Do you have a link to the Teslascope report of the second collision, as I can only find reference to a single collision, happening about a week after the rollout:
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD v12.x (end to end AI)
- Thread starter Buckminster
- Start date
-
- Tags
- #fixphantombraking autopilot chill is not chill elon is an ass fix fsd extreme late deceleration fsd does not commit to go fsd does not maintain proper speed fsd hard and late braking fsd jerky driving fsd late deceleration fsd stops too early too quickly fsd stops too far behind line fsd tailgating in city streets
That is referencing the parking lot incident.
Click on my quote and it will take you to the screen shot. It happened on March 3rd.
Why do you think that? They said a crash involving FSD V12 on March 3. They reported on the parking lot incident when it originally occurred in February.
As well as the overlord. TBH, I'm surprised there haven't been more reported v12 accidents given v12's performance and the crew driving it.
gottagofast
2022 Model 3 Performance
unforeseeable future makes it sound like a very long delay, why would it take so long to update 1 mistake
We waiting another 6 months to a year?
We waiting another 6 months to a year?
No, it does not say that.
This is what they say:
I think V12.2.1 came out about 2/19, so the collision they are talking about occurred before 2/26. Which lines up exactly with the parking lot collision that was reported on the 25th.
Where do they say it happened on the 3rd? That is just when they started reporting about it causing a delay in new V12 deployments/versions.
You can read back a few pages to see the issues but for certain it's not just one mistake. It's a design feature. V12 put all the burden on NN training and the NN training has never been an FSD strength. It might take a while if they need to substantially bolster training.
You mean this one?
It is the same event. They report the delay on the 3rd because of a collision, and provide the details of the delay on the 4th saying that the collision happened about 2/25. (i.e. the parking lot incident.)
3/3 Post:
Follow-up on the 4th:
That is literally the tweet I linked to. They do not say that it happened on the 3rd. The more information they said they would publish after 2:30AM was what I linked to that said it happened before 2/26.
Note: they clarified in another tweet that the V11.2.1 was a typo and it was really V12.2.1.
Last edited:
Correct. But the delay was not know when it was originally reported, which is what they were reporting on the 3rd. (And saying that the delay was because of the parking lot collision, without actually saying it.)
PianoAl
Active Member
Exactly what I do. Works well.
Same here. We'd hear more about this if more testers lived in rural areas with less traffic.
Yes. There's a spot where a sign says "End 45 MPH" and there's a "55" on the road surface. V11 got it, V12 doesn't.
It still misinterprets speed limit signs with "TRUCK" above them.
I think I've noticed some other problems with signs. I'll pay attention to that.
Sounds like Teslascope just doesn't know how long it'll take to address the issue. Assuming it was the parking lot crash on February 24th that "required a full internal investigation the following day," delaying 12.2.2 or 12.3 or whatever expected deployment originally scheduled for March 3rd would suggest they're still trying to address the parking lot issue as it was "the reason for the missed deployment date."
Depending on the underlying issue, Tesla could have a range of fixes such as preventing usage in parking lots, limiting end-to-end based on ultrasonics, retraining 12.x to avoid creeping into parked vehicles, etc. Perhaps the incident revealed some more fundamental issue that Tesla didn't notice earlier from smaller internal testing, or it could have been a "simple" bug of heuristics that prevented end-to-end from reversing.
Captkerosene
Member
James Douma: This is so good that I wonder if the AP team might not shift development approach. Up until now big improvements have required big changes. But the range of behaviors that are being effectively implemented with E2E path planning and control suggests to me that they can get the rest of the way to "better than human" solely on top of more and better data and training.
Elon: Thanks, that is an accurate assessment. Training compute is currently our limiting factor, but that issue is being resolved fast.
Elon: Thanks, that is an accurate assessment. Training compute is currently our limiting factor, but that issue is being resolved fast.
I do feel duped, and I do feel an owners foru m is the proper place for an owner to express his dissatisfaction with a product that was sold as self driving
Meanwhile, I am waiting for the promises to come true. If I am to believe the shills, Tesla. And Elon, the product I purchased full self driving, turning my car into an autonomous vehicle, is still a reality and I should have the product by the end of the year. I have the beta but not the employee and youtube version 12.
I am ready to test and criticize it
The current version of automated driver assistance, certainly is not full self driving, and is more of a headache to use then drive the car myself. For cruise control on a highway, it is OK, but so is every other driver assisted advanced cruise control. That is not what I paid for.
That is not what I am still expecting. That is not what I was promised. That is not what I am still being promised.
Cruise control became popular because it made your experience better on the road more relaxing, and was a good experience for the other drivers of other vehicles. I thought that would be the case with self driving. That was the promise, but that is not the reality.
I am still waiting for that reality, and as soon as it comes, we can stop talking about it and actually start talking about the product.
When will they define the product, and at least release the specification?
But most of all I want to express thanks for your sympathy that I feel I was duped. I’m sure you agree with me because you express no reasons why I was not duped. so again thank you and I will let you know as soon as I receive the product I paid for over five years ago.
And if it is amazing, I will sing it praises from the rooftop, and even give it a couple
Here's the full context if people were wondering:
powertoold
Active Member
100x intervention reduction over in 11.4.9 is a tall order
Seems impossible based on my experience with V12.2.1
Once again, the next release is our last hope
Seems impossible based on my experience with V12.2.1
Once again, the next release is our last hope
Similar threads
- Article
