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

FSD Beta 10.69

This site may earn commission on affiliate links.
It's really too bad that people were mashing the Report button so much that Tesla removed it from the vast majority of the fleet.

@ACPark and myself are experiencing a horrible flaw that needs to be reported; having a traffic light being shown as flashing red and green at the same time isn't ideal. Not having a way to report it is even less so.

If only there were a forum for FSDb "testers" (I mean, are we *really* testers at this point, or just gatherers of data?) to make Tesla aware of issues like this that they otherwise would never know existed.

I still stand by my recent exuberance on how much progress has been made, but I am left wondering how much *more* progress has been left on the table due to how poorly run this whole "beta" has been; IMHO at this point in time (December 2, 2022) we could be much further down the road (please forgive the pun) toward an actual release product if Tesla had been more professional in their running of the entire beta program.

Not only have they not opened a forum for Beta testers to discuss issues, educate other testers, inform Tesla of issues, etc., they have now closed off the only avenue we had of reporting potentially serious problems.

The hubris Tesla has shown by taking the "we know what's wrong, and we alone will be in charge of collecting the data we need" attitude has not served them well, and will not continue to do so in the future. I've been involved in many beta programs over the years, from both the development (I was involved in developing an in-house software maintenance tracking program to prevent our planes from sitting on the ground on January 1, 2000) and user side, and I am NOT impressed.

I have been trying to get into the "pole position" at lights I know this is a problem at when the light is red to see if the car would run it or not, but haven't been successful yet. Honestly, I'm not sure I really should do this, as there isn't a point to it; I can't report the findings to anyone other than the people on this forum. Tesla has shown they don't care, so why should I?

TL;DR: Great progress has been made. Greater progress *could* have been made if Tesla's hubris and unprofessional running of this program hadn't hobbled it.
 
@ACPark and myself are experiencing a horrible flaw that needs to be reported; having a traffic light being shown as flashing red and green at the same time isn't ideal. Not having a way to report it is even less so.

As long as you disengage Tesla might eventually find it. At this point they have low hundreds of thousands of vehicles operating the software so manual reports would be overwhelming. They'll just do campaigns or whatever.

Just be sure to disengage or apply accelerator whenever FSD does anything wrong (this is what I do when I'm not filming). They'll figure it out. Eventually.

So it's very important to correct every mistake. As few miles (or yards) as possible between interventions is the goal.

We definitely want to correct all the incorrect behaviors and presumably the vast data set will help - as long as we allow it to.

The red light running is a normal thing and has been there for a while (videos posted of course). Just have to pay extremely close attention. FSD Beta is not meant to be relaxing or helpful, at the current time!
 
Just have to pay extremely close attention.
I'm still using FSDb the way I always have, as follows:

I "phantom drive" the entire time FSDb is engaged. This means that I am constantly applying torque to the steering wheel as I would as if I were actually driving the car. And I do this the ENTIRE time that FSDb is engaged: it's like having two drivers driving the car at the same time. As such, I never get "please apply slight torque to the wheel" messages, nor do I ever get the "please pay attention to the road" camera nag, either.

While I do have some leniency in the way FSDb picks its path (in other words, I let it get away with a path I wouldn't have picked to a certain degree), using FSDb in this manner allows for very smooth disengagements, as the torque I'm applying to the wheel gradually gets larger and larger to the point that FSDb disengages when the situation develops to the point that I'm not gonna let FSDb continue with what it wants to do.

This results in me very intently driving the car at all times, even more so than when FSDb is not in use. I'm going to continue to use FSD in this manner until the steering wheel is removed from the car. :)
 
It's really too bad that people were mashing the Report button so much that Tesla removed it from the vast majority of the fleet.

As long as you disengage Tesla might eventually find it
That’s my assumption - that Tesla uses brake pedal & steering wheel disengagements to trigger reports.
 
This results in me very intently driving the car at all times, even more so than when FSDb is not in use
I find that even with all the interventions and attentiveness I still drive a lot worse with FSDb on.

I drive in people’s blind spots, I don’t move position in my lane when others are off-center in adjacent lanes, etc.

There is a constant nagging feeling that I should just disengage and drive myself.
 
I don't know, but in my mind, tracking disengagements would result in a lot more reports to the mothership than what they receive in the way of Report button pushes.

A lot of guys here on the forum have stated that they always disengage prior to entering a situation that they know FSDb won't handle well. In other words, prior to where Tesla actually needs the data: the area that it doesn't handle well. If all of these prior-to-the-situation disengagements are reported back, Tesla is left with bupkiss as far as good data is concerned. I must admit, I'm in this group... I will *always* disengage before I get into a situation I know FSDb isn't going to work well in.

I guess that leaves the onus on us... we need to go ahead and let FSDb clunk its way into these scenarios and then disengage when FSDb forces us to.
 
  • Like
Reactions: FSDtester#1
I find that even with all the interventions and attentiveness I still drive a lot worse with FSDb on.

I drive in people’s blind spots, I don’t move position in my lane when others are off-center in adjacent lanes, etc.

There is a constant nagging feeling that I should just disengage and drive myself.
Completely agree.

But to be honest, I feel that way whenever I'm in a car that I'm not driving. I always end up with less tooth enamel at the end of a drive when my wife was driving.
 
I don't know, but in my mind, tracking disengagements would result in a lot more reports to the mothership than what they receive in the way of Report button pushes.

A lot of guys here on the forum have stated that they always disengage prior to entering a situation that they know FSDb won't handle well. In other words, prior to where Tesla actually needs the data: the area that it doesn't handle well. If all of these prior-to-the-situation disengagements are reported back, Tesla is left with bupkiss as far as good data is concerned. I must admit, I'm in this group... I will *always* disengage before I get into a situation I know FSDb isn't going to work well in.

I guess that leaves the onus on us... we need to go ahead and let FSDb clunk its way into these scenarios and then disengage when FSDb forces us to.
True, but if they are consistently getting disengagements at the same location every time they know it’s an issue and they should have at least some cases to look at where it didn‘t disengage.

There’s a stretch near my house where FSD always gets the lane selection wrong in 2 or 3 locations. First it selects a right turn lane coming up to a light rather than the straight lane, then it will try to merge over to the shoulder (separated by a solid white line) then it will select another right turn lane. Overall, lane selection has gotten better but this are has persisted. I’ve reported it dozens of times and now I just disengage. Tesla clearly has the data; they just need to act on it.

Edit - as a single driver it’s easy to get frustrated over areas like this but I try to remember that Tesla is trying to solve FSD for the entire country so my guess and assumption is that they look at failures heuristically rather than from an individual failure perspective. It would be easy for them to program an exception based on specific GPS data but then they’d have a hopelessly complicated system and need to reprogram it every time a new road was built. I’m pretty sure they’re trying to figure out why the car is getting it wrong to change the underlying algorithm, which also results in the 2 steps forward, one step back dance we see. As they tweak the algorithm to fix one set of problems another problem crops up.
 
Last edited:
Actually that is a myth. Drains are far too small to be effected by the Coriolis effect and can drain in either direction depending of their construction (toilets too).

Also I say topato.

That myth only applies south of the equator, you know.........which is in the northern half of the world half of the time..........not affected by daylight savings...........
 
  • Funny
Reactions: FSDtester#1
I had an interesting observation yesterday morning - we had a snow storm and I was driving to work 'the old fashioned way.' The roads were covered in snow and not only were the lane markings not visible, for much of the drive there were not even tire tracks visible. I glanced down at the road visualizations a few times and not only did the computer do a pretty good job of estimating the edge of the road, it actually had lane markings in places. To be clear, there was nothing visible; I just drove based on where the edge of the road was, likely straddling the center line half the time, then slowing down and moving closer to the edge when an oncoming car approached.

Based on this it appears that FSD must use some map information combined with the cameras to generate the visualizations.
Here FSDb has done a good job of faking the road markings on snow-obscured roads. Even when I myself have a hard time guessing where the road edge/center is. That said I have tried actually engaging FSDb under those conditions and it disengages rather quickly. Every time.
 
The cameras also see just beyond human visual spectrum, and can compensate for glare. There have been many times, driving into the sun, that my human eyes lost the lanes - they were just gone, lost in a sea of glare and mirage reflections, but the car saw them perfectly. I'm not saying it saw through the snow, but if the snow was thin enough, they may have picked up the markings through the glare that we can't see.
The glare thing also seems to work well with my cameras, or does it? Is it just faking it like road markings in the snow, making assumptions? Here in Montana, driving east on the interstate when the sun comes over the horizon, flat land, no trees, even in the summer it's literally impossible to see pretty much anything ahead of you. I never drive those hours on the interstate and also try to avoid the back highways after dark, you know, when at risk of hitting a flock of flying deer escalates. With the glare the cameras seem to do a better job than my eyes, but I never know if it's accurate or just faking it, like it obviously does in the snow. Even without glare the cameras still suck at spotting and dealing with dead skunks in the middle of the road and such, you know, stinking to high-high heaven...........Loudon.......
 
The glare thing also seems to work well with my cameras, or does it? Is it just faking it like road markings in the snow, making assumptions? Here in Montana, driving east on the interstate when the sun comes over the horizon, flat land, no trees, even in the summer it's literally impossible to see pretty much anything ahead of you. I never drive those hours on the interstate and also try to avoid the back highways after dark, you know, when at risk of hitting a flock of flying deer escalates. With the glare the cameras seem to do a better job than my eyes, but I never know if it's accurate or just faking it, like it obviously does in the snow. Even without glare the cameras still suck at spotting and dealing with dead skunks in the middle of the road and such, you know, stinking to high-high heaven...........Loudon.......
I should conduct an experiment here, drive a local dirt, public, mapped road, through a flat field, in the snow, and see if is guessing solely from the map or other........
 
  • Like
Reactions: FSDtester#1
So I used the AP on the highway on the way home today, and this huge pile of sugar gave me a strike lol, I don’t even know what I did wrong. Just got a random large warning to grab the yoke and steer manually.

I should’ve reported it, oh wait the report button isn’t there anymore.

I guess the AP strike affects the FSD strike count haha.

View attachment 880659
Dude, I had 4 of those today just going up to Hernando from Orlando on the turn pike. Bitch just started screaming at me for no reason. I swiftly reminded her that’s why I left my first wife.

4 strikes for me