Or a bug smear/water spot catches the light just right (there are other bug smears). Might be worth really cleaning the glass in front of the cameras. Not saying that this should happen. But it others have improved PB by doing that.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Phantom Braking
- Thread starter 2101Guy
- Start date
Dont hold your breath. The better cameras might help reduce the number of false positives, but the issue is mostly related to the interpretation of the camera data, not the quality of the camera data itself. If HW4 helps, it will because of better and more sophisticated NNs.
While most of my FSDb experiences have been positive, I've had met car change lanes to avoid an obstacle that wasn't there, but clearly the car "saw" something about 30-40 feet ahead of me. At such a range this is not an issue with camera resolution.
It's already had two service appointments, they say the hardware is all fine.
It's heat mirages. Sometimes all by themselves but mostly combined with an actual vehicle that makes the vehicle appear as if its doubled in size in an instant i.e. is approaching very rapidly.
My phantom braking occurs CES are not heat related, mirage related, or shadow on road related, and occur in chill mode (only one I’ve ever used), in FSDb, in TACC and in the absence of any other car for miles in any direction. Sometimes a hill is involved but will not occur every time I drive that spot, but will also occur on totally flat roads and is not consistent Re location. These are routes I drive frequently.
So, as I said, we drove through Nevada this past weekend and I am reporting back.
We took US95/I-80, which you specifically mentioned, from the Oregon border to Hawthorne. It is both dual lane and divided highway, mostly the former. We returned through southern Nevada on I-15 through Utah back to Idaho.
Most of the time we were using TACC and Autosteer (and of course AEB was active).
We had several, along the way, what I call irrational slowings. They are mild (2-5 mph). These are more like hesitations that don't require interventions. With some, but not all, of those, autosteer would disengage (without intervention) and TACC would resume it's speed on it's own. All without us doing anything (except reengage autosteer, which then was fine). Several happened when we were passing trucks with their flashers on (in the slow lane going up hill, which they are required to do). I don't think it's bad that a car slows when passing a vehicle with it's emergency flashers on!
We did have some hard slowings (PB) that required intervention (hit the accelerator and disengage autosteer and TACC). Two of them happened in fairly quick succession, happened while my wife was driving, and we were on 95 where it is dual lane. So I cleaned the windshield bugs off at the next stop. Microfiber cloth and windshield cleaner spray. No more PBs that day. We did have one late the next day in Southern California, and again I cleaned the windshield bugs in front of the front facing cameras. No more after that the entire trip. I did clean the windshield periodically.
So, as you say, there were mirages when they happened. Once, there was also a raven on the fog line eating something between us and the mirage, and the PB let up when we passed the raven and it flew away, but a long time before we reached the area of the mirage. Another time there was, as you described, a dark vehicle combined with a mirage a quarter of a mile away.
But there were mirages all the time, all day. I counted 15-20 per minute, sometimes several at once. And of course, many of them had vehicles of all colors in them. So I had to ask myself. "Was it the mirages? How do you know when only one in thousands cause PB?"
Anyway, maybe you are right. But I did seem to be able to end the problem by cleaning the windshield, when it happened. I don't really know. I am curious how many you are experiencing in a day of driving that route.
Best.
Last edited:
At a minimum dozens of PBs. I remember one particularly bad trip on a two lane in Colorado that IIRC I stopped counting at around 20 PBs in 2 hours.
As I said, it's not usually just mirages it's mirage+vehicle in the "right" place so I wouldn't really say it's "one in thousands" unless you really think you encountered 3,000+ vehicles adjacent to mirages on your trip which seems like it's probably not accurate.
Also size and placement of the mirages maybe. Nevada has been relatively cool and cloudy that last couple weeks maybe that made a difference.
How fast do you drive? Passing more people probably equals more vehicles ending up in the "right" place.
It's very clearly not just bugs doing it since for me it is always in conjunction with mirages and nearly always also with a vehicle adjacent to the mirage (not really "in" the mirage), but If it's bugs+mirages+vehicle then it's the type of thing that could only be remedied by stopping every ~45 minutes to clean the windshield because the PBs start up at basically the beginning of every single road trip I've been on. I'll try it the next trip I'm on though, literally pull over and clean immediately after the first PB and see if it makes a difference.
And I've never had these sort of problems with the Ford's and Honda's TACCs. It's either Tesla's lack of radar or some problem with their software or both. Or I guess their windshields.
As I said, it's not usually just mirages it's mirage+vehicle in the "right" place so I wouldn't really say it's "one in thousands" unless you really think you encountered 3,000+ vehicles adjacent to mirages on your trip which seems like it's probably not accurate.
Also size and placement of the mirages maybe. Nevada has been relatively cool and cloudy that last couple weeks maybe that made a difference.
How fast do you drive? Passing more people probably equals more vehicles ending up in the "right" place.
It's very clearly not just bugs doing it since for me it is always in conjunction with mirages and nearly always also with a vehicle adjacent to the mirage (not really "in" the mirage), but If it's bugs+mirages+vehicle then it's the type of thing that could only be remedied by stopping every ~45 minutes to clean the windshield because the PBs start up at basically the beginning of every single road trip I've been on. I'll try it the next trip I'm on though, literally pull over and clean immediately after the first PB and see if it makes a difference.
And I've never had these sort of problems with the Ford's and Honda's TACCs. It's either Tesla's lack of radar or some problem with their software or both. Or I guess their windshields.
What I said was that we encountered thousands of mirages, and some of them (I did not try to count, but many) had vehicles in the vicinity or adjacent.
Certainly we did not have "dozens" of PBs, and on the way home it hit 100 degrees, during which, no PBs. We drive the speed limit. We had every weather but snow and fog for periods on the trip. On 95/I80 it was sunny or occasional clouds casting shadows.
I don't doubt your experience. All I am saying is that it wasn't ours. I hope yours gets better, as ours has improved tremendously over the last year.
Other simpler TACCs will be less likely to respond to potential danger threats. Their assumption is "drive ahead unless proven for sure there is a safety threat head on", while something intended eventually for autonomy is the other way around, "everyone else is a potential danger until proven they are operating safely in a way that I expect".
The obvious tradeoff is phantom braking on the second in uncertain cases, though it would be more likely to avoid some accidents that aren't straightforward rear-ending a car in front in the same lane.
Are there a rash of TACC caused crashes in Fords and Hondas?
Is it really accurate to say TACC is "intended eventually for autonomy"? It's not FSD. I just want cruise control that works, I don't want to participate in Musk's delusions regarding full autonomy.
Other brand TCAA relies quite heavily on radar, and that doesn't get "tricked" by mirages like the vision only Teslas do. Radar however gets "tricked" by stationary objects.
Your video is a perfect demonstration of the worst aspect of Tesla TCAA at the moment. Maybe the new unified stack of FSD that drives on the highway will react better than the AP that everyone else has, and when that unified stack becomes an update for everyone, your Tesla will be better. But these are big IFs, and none of the FSD beta tester videos I've seen ever test them out in the open desert roads.
dbldwn02
Active Member
Since losing radar, mine does pretty much the same thing with nothing in front of it. It initially applies the brakes and then just continues to slow down as if it's still thinking about if there's something in the road or not....just like what yours did.
If I'm following traffic it does ok. I know the spots where it's going to happen on my normal commute to work. It's when I go somewhere new, that it becomes more of an issue. Always hovering my foot over the accelerator when I should be hovering it over the brake. Over the past 6 years my car has re-written the code in my head to "IF car braking: THEN press accelerator"
On the other hand, my wife's Ioniq5 is almost flawless when it comes to smoothness of TACC. (HDA2) It catches slowed/stopped traffic much faster too, instead of slamming on brakes last second.
Similar threads
