I think you're fine - as long as your version is lower than the FSD Beta version, then you'll get in if they offer it to more people with the release. There is no guarantee they will open it further though with this release, so don't get your hopes up too much.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD Beta 10.69
- Thread starter Buckminster
- Start date
planetary
EndlessCheese spreader
Thanks. My hopes for anything could not be any lower.
EVNow
Well-Known Member
I'm too afraid to press accelerator when this happens. I usually disengage - because I don't want it to mistake my accelerator push and run the red light. But its very rare ...
It still struggles to come to a smooth stop at the stop line though - usually slows down some 20 ft before the stop line and slowly moves to the line and stops.
Oddly I don't have that problem. My car is very smooth with slowing for red lights. Many times, especially if there is no lead car, I'll slow down and stop at the light without engaging brakes at all - purely regen. If there is a line of cars behind the light, it's usually a bit rougher and there is some brakes engaged.
diplomat33
Average guy who loves autonomous vehicles
In this case, there was no risk of running the red light. I only pressed the accelerator very gently to push the car passed the railroad crossing and then released the accelerator and FSD Beta came to a stop again at the red light.
Yes, I experience this as well.
Yeah very valid point. At some intersections it misjudges the correct stopline and will in fact go if you push the accelerator, while in other cases it simply stopped too short and can be nudged forward without risk. Wish they rendered the perceived stopline so we could know where it thinks the intersection begins
Has anyone noticed any changes to the 69.2.4 vs the previous version?
Just got mine, so installing it right now...
Just got mine, so installing it right now...
Terminator857
Active Member
It would get more respect if it wasn't over sold. Was there suppose to be a million Tesla Robotaxis in 2019?
Why are they pushing 2 updates simultaneously: 69.2.4 and 69.3? Obviously .3 is only to employees right now. I guess we will be waiting quite a while if they are still pushing an update to the .2 branch. Oh well.
In my area the car gets into turning lanes incorrectly before arrows are visible. This means that being able to see the arrows (at all or better than the way it currently sees them) wouldn’t help if the car makes the decision to change lanes before it sees the arrows. I think bad map data is the culprit in my area. They will need at least near-perfect map data to conquer lane decisions. Think about it. We humans use our memory to drive well in areas we know. We can drive fine in areas we don‘t know, but we make poor lane choices more often due to not knowing the area. We wouldn’t accept this behavior from the passenger seat of a self-driving car.
Without good map data, FSD is always blind to the distant road ahead. The cameras don‘t see nearly far enough to make lane decisions just off of vision. Worse, with bad map data, FSD is told poor information about where it currently is and needs to be Creating a conflict between vision and map data.
All of this could be proven wrong by 69.2.3. Hopefully it is, but I am not optimistic. 80-90% of my disengagements are lane-selection based at this point…
In my area the car gets into turning lanes incorrectly before arrows are visible. This means that being able to see the arrows (at all or better than the way it currently sees them) wouldn’t help if the car makes the decision to change lanes before it sees the arrows. I think bad map data is the culprit in my area. They will need at least near-perfect map data to conquer lane decisions. Think about it. We humans use our memory to drive well in areas we know. We can drive fine in areas we don‘t know, but we make poor lane choices more often due to not knowing the area. We wouldn’t accept this behavior from the passenger seat of a self-driving car.
Without good map data, FSD is always blind to the distant road ahead. The cameras don‘t see nearly far enough to make lane decisions just off of vision. Worse, with bad map data, FSD is told poor information about where it currently is and needs to be Creating a conflict between vision and map data.
All of this could be proven wrong by 69.2.3. Hopefully it is, but I am not optimistic. 80-90% of my disengagements are lane-selection based at this point…
willow_hiller
Well-Known Member
I saw at least one person saying 10.69.2.4 downloaded over LTE for them. Tesla generally reserves that for critical bugs, so maybe they found something worth fixing while the employees validate 10.69.3:
powertoold
Active Member
Elon is the greatest fsd meme in the world. It will be forever remembered when Tesla is the first and only company to achieve generalized fsd for years to come.
The way that Elon thinks is out of this world (or a big fraud to some people). Karpathy said in 2016 when he joined the AP team, there were 2 dudes doing NN training on their desktop computers. In 2016 was the year Tesla released that infamous Model X FSD video, "Driver's seat is only there for legal reasons." Lol!
And once you introduce snow covered roads which dynamically changes lane availability HD maps will be inaccurate.
Just went out for a ride to a close by mall. According to my wife, this version is slightly smoother still. It breaks smoother, accelerates smoother. We usually get 1 or 2 disengagements during this drive, although once in a while, we get zero. Today, both directions were zero.
ALL turns were quite smooth - as good as my own. One section is a 2 lane road with no road markings at all. No problem there either.
In fact, the ride was rather boring, which must be a complement for any self driving technology...
ALL turns were quite smooth - as good as my own. One section is a 2 lane road with no road markings at all. No problem there either.
In fact, the ride was rather boring, which must be a complement for any self driving technology...
ISSUES ISSUES ISSUES
I’ve had nothing but PROBLEMS with 10.69.2.4
Turn right on street X. Sorry but that street is a mile away. I do need to turn on it but there are like six street turns before I even get there. This has happened several times where it thinks I’m no where near where I am.
Turn left of street Y. Which is the correct street. I do need to turn left on it. Guidance and route both show left. But the car turns right. The opposite direction from where I need to go.
NO GPS AVAILABLE. With a red band at top. Where you normally get the flashing flash to interact with car. This time though it’s red. I have NEVER ever had the car say it couldn’t find gps. Clear sky’s, no buildings.
Ugh. Every drive has had these problems. All new for this version for me.
I’ve had nothing but PROBLEMS with 10.69.2.4
Turn right on street X. Sorry but that street is a mile away. I do need to turn on it but there are like six street turns before I even get there. This has happened several times where it thinks I’m no where near where I am.
Turn left of street Y. Which is the correct street. I do need to turn left on it. Guidance and route both show left. But the car turns right. The opposite direction from where I need to go.
NO GPS AVAILABLE. With a red band at top. Where you normally get the flashing flash to interact with car. This time though it’s red. I have NEVER ever had the car say it couldn’t find gps. Clear sky’s, no buildings.
Ugh. Every drive has had these problems. All new for this version for me.
powertoold
Active Member
That's interesting.
During my earlier drive, I did notice that it routed me back the way I normally take, as opposed to the one it usually suggests. I thought that was odd, but in a positive way.
I absolutely felt nothing was wrong with the drive.
Such a big differences in quality of drives!
I'll be driving much more tomorrow, so I'll get a better picture by then.
During my earlier drive, I did notice that it routed me back the way I normally take, as opposed to the one it usually suggests. I thought that was odd, but in a positive way.
I absolutely felt nothing was wrong with the drive.
Such a big differences in quality of drives!
I'll be driving much more tomorrow, so I'll get a better picture by then.
You need to get your GPS issues resolved. Have service check your cabin camera for shielding issues.
There's a couple of youtube complaints about a slightly inaccurate map with this build.
Lidar approaches is having a more powerful compute, high res cameras (8+ mp) rather than a low res (1.2 mp) camera, having more cameras to eliminate blindspot, adding other sensors like high res radar and lidar to improve accuracy...that lidar approach?
What exactly was impossible? Please don't say "vision only" because the very first vision only was Mobileye in 2018 (no radar, lidar or ultrasonics).
You can praise Tesla without turning it into a series of statements that is demonstrably false.
As there are others that ARE doing this. Namely Huawei and Xpeng and others that will join them soon.
Why do you have to turn everything to "Tesla is the greatest"?
Nothing Tesla is doing is unique. Bird's Eye View is industry standard. Occupancy Network is industry stand. Even Xpeng who isn't even a player in AV, you can see their transformer based BEV network here called XNet that uses multi-video input from all cameras to do static recognition, dynamic recognition and motion prediction.
Transformers are the most overused architecture in ML today, I saw a stat saying something to the effect that almost 90-95% of projects today are based it.
I don't understand where you see this uniqueness or greatest in Tesla.
Similar threads
- Article
