Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
Actually you are the one incorrect. On my last road trip, maybe 3 months ago, on whatever the current FSDb version was, on a 2-lane highway I'd be right about at the middle of the tractor-trailer when I got a serious brake. Absolutely horrible. I have my own way of passing trucks, a habit from when stuff that kicks up would ding my paint (not currently a problem) where the key is to overtake smoothly, so that sudden brake for no friggin' reason got real old real fast. After 4 or 5 times, I'd just disengage to get past the trucks.
OK, communication confusion I think, so sorry for my part. I thought you were saying that FSD "does not" slam on the brakes while in the process of overtaking a truck, and I'm saying "it does" based on my experiences lately.
 
  • Like
Reactions: beachmiles
OK. Your experience doesn't match mine.
It would be statistically freaky if every 18-wheeler waited until I was precisely half-way through going by him and then swerved toward me at that exact moment. Yet the brakes got slammed only when I was abreast of these trucks. And I actually got no other phantom braking events on that entire trip except for all those next to trucks.
 
It would be statistically freaky if every 18-wheeler waited until I was precisely half-way through going by him and then swerved toward me at that exact moment. Yet the brakes got slammed only when I was abreast of these trucks. And I actually got no other phantom braking events on that entire trip except for all those next to trucks.
OK. Your experience still doesn't match mine. You don't need to prove it.
 
  • Like
Reactions: Yelobird
My top V12 desired "fixes" are:

1) No more hesitation after a right turn
2) No jerkiness while doing unprotected rights
3) No more lunging towards lead cars
4) Less hesitation whenever there's an out of the ordinary situation (lol)
My number 1 (and I think this is more a HW4 problem than a HW3 problem) is to stop trying to pull out into traffic when cars are coming, and to stop pulling out aggressively from blind corners.

I don’t have that many pairs of clean underwear left.
 
  • Like
  • Funny
Reactions: Joesmoe3 and Hiline
My number 1 (and I think this is more a HW4 problem than a HW3 problem) is to stop trying to pull out into traffic when cars are coming, and to stop pulling out aggressively from blind corners.

I don’t have that many pairs of clean underwear left.
Agree completely. I have a HW4 car and both problems you list are now worse than my old HW3 car but they were still a problem with HW3. Just worse now. It's the exception now when I have a zero disengagement drive.
 
V12 improvements don't seem good enough to get to generalized robotaxis
Anything in particular that led you to that? The improvement or regression rates from 12.1 feedback?

FSD Beta 10.x had over a dozen releases over a year while 11.x had a handful over a similar time-frame. For 12.x rewrite to already maintain core driving capability and improve some other aspects of a system with over 3 years of public development and testing seems to be quite the accomplishment already. We'll have to see if the new architecture actually results in a faster rate of progress as it seems to be very much dependent on Tesla's ability to continuously train the neural networks even more than before.
 
Anything in particular that led you to that? The improvement or regression rates from 12.1 feedback?

Well, if those reviews and impression percentages are legit, they seem too low to be vastly better than 11.4.9, and I think fsd will have to be vastly better than 11.4.9 to deploy as robotaxis without micromanagement and excessive liability.

Granted, Tesla may still deploy robotaxis by scaling back V12 training data to a smaller region, like rather than data from USA-wide, they can concentrate all the training to CA and use all billion parameters for a CA-only robotaxis service for example.
 
Last edited:
My top V12 desired "fixes" are:
1) Drive WAY more chill especially while FSD is in beta and chill mode is selected. Limit lane changes should be the default for instance. FSD should not be programmed to have the Dunning Kruger syndrome
2) Start to decelerate WAY earlier for stopping/slowing situations instead of waiting till the last second and using hard friction brakes. (See desired fix #1)
3) Phantom braking needs to end especially with cars right behind you. (See desired fix #1)
4) Stop swerving in large right lanes with unmarked parking lanes / unmarked right turn lanes / unmarked biking lanes and instead hug the left side of large right lanes. (See desired fix #1)
5) Stop tailgating cars in city streets and give larger following distances especially in chill mode. (See desired fix #1)
6) When getting in a dedicated left turn lane don't only drive halfway into the turn lane and hog 2 lanes. Instead just fully get in the left turn lane ASAP. The car needs to signal way earlier for turning situations as well.
 
Last edited:
While I'm ot expecting v12 to be an immediate miracle, I'm hopeful and enthusiastic to see how it works on my familiar routes.

My top two categories of v11 problems are both associated with lane selection:
  • The often-mentioned tendency to interpret a turn lane as a new driving lane, and then wanting to move into it. I still haven't figured out why, even if it were correct about the identification of a new driving lane, it's so anxious to change into it. I have a place where there really is a new third lane that opens up on the right. FsD reliably wants to move into it, (lnot helpful because I need to turn left about a mile on.
  • The entirely perverse desire to want to move out of the lane we need to be in, for a turn that's upcoming in a half mile or so. It fights very hard to accomplish this then, then finally stops fighting when the lady says the turn is coming in 1000 ft. I've said before that this really, really feels like a heuristic software bug. As in there's a subroutine telling it to move into the correct lane well ahead of time - but to accomplish that it thinks it needs to be in the wrong lane first.
The point of bringing up these v11 foibles in a v12 thread? I wonder whether these things, especially the second one, are related to the navigation map-hints infrastructure. And if so, I wonder if that infrastructure will be replaced by NN or maybe new heuristic code in v12.

It would be a shame if these, the most common reasons for intervention and frustration (and embarrassment in front of other drivers) are still with us in v12. Hopefully not, but my optimism is guarded because they don't seem to be really part of the perception and control tasks, they seem more like problems with the map / navigation system.

Anyway it should be a lot of fun two weeks from now er, whatever time that it's two weeks from... :)
 
  • Like
Reactions: Jeff N