FSD Beta 10.69

[Supcom]

It's actually waiting for solid torque on the steering wheel. If you keep constant pressure on the wheel, it will merge much faster.

I drive with constant torque on the wheel and rarely see a nag. Once in a while, the car will flash blue on the screen when wanting to change lanes and will not do so until I apply a little extra torque. But I have never seen my car change lanes automatically in less than five seconds after the turn signal is activated. I've got many thousands of miles on NOA and the car is very consistent about this.

FSDb, on the other hand, can be a weaving idiot when it wants to. Sometimes it seems like it hits the blinker as it turns the wheels. It's like Dr. Jekyll and Mr. Hyde between the two stacks.

 

[JHCCAZ]

Just making sure, you went to your vehicle and checked the Software settings screen and the last checked time was recent / within the last 24 hours? This relaxed pace also shows up in TeslaFi stats too where the pending number for 2022.44.30.5 has been slowly dropping from around 400+ on January 1st to currently around 200, but since the 1st, there's been nearly 1400 updates to this version. I believe this reflects people manually checking for updates and installing it whereas the "pending" count is mostly vehicles that were pushed the update notification (and a smaller portion for people who manually checked and found the update but didn't complete installation yet).

Thanks for the information; it helps in proper interpretation of the Teslascope /TeslaFi data.

My car has since received the 30.5 update without any manual checking or intervention on my part, and this seems consistent with the relaxed-pace conclusion. I cannot say whether I could have forced it earlier, only that it showed up within the time span being seen fleetwide.

 

[powertoold]

Anyone have thoughts on what's going on here:

I've noticed with .69 in general that there is a 2-3 second delay in making a new decision once an incorrect one is made, e.g.:

1) Light turns green, lead car starts moving, ego car appropriately follows closely, but lead car, for some reason, brakes briefly and then proceeds again. Ego car then takes 2 seconds or so before committing to follow the lead car, leaving a much larger gap than previously

2) Unprotected right: ego car creeps up, kinda commits to go but decides to stop for oncoming car, once oncoming car passes, ego car takes 2-3s longer to commit than if it had just committed on the initial decision

 

[DarkForest]

Care to expound? I've not noticed any different sounds.

The audio the car plays when it’s in drive or reverse.

The reverse audio is much more prominent to me then the drive audio. Months ago, I wasn’t 100% sure if it was audio played through the sound system or something about the mechanics of the vehicle.

Then one of the software updates messed up the audio system in my M3, and I temporarily lost all audio. Music wouldn’t play, the turn signals wouldn’t make a sound, and the drive/reverse sounds were gone! So then I knew it was something they were playing. A restart fixed it, of course.

Anyway, it’s something I pay attention to every time I drive the car. It’s much more obvious when I’m in my garage. With 25.1, the audio it plays changed.

 

[Ramphex]

Got another angry TAKE OVER NOW NOW NOW DING DING DING the other day when on a flat straight road. Thought that would be my 4th strike, but it wasn’t, FSD just had a random TIA.

 

[GranularHail]

"2 weeks" to v11.3....

https://twitter.com/x/status/1611787267340926976

 

[Fiver]

Translation: Expect it around Valentines Day, or maybe realistically St Paddy's day.

 

[FSDtester#1]

"2 weeks" to v11.3....

https://twitter.com/x/status/1611787267340926976

So April according to my Elon time conversion

 

[nvx1977]

Looks like V11.3 will be ready for wide release in about 2 weeks. Many major improvements.

Translation: Expect it around Valentines Day, or maybe realistically St Paddy's day.

So April according to my Elon time conversion

with barely noticeable improvements.

 

[aronth5]

with barely noticeable improvements.

I'm going to go out on a limb and say this time there will be noticeable improvements. Of course also a bunch of regressions but everyone expects that as well. I know I'm probably being foolish,

 

[2101Guy]

"2 weeks" to v11.3....

https://twitter.com/x/status/1611787267340926976

is this the one that gets LA to NYC with no human input and/or robotaxi capabilities? Just checking in on status of those promises

 

[mjwojtukiewicz]

Downloaded it yesterday. JUST AWFUL. I was at a storage locker facility. Punch in address to store. Double clicked for FSD. Car turned when left and right about 4 times. Pulled out, go to stop light waiting on making a left. Light turns green. Car won't move.............DOA

 

[kabin]

Anyone have thoughts on what's going on here:

I've noticed with .69 in general that there is a 2-3 second delay in making a new decision once an incorrect one is made, e.g.:

1) Light turns green, lead car starts moving, ego car appropriately follows closely, but lead car, for some reason, brakes briefly and then proceeds again. Ego car then takes 2 seconds or so before committing to follow the lead car, leaving a much larger gap than previously

2) Unprotected right: ego car creeps up, kinda commits to go but decides to stop for oncoming car, once oncoming car passes, ego car takes 2-3s longer to commit than if it had just committed on the initial decision

My WAG - is the software requirements overgrew current hardware capabilities. If so it might be related to frequently loading various NNs and support code for different scenarios. Once those NNs and support code are loaded they probably require fresh data buffering, data processing, and time to achieve a level of confidence before making decisions.

Sometimes it feels like FSDb is out to lunch, sees a different scenario, flips a coin, and takes a chance knowing the driver is responsible. It works for an accepting customer base but the time is coming for real advancements or the jig will be up and FSD's price will need to drop.

Craptastic roadway use software releases would've never passed muster 20 years ago.

 

[EVNow]

So April according to my Elon time conversion

Late March (so, Q1) or early April ( so can be talked about in Q1 ER).

 

[mjwojtukiewicz]

Sure you did.

Sure I did what?

 

[jebinc]

Got another angry TAKE OVER NOW NOW NOW DING DING DING the other day when on a flat straight road. Thought that would be my 4th strike, but it wasn’t, FSD just had a random TIA.

Junk

 

[vamike]

Late March (so, Q1) or early April ( so can be talked about in Q1 ER).

March makes sense, in fact I expect the typical timeline:

• 2 Weeks from Now: "Surprised" TMC posters begin commenting 11.3 is late (the "2 week" jokes are exhausted, Elon's reliability revisited, etc.)
• February: early non-employee tester rollout, followed by critical comments of "shill's posting youtube videos"
• March: wide rollout begins
• March wide rollout + 3 hours: concerns emerge about install getting stuck at 50%
• March wide rollout + 1 to 2 days: the usual posters comment that 11.3 is worse than previous version

 

[BobbyTables]

March 2024: 11.x highway fsd is close enough to autopilot that people are not so annoyed that they keep switching profiles to the one with fsd disabled.

And a bonus
January 2024: new fsd computer, cameras rolled out. Everyone hopes their cars will be upgraded, but it obviously isn’t possible due to the computer’s cooling requirements and the size and quantity of the sensors.

 

[bhakan]

I have been driving the past week (12/31 install) on this 25.1 and the car seems to drive well on one of the roads that has lot of curves ( Rt 410, East West hwy) and dealing with pedestrians including those that run to catch the metrobus in our congested urban area except for few errors:
One of the left turns where it has to turn on a T intersection to 16th street the car moved very fast and I had to disengage it as I was worried otherwise it would hit the metal barrier.

After that turn it had to go a very short distance and turn right onto Rt 410. I engaged FSD and the car was able to turn right but the lane ended and with a lot of traffic I did not think it would have changed lane quickly before getting stuck. So, I disengaged out of caution.


The stop at the Stop sign seems to have definitely improved.

 

[bhakan]

When I had the issues with cabin camera and FSD Strikes occurring in the evening I requested Tesla to look into the issue. So, I am limiting my evening drives in my Tesla. I have been rescheduling my appts to bring my car as I did not want anything to be screwed up. Tesla said today that they want to change the computer in my car and my appt is for Jan 9th. Finally, will they be able to erase the 3 Strikes I have?