Phantom braking so bad I want to return my car

[mangrove79]

Yes, others have reported that phantom BRAKING is becoming less of an issue of late. Each update seems to have improvements to this, AFAIK.

Just to add another data point. Within 100 miles of where I live, I've only had one location where PB was a sure thing like filing taxes every year. It's the interstate I-75 at one of the exits. It's an overpass with no signs of any kind that could possibly affect AP, but up until Jan, it would slam the brakes from 70 mph to 35 every time without fail! It's interesting the PB only happens on the northbound side. Since Feb 1, the PB went away. The road hasn't changed in any way, so it's probably either the Tesla map got updated or software update fixed something.

 

[j_grange]

Just to add another data point. Within 100 miles of where I live, I've only had one location where PB was a sure thing like filing taxes every year. It's the interstate I-75 at one of the exits. It's an overpass with no signs of any kind that could possibly affect AP, but up until Jan, it would slam the brakes from 70 mph to 35 every time without fail! It's interesting the PB only happens on the northbound side. Since Feb 1, the PB went away. The road hasn't changed in any way, so it's probably either the Tesla map got updated or software update fixed something.

This is kind of my point , we should be reporting to Tesla (if it is not already happening automatically ?) this is how the problem will get fixed. Some government agency will only serve to slow the fixes down.

 

[CyberGus]

The best course of action is to report this incidents to the NHTSA. They will force a recall especially if people are reporting accidents caused by PB. Tesla is like the Russians. Cheat as much as you can until you get caught then blame someone else for it. Couldn’t resist the Olympics tie, sorry…

In Soviet Russia, car reports you

 

[DrGriz]

I have a family member, an engineer who works in the field. He feels that camera is the way to go. Adding radar produces more information but radar is far rougher (blurry images) than optical. Combining the two can produce conflicting information which causes interpretation errors. Sometimes it is better, sometimes worse, but more often it is worse when there are conflicts.

The thing is, programmers can set the threshold for braking based on the input. It's a judgement call and it's not going to be perfect.

Tesla is really safety conscious and has likely set the threshold for avoiding collisions lower than some other systems. First, they are striving for FSD and they want there never to be head ons or pedestrian deaths under that system. In other words, they accept more false positives than false negatives. Maybe if they weren't working toward FSD and were just doing driver assist with the human driver primarily in control they would set the threshold higher.

The problem we all fear is, of course, that hard braking for phantoms will result in other consequences including rear end collisions. If Tesla considers these an acceptable risk they may choose to leave the thresholds lower and the false positive higher. That's where reporting to the NHTSA is worthwhile. It may pressure them to accept more false negatives and keep the human driver in control until they have advanced their programming to a better solution.

In other words, keep EBA driver assist and not FSD until FSD is ready for prime time.

 

[drtimhill]

i think the problem is that tesla had radar capability where this issue was less.

This is purely anecdotal .. I dont think anyone actually knows if removal of radar made PB worse (or better for that matter).

 

[drtimhill]

Here Tesla steps over itself a bit by issuing new games, new UI’s, and so on with no outward transparency into its efforts to ameliorate one problem the cars have had since inception for at least some of us.

There is no way the same teams work on UI, games and AP/TACC . so releasing games or UI changes wont have any impact at all on the timeline for those working on core driving functions that are involved in PB.

 

[j_grange]

There is no way the same teams work on UI, games and AP/TACC . so releasing games or UI changes wont have any impact at all on the timeline for those working on core driving functions that are involved in PB.

Thank you for bringing this up, as a software engineer I always laugh when people try to say that they should just have everyone working on one thing. There is only so many people who to work on a single project.

 

[SalisburySam]

There is no way the same teams work on UI, games and AP/TACC . so releasing games or UI changes wont have any impact at all on the timeline for those working on core driving functions that are involved in PB.

Don’t disagree, but also don’t care. The product comes from a company, the company delivers whatever it delivers, from UI changes to games to taillights to cars to charging adapters to flamethrowers to tequila. How those things happen are of little concern, just that they do…or don’t. Optics are generated by announcements (Elon’s twitter account?), reports of delivered results in the lab (Alpha, Beta testing) and field (in consumers’ hands), and concomitant reviews. Doesn’t matter how many people ARE working on an issue, or how many COULD be working on that problem, or whatever else they are doing, etc. The several posts and threads here explaining what MIGHT be going on at Tesla are thought-provoking, somewhat insightful, and may even be accurate. That said, as a consumer, my experience, anecdotal to be sure but extant for me nonetheless, is that phantom braking is a very real issue, a really hard problem to solve, is LIKELY but not definitively or even exclusively a software problem (since for me it gets better and worse from update to update but doesn’t go away), and no delivered update so far gives me reason to believe it will be solved for my car during my ownership, or maybe even at all. Some posters posit that this is the way not-dumb cruise control works so deal with it, and obviously those of us with PB have had and continue to do so.

 

[Crimpcap]

This is purely anecdotal .. I dont think anyone actually knows if removal of radar made PB worse (or better for that matter).

I can say that my 2021 MY was in service because of PB and was given a 2018 M3 loaner. The 2018 M3 did not have any issues on the same route that my 2021MY is almost undriveable. The 2018 M3 has radar, my 2021 MY doesn't. Confirmed by service center.

 

[Bill.R]

Just to add another data point. Within 100 miles of where I live, I've only had one location where PB was a sure thing like filing taxes every year. It's the interstate I-75 at one of the exits. It's an overpass with no signs of any kind that could possibly affect AP, but up until Jan, it would slam the brakes from 70 mph to 35 every time without fail! It's interesting the PB only happens on the northbound side. Since Feb 1, the PB went away. The road hasn't changed in any way, so it's probably either the Tesla map got updated or software update fixed something.

I was advised that anytime you have a pb instance push the voice command button on the steering wheel and say “bug report“ . Apparently the car collects data and can report back to the fsd team.

 

[Fourdoor]

This is purely anecdotal .. I dont think anyone actually knows if removal of radar made PB worse (or better for that matter).

The problem with a forum discussion is that EVERYTHING is anecdotal. The simple fact is I have the choice to not use cruise control at all, or be used as a beta tester for an advanced cruise control system. There is no third option of using dumb cruise other than the speed limit mode work around.

Tesla as a company doesn't consult with customers on a message board, and neither do other manufacturers of virtually any product. We are not here to generate solutions, we are here to bitch and moan about Tesla not generating a solution when one is so obvious to anyone with a few brain cells to rub together.

Tesla refuses to do a software update to enable dumb cruise control as an option. This is a fact. Now, on the opinion side of things there are all kinds of conspiracy theory rabbit holes you can go down as to why they don't do this update... the most plausible to me being that they don't want to lose beta testers for the TACC system... but that opinion could be way off base.

My TACC worked fantastic when I first got the car, lack of radar didn't cause phantom braking for me. Software updates to make the cameras do more work since radar was no longer being used caused TACC performance to tank hard, and during daylight hours it has slowly been getting incrementally better over time. The original configuration of HW 3.0 AP was to have the cameras do detailed close range analysis while the radar looked far down the road for a "coarse" view of what was down the road. Now the cameras are being used to look further down the road putting much more work load on the camera system and AI analysis of the camera views. Using the camera to look further down the road makes it much harder for the system to tell if an approaching object is in my lane or the oncoming lane. This problem is amplified by a huge amount at night when the cameras only see oncoming headlights.

My main problem with this is a manufacturer releasing a product that doesn't work correctly to the public with the promise that "we are sure that it will work properly someday, if you don't like it don't use it".

Keith

 

[WhiteWi]

The problem with a forum discussion is that EVERYTHING is anecdotal. The simple fact is I have the choice to not use cruise control at all, or be used as a beta tester for an advanced cruise control system. There is no third option of using dumb cruise other than the speed limit mode work around.

Tesla as a company doesn't consult with customers on a message board, and neither do other manufacturers of virtually any product. We are not here to generate solutions, we are here to bitch and moan about Tesla not generating a solution when one is so obvious to anyone with a few brain cells to rub together.

Tesla refuses to do a software update to enable dumb cruise control as an option. This is a fact. Now, on the opinion side of things there are all kinds of conspiracy theory rabbit holes you can go down as to why they don't do this update... the most plausible to me being that they don't want to lose beta testers for the TACC system... but that opinion could be way off base.

My TACC worked fantastic when I first got the car, lack of radar didn't cause phantom braking for me. Software updates to make the cameras do more work since radar was no longer being used caused TACC performance to tank hard, and during daylight hours it has slowly been getting incrementally better over time. The original configuration of HW 3.0 AP was to have the cameras do detailed close range analysis while the radar looked far down the road for a "coarse" view of what was down the road. Now the cameras are being used to look further down the road putting much more work load on the camera system and AI analysis of the camera views. Using the camera to look further down the road makes it much harder for the system to tell if an approaching object is in my lane or the oncoming lane. This problem is amplified by a huge amount at night when the cameras only see oncoming headlights.

My main problem with this is a manufacturer releasing a product that doesn't work correctly to the public with the promise that "we are sure that it will work properly someday, if you don't like it don't use it".

Keith

Autopilot is designed to be used on restricted access highways. The end. Everything else is a bonus and should be used carefully and be ready to take over.

 

[soumak]

This is purely anecdotal .. I dont think anyone actually knows if removal of radar made PB worse (or better for that matter).

Perhaps. But there are more reports to nthsa about phantom breaking once it was switched to vision. So you do have some evidence from that data.

 

[j_grange]

Perhaps. But there are more reports to nthsa about phantom breaking once it was switched to vision. So you do have some evidence from that data.

I here a report about phantom breaking after switching to vision, I report "me too!" it must be the vision...

 

[ScubaBri]

I just spent 6 days driving from Utah to MSP and back. The phantom braking just sucked. Back roads where the worst, autopilot or just cruise control, anytime there was a oncoming big rig at minimum it would just slow down, but the worst was when it almost came to a stop. Interstates were not AS bad, but passing trucks if someone was following to close I was forced to disengaged cruise control because I was always worried that it was going to slam on the brakes and have the person following me slam into the back of me, which I came close a couple times.

 

[drtimhill]

Perhaps. But there are more reports to nthsa about phantom breaking once it was switched to vision. So you do have some evidence from that data.

Suggestive yes, but even then you need to be careful. Tesla have ramped up production a lot, and so it may just be more cars no the road, or a result of a change that was unrelated to the switch to vision, or even just people jumping on the "me too" when they see some people making a report. Correlation is one thing and reasonably easy to establish with a sufficient sample size, but causation is another matter entirely.

 

[soumak]

Suggestive yes, but even then you need to be careful. Tesla have ramped up production a lot, and so it may just be more cars no the road, or a result of a change that was unrelated to the switch to vision, or even just people jumping on the "me too" when they see some people making a report. Correlation is one thing and reasonably easy to establish with a sufficient sample size, but causation is another matter entirely.

True. I’m sure they will get to the bottom of it sooner or later. Either way there is still a problem that needs to be corrected.

 

[karuna]

I have a 2021 Model Y Performance. Phantom Braking was a THING until V11. Now its no longer an issue.

 

[TheLex]

Again I ask - why is Tesla not able to solve the phantom braking issue when even cheap Honda/Toyotas/Subarus don't have this issue. My Subaru Ascent has never phantom braked me. This shouldn't be rocket science at this point. Why is Tesla seemingly the worst at this when Tesla touts itself as being the most advanced automobile company?

 

[sleepydoc]

Autopilot is designed to be used on restricted access highways. The end. Everything else is a bonus and should be used carefully and be ready to take over.

And yet every other car with adaptive cruise that I and everyone I know has driven manages these roads just fine. Try a new excuse.