowdown - so no where as severe as the previous days incidents.

My current theory is that this occurs > 75mph and somehow the "emergency breaking" gets activated which causes the severe slowdown.
 
Just drove back to Colorado. Had several more Phantom Breaking events occur - particularly driving from Lubbock to Amarillo TX. I had emergency breaking switched off - so that theory did not work out. Pretty much all occured between 75-80mph. Not much traffic arround - so not any issues.
 
I’m calling BS on the squealing tires.

Here are some samples I captured on my way back.
The first is 13mph in around 1.1sec, a solid 0.53g (average - the peak would be a bit higher).

The last one the brake got hit by the driver, which explains the higher reduction in speed and also the slightly greater acceleration (I think it might have been as high as 0.6-0.7g for that one but I'm too lazy to count frames).

You can see the brake pedal move significantly in all of these events on its own volition. Kudos for an excellent video.

That is solid! If it is 0.7g with poor tires and questionable surface you might hear a chirp, and it's hard to assess the actual peak numbers here. Especially if you tap the brake as well. But overall I think you'd just hear the tires "grip" a little. Tires make noise when they're working well!

Definitely some solid decel. Well done Tesla. Driver alertness is important, and this will get the job done.

2 out of 4 were associated with tire treads. Kudos to Tesla for picking up on those. One was associated with a complicated sign - it may have wanted to have a little bit more time to read it. The other seemed to have nothing in particular which might have caused it.
 
Last edited:
This is the best phantom braking video I've seen. Really well done.
This, to me, is the biggest problem with FSD beta and is the reason that I've stopped using it on surface roads. Seems to be less of an issue on the highway.
It sucks for sure and it's just as predictable as that. There's no reason for any driver to apply brakes in these scenarios but yet FSD all too frequently applies needless stiff braking. Speed plays a roll although I experience it at 45mph in city driving.

Obviously there's cause and effect. These appear to be deterministic false alarms likely indicating general poor design. Duh! More specifically poor processing of images, insufficient time to achieve high confidence of object threat, inadequate image sensors, ...

FSD continues to woefully lack good system engineering, integration engineering, quality engineering, ...
 
Last edited:
Here are some samples I captured on my way back.
Seems ike you car is telling you to slow down. The difference is that if you applied the brakes yourself, you hands would be on the steering wheel, AND you foot on the brake peddles. Because of those extra supports, you wouldn't even notice it.
I had the privilege of driving an ICE Hyundai Kona. It is so light weight that every time the brakes are used they snatched and I felt like being tossed out of my seat. Go drive a small ICE car for a trip, and you will realize that your T is way smoother
 
  • Funny
Reactions: AlanSubie4Life
I now have FSD 11.3.6 Installed. Just did a test 300m test drive up the 25 into Wyoming to see if Phantom Breaking issues have been resolved. The good news was no Phantom Breaking incidents like I experienced on my last trip - but I did see 3 minor slowdowns all occur in similiar situaitons. That was going over a small crest of a hill at around 80mph - saw a minor slowdown - by 2-3 mph - and in each case - with zero traffic in any direction the car did a lane change from the right hand lane to the left hand lane - in a dual lane road. Identical in all 3 cases.
 
I now have FSD 11.3.6 Installed. Just did a test 300m test drive up the 25 into Wyoming to see if Phantom Breaking issues have been resolved. The good news was no Phantom Breaking incidents like I experienced on my last trip - but I did see 3 minor slowdowns all occur in similiar situaitons. That was going over a small crest of a hill at around 80mph - saw a minor slowdown - by 2-3 mph - and in each case - with zero traffic in any direction the car did a lane change from the right hand lane to the left hand lane - in a dual lane road. Identical in all 3 cases.
Try enabling Minimize Lane Changes, see if that helps.
 
I now have FSD 11.3.6 Installed. Just did a test 300m test drive up the 25 into Wyoming to see if Phantom Breaking issues have been resolved. The good news was no Phantom Breaking incidents like I experienced on my last trip - but I did see 3 minor slowdowns all occur in similiar situaitons. That was going over a small crest of a hill at around 80mph - saw a minor slowdown - by 2-3 mph - and in each case - with zero traffic in any direction the car did a lane change from the right hand lane to the left hand lane - in a dual lane road. Identical in all 3 cases.
I also have FSD 11.3.6 installed. I have a 6 mile drive into town on 2-lane roads with low traffic. On every trip, I can depend on at least one hard braking event caused by safe, normal traffic in on-coming traffic lane. The car brakes HARD. I then touch the brakes myself to get out of FSD, and then use Tesla's new system to report these events when it asks "why did you get out of FSD?". So I report these events at the rate of once or twice on the way into town, and once or twice on the way home.

I have not seen many other comments about phantom braking from on-coming traffic -- though it isn't actually "phantom" braking -- that is, it is not misinterpreting a shadow in the oncoming lane as a car or truck -- the vehicle in the oncoming lane is definitely there -- but the FSD is misinterpreting the cars/trucks as a threat when they are just doing their ordinary, normal thing.

Very frustrating, dangerous because of possibility of being rear-ended, and a source of considerable angst for my wife.

Bill
 
Some people would make an excuse for an undesirable slowdown as if it's not a "phantom" braking.

An undesirable slowdown is still undesirable and that needs to be fixed.
I think you missed the part where he says

"Very frustrating, dangerous because of possibility of being rear-ended, and a source of considerable angst for my wife."

He was making no attempt to downplay the significance of the braking. He was suggesting that it wasn't a true "phantom" because he knew the source.
 
  • Like
Reactions: Robipad