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.
If they can solve lane selection, 95% of my drives would be zero disengagement. Even though they are supposed to be reading arrows now, it seems like the car still just ignores them…
I wonder if the same lane selection behavior is used for the new "auto-cancel turn signal" feature. Auto-cancel will deactivate signals if you change lanes, but it will remain activated if the car "thinks" you're in a turn lane and still need to complete the turn. I've noticed this behavior breaks sometimes, like on some dual right-turn lanes, which maybe reflects the same limitations that FSD has.
 
  • Like
Reactions: FSDtester#1
Makes one wonder if anyone at TSLA really tests any of this stuff.
Seems like they need to keep working on their simulator/model, and adjust the test program to continue to look for a whole host of possible regressions. Getting good test coverage is tough.

Practically this is not something that can be done in the real world, so the simulation environment has to work really well, and properly capture all the intricacies and special cases of the real world.

I imagine it is an ongoing serious effort to improve this. It’s a difficult and very compute-intensive problem!
 
  • Like
Reactions: FSDtester#1
haven't seen many (any?) new car being added to the pending list.
Both 44.30.5 and 44.25.3 seem to have just been installs from the existing pending count.
With 44.30.5 getting mostly existing FSD cars and 25.3 the majority of the non-FSD cars.
Maybe that will change, but we all know by now that despite some sources claiming victory, the reality is the typically opaque Tesla.
 
Curious, anybody here "stuck" on 10.69.25? As in you go to the Software settings screen and it doesn't find 2022.44.30.5 / FSD Beta 10.69.25.1 available? I wonder why Tesla hasn't pushed the current version if people are experiencing improvements, but it seems to be available if you pull for an update.

And just checking, release notes for 2022.44.25.5 still show Early Access Program? (Re-check today as it seems to be remotely hosted.)
Our 3 was stuck at a 10.69.25 build (I don't remember the exact one) for quite a while. It was for so long that I almost posted about it. Our Y was keeping up with the updates. The 3 finally got the current 10.69 build a couple of days ago. I did end up checking on the software version in the car and doing a reboot to see if I could help it along. None of the things I did result in an immediate effect. The fact that it did eventually get an upgrade may or may not have been triggered by the actions I took.
 
I wonder if the same lane selection behavior is used for the new "auto-cancel turn signal" feature. Auto-cancel will deactivate signals if you change lanes, but it will remain activated if the car "thinks" you're in a turn lane and still need to complete the turn. I've noticed this behavior breaks sometimes, like on some dual right-turn lanes, which maybe reflects the same limitations that FSD has.
I was thinking similar thoughts. Although, cars without FSD beta have this feature, which may point to it using something different. Who knows.
 
Last night, after sunset, the car did something it had never done before. On a two lane well marked road that I have driven often if failed to make a left turn. This is a rural road not much traffic, there is a T intersection that the car has negotiated from both directions day and night many times but last night there was a car at a stop at the intersection and as I came abreast of the intersection, with the stopped car even with the driver side car door, my car came to a complete stop instead of continuing the turn. I disengaged FSD and made the turn without any problem. The only thing different about this intersection is that it is slightly down hill from the road I was on and that caused the headlights from the car stop to my left to shine a little higher on the left side of my car. I did not look at the monitor to see what the car was seeing so I don't know why the FSD stopped forward movement. The weather was clear no rain just dark on a asphalt well marked road.
 
  • Informative
Reactions: FSDtester#1
Not yet but I have a 150 mile highway drive tomorrow so I'll know after that.
Has anyone else noticed the new build 10.69.25.1 works WAY better on highway stack than 10.689.25. It doesn't drift over nearly as far when the lines widen for merge lanes. Its more confident in exit ramps / on ramps. It almost feels like they might be pushing more of the FSD logic into the EAP stack..

Anyone else???

@MindBlend
Well drove 150+ miles yesterday mostly highway and found very little difference on 25.1.
I knew as soon as I used the on ramp there would be no magic today. A major problem for me is when merging into 25mph-40mph rush hour traffic behind a line of cars entering the highway. As the cars in front of me match the rush hour traffic and merge into traffic FSD sees empty ramp space and increases speed passing rush hour traffic. That is stupid and what typically happens is FSD begins to run out of ramp space and would have to do a nasty merge. Knowing what FSD is doing I almost always have to disengage. Merging with highway traffic at higher speeds has worked "reasonably" well for a long time.

My other major highway problem is when FSD wants to move into a passing lane with a car fast approaching continues. I just cancel the move.
I'm hoping V11 will address these types of issues. With these 2 problems resolved my highway driving would be 99% intervention/disengagement free.
My 10 miles of city/street driving was surprisingly good.
 
  • Like
Reactions: VanFriscia
Last night, after sunset, the car did something it had never done before. On a two lane well marked road that I have driven often if failed to make a left turn. This is a rural road not much traffic, there is a T intersection that the car has negotiated from both directions day and night many times but last night there was a car at a stop at the intersection and as I came abreast of the intersection, with the stopped car even with the driver side car door, my car came to a complete stop instead of continuing the turn. I disengaged FSD and made the turn without any problem. The only thing different about this intersection is that it is slightly down hill from the road I was on and that caused the headlights from the car stop to my left to shine a little higher on the left side of my car. I did not look at the monitor to see what the car was seeing so I don't know why the FSD stopped forward movement. The weather was clear no rain just dark on a asphalt well marked road.
It might be the planner isn't ready. Sometimes it seems to be asleep at the switch.
 
@MindBlend
Well drove 150+ miles yesterday mostly highway and found very little difference on 25.1.
I knew as soon as I used the on ramp there would be no magic today. A major problem for me is when merging into 25mph-40mph rush hour traffic behind a line of cars entering the highway. As the cars in front of me match the rush hour traffic and merge into traffic FSD sees empty ramp space and increases speed passing rush hour traffic. That is stupid and what typically happens is FSD begins to run out of ramp space and would have to do a nasty merge. Knowing what FSD is doing I almost always have to disengage. Merging with highway traffic at higher speeds has worked "reasonably" well for a long time.

My other major highway problem is when FSD wants to move into a passing lane with a car fast approaching continues. I just cancel the move.
I'm hoping V11 will address these types of issues. With these 2 problems resolved my highway driving would be 99% intervention/disengagement free.
My 10 miles of city/street driving was surprisingly good.
I did a 150 mile drive yesterday on NoA, and 25.1 sucked.

The speed based lane changes were way too early compared to previous versions. Like, not “I think they changed something.” They definitely changed something. It wanted to change lanes so damn early.

And it wanted to get out of the passing lane too early, too. It would turn the blinker on before I had even cleared 18-wheelers every time. It didn’t change lanes until it was safe to do so, but I’m sure it made the 18-wheeler drivers uncomfortable with this silly Tesla signaling it wanted to cut them off.

It also did a poor job figuring out what speed to go when changing lanes. Before, it used to be anticipatory which was perfect. If I was passing someone going slower, it’d speed up perfectly. On 25.1, it would often consider the wrong lane’s speed when it didn’t impact us at all (and not in the “lane going slow so we’re slowing down” way).

Other changes I noticed on 25.1 include:

  • Steering wheel feedback felt a bit mushy
  • Driving sounds changed (in Drive and Reverse)
 
@MindBlend
Well drove 150+ miles yesterday mostly highway and found very little difference on 25.1.
I knew as soon as I used the on ramp there would be no magic today. A major problem for me is when merging into 25mph-40mph rush hour traffic behind a line of cars entering the highway. As the cars in front of me match the rush hour traffic and merge into traffic FSD sees empty ramp space and increases speed passing rush hour traffic. That is stupid and what typically happens is FSD begins to run out of ramp space and would have to do a nasty merge. Knowing what FSD is doing I almost always have to disengage. Merging with highway traffic at higher speeds has worked "reasonably" well for a long time.

My other major highway problem is when FSD wants to move into a passing lane with a car fast approaching continues. I just cancel the move.
I'm hoping V11 will address these types of issues. With these 2 problems resolved my highway driving would be 99% intervention/disengagement free.
My 10 miles of city/street driving was surprisingly good.
I agree 100% with your assessment. I'd add one more gripe to your list though - NoAP does a lousy job with traffic merging from an on-ramp into its lane. It never speeds up to accommodate this traffic, nor does it temporarily change lanes to make room. What it DOES do is wait until the last minute and then slam on the brakes. Drives me nuts. I intervene/disengage pretty much every time.
 
  • Like
Reactions: old pilot
I agree 100% with your assessment. I'd add one more gripe to your list though - NoAP does a lousy job with traffic merging from an on-ramp into its lane. It never speeds up to accommodate this traffic, nor does it temporarily change lanes to make room. What it DOES do is wait until the last minute and then slam on the brakes. Drives me nuts. I intervene/disengage pretty much every time.
If a merge is coming, can't you use the turn signal to change lanes to the left prior to the merge, then change back after the merge?
 
  • Like
Reactions: VanFriscia
If a merge is coming, can't you use the turn signal to change lanes to the left prior to the merge, then change back after the merge?
Sure can, and that's most often what I do. It's an admittedly painless intervention on my part, but NoAP should still be able to do that on its own nonetheless. FSD should surely be able to do this, eventually.
 
I did a 150 mile drive yesterday on NoA, and 25.1 sucked.

The speed based lane changes were way too early compared to previous versions. Like, not “I think they changed something.” They definitely changed something. It wanted to change lanes so damn early.
Thats why I sometimes use "chill" mode on the highway. That does seem to help but then it sometimes takes too long, but I'd rather that problem since I can always initiate a lane change. Despite a couple of issues FSD/NoA on the highway was very helpful especially since it was light rain in the dark.
 
  • Like
Reactions: DarkForest
Sure can, and that's most often what I do. It's an admittedly painless intervention on my part, but NoAP should still be able to do that on its own nonetheless. FSD should surely be able to do this, eventually.
I think it's the mindset - I treat it as an L2 system - a driver assist. I frequently, while on Advanced Driver Assist, force lane changes well before the system would initiate it, just because I know what's coming up. The system is assisting me, but I'm in control. It helps reduce frustration with the system. If you think of it as an L3+ system, then I could see it being more frustrating.
 
I think it's the mindset - I treat it as an L2 system - a driver assist. I frequently, while on Advanced Driver Assist, force lane changes well before the system would initiate it, just because I know what's coming up. The system is assisting me, but I'm in control. It helps reduce frustration with the system. If you think of it as an L3+ system, then I could see it being more frustrating.

That’s the right strategy and mindset for sure, but unfortunately there is just too much intervention required at the moment to make it worth using (3 interventions per mile in a simple suburban type drive with excellent markings and simple traffic lights, with light traffic).

But when they sort out the major persistent problems with brake use and lack of acceleration and lack of smoothness, this will be the right idea.

The lane changes which are unnecessary are annoying, and biasing it towards not making them or just having the driver request most of them would be great. Maybe something to look forward to where you can just do it yourself most of the time, having it only initiate for actual turns (and missing the turn if the driver has not positioned the car in the correct approach lane to allow a simple turn lane entry - basically identical to NOA behavior - that would be awesome since then the driver would be basically in complete control of all the important decisions).

2023 hopefully will be the year of good Tesla decisions where they start giving customers what they want. That is the really great thing about competition - will make everything so much better and Tesla has an advantage in their culture and approach - as long as they are forced to make changes.
 
Last edited:
That’s the right strategy and mindset for sure, but unfortunately there is just too much intervention required at the moment to make it worth using (3 interventions per mile in a simple suburban type drive with excellent markings and simple traffic lights, with light traffic).
If I had 3 interventions per mile I'd think the same but I don't. My guess is you intervene far more than most drivers in the same situation. Just a guess though :)