-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Elon: FSD Beta tweets
- Thread starter Terminator857
- Start date
Where was that cited?
Tesla is still selling FSD and letting people in the program just as they did with the NHTSA rolling stop "recall".
That's maybe the 20th time Elon missed a window for V11 release.
Knightshade
Well-Known Member
You (and the guy you are replying to) are mixing up FSD (which was never recalled and can still be sold)
and FSDb, which is not a paid thing at all, has to be specifically requested by itself, and which Tesla can't put on new vehicles until the recall is fixed (and the recall notice specifically lists the fact it's NOT installed on new vehicles under the required explanation of how the issue was (past tense) corrected for new vehicles.
No.
My next door neighbor applied for Beta on Thursday and got it today.
They aren't stopping new testers. There's no reason to believe that.
Edit: He got it last night.
sleepydoc
Well-Known Member
The recall was just announced, so it’s too early to make any conclusions. My assumption is that NHTSA probably has a few different categories based on risk severity. Something like a faulty brake line that poses an immediate danger to life and limb would warrant a ban on sales until the remedy is in place where as a more trivial issue like the one with the seatbelt chime last year doesn’t require such urgency.
(this is just my assumption, I have no proof whether this is the case or not)
n.one.one
Info w/o Drama
KArnold
Active Member
(
)
**FSD Beta v11.3 Release Notes**
- Enabled FSD Beta on highway. This unifies the vision and planning stack on and off-highway and replaces the legacy highway stack, which is over four years old. The legacy highway stack still relies on several single-camera and single-frame networks, and was setup to handle simple lane-specific maneuvers. FSD Beta’s multi-camera video networks and next-gen planner, that allows for more complex agent interactions with less reliance on lanes, make way for adding more intelligent behaviors, smoother control and better decision making.
- Added voice drive-notes. After an intervention, you can now send Tesla an anonymous voice message describing your experience to help improve Autopilot.
- Expanded Automatic Emergency Braking (AEB) to handle vehicles that cross ego’s path. This includes cases where other vehicles run their red light or turn across ego’s path, stealing the right-of-way.
- Replay of previous collisions of this type suggests that 49% of the events would be mitigated by the new behavior. This improvement is now active in both manual driving and autopilot operation.
- Improved autopilot reaction time to red light runners and stop sign runners by 500ms, by increased reliance on object’s instantaneous kinematics along with trajectory estimates.
- Added a long-range highway lanes network to enable earlier response to blocked lanes and high curvature.
- Reduced goal pose prediction error for candidate trajectory neural network by 40% and reduced runtime by 3X. This was achieved by improving the dataset using heavier and more robust offline optimization, increasing the size of this improved dataset by 4X, and implementing a better architecture and feature space.
- Improved occupancy network detections by oversampling on 180K challenging videos including rain reflections, road debris, and high curvature.
- Improved recall for close-by cut-in cases by 20% by adding 40k autolabeled fleet clips of this scenario to the dataset. Also improved handling of cut-in cases by improved modeling of their motion into ego’s lane, leveraging the same for smoother lateral and longitudinal control for cut-in objects.
- Added “lane guidance module and perceptual loss to the Road Edges and Lines network, improving the absolute recall of lines by 6% and the absolute recall of road edges by 7%.
- Improved overall geometry and stability of lane predictions by updating the “lane guidance” module representation with information relevant to predicting crossing and oncoming lanes.
- Improved handling through high speed and high curvature scenarios by offsetting towards inner lane lines.
- Improved lane changes, including: earlier detection and handling for simultaneous lane changes, better gap selection when approaching deadlines, better integration between speed-based and nav-based lane change decisions and more differentiation between the FSD driving profiles with respect to speed lane changes.
- Improved longitudinal control response smoothness when following lead vehicles by better modeling the possible effect of lead vehicles’ brake lights on their future speed profiles.
- Improved detection of rare objects by 18% and reduced the depth error to large trucks by 9%, primarily from migrating to more densely supervised autolabeled datasets.
- Improved semantic detections for school busses by 12% and vehicles transitioning from stationary-to-driving by 15%. This was achieved by improving dataset label accuracy and increasing dataset size by 5%.
- Improved decision making at crosswalks by leveraging neural network based ego trajectory estimation in place of approximated kinematic models.
- Improved reliability and smoothness of merge control, by deprecating legacy merge region tasks in favor of merge topologies derived from vector lanes.
- Unlocked longer fleet telemetry clips (by up to 26%) by balancing compressed IPC buffers and optimized write scheduling across twin SOCs.
**FSD Beta v11.3 Release Notes**
- Enabled FSD Beta on highway. This unifies the vision and planning stack on and off-highway and replaces the legacy highway stack, which is over four years old. The legacy highway stack still relies on several single-camera and single-frame networks, and was setup to handle simple lane-specific maneuvers. FSD Beta’s multi-camera video networks and next-gen planner, that allows for more complex agent interactions with less reliance on lanes, make way for adding more intelligent behaviors, smoother control and better decision making.
- Added voice drive-notes. After an intervention, you can now send Tesla an anonymous voice message describing your experience to help improve Autopilot.
- Expanded Automatic Emergency Braking (AEB) to handle vehicles that cross ego’s path. This includes cases where other vehicles run their red light or turn across ego’s path, stealing the right-of-way.
- Replay of previous collisions of this type suggests that 49% of the events would be mitigated by the new behavior. This improvement is now active in both manual driving and autopilot operation.
- Improved autopilot reaction time to red light runners and stop sign runners by 500ms, by increased reliance on object’s instantaneous kinematics along with trajectory estimates.
- Added a long-range highway lanes network to enable earlier response to blocked lanes and high curvature.
- Reduced goal pose prediction error for candidate trajectory neural network by 40% and reduced runtime by 3X. This was achieved by improving the dataset using heavier and more robust offline optimization, increasing the size of this improved dataset by 4X, and implementing a better architecture and feature space.
- Improved occupancy network detections by oversampling on 180K challenging videos including rain reflections, road debris, and high curvature.
- Improved recall for close-by cut-in cases by 20% by adding 40k autolabeled fleet clips of this scenario to the dataset. Also improved handling of cut-in cases by improved modeling of their motion into ego’s lane, leveraging the same for smoother lateral and longitudinal control for cut-in objects.
- Added “lane guidance module and perceptual loss to the Road Edges and Lines network, improving the absolute recall of lines by 6% and the absolute recall of road edges by 7%.
- Improved overall geometry and stability of lane predictions by updating the “lane guidance” module representation with information relevant to predicting crossing and oncoming lanes.
- Improved handling through high speed and high curvature scenarios by offsetting towards inner lane lines.
- Improved lane changes, including: earlier detection and handling for simultaneous lane changes, better gap selection when approaching deadlines, better integration between speed-based and nav-based lane change decisions and more differentiation between the FSD driving profiles with respect to speed lane changes.
- Improved longitudinal control response smoothness when following lead vehicles by better modeling the possible effect of lead vehicles’ brake lights on their future speed profiles.
- Improved detection of rare objects by 18% and reduced the depth error to large trucks by 9%, primarily from migrating to more densely supervised autolabeled datasets.
- Improved semantic detections for school busses by 12% and vehicles transitioning from stationary-to-driving by 15%. This was achieved by improving dataset label accuracy and increasing dataset size by 5%.
- Improved decision making at crosswalks by leveraging neural network based ego trajectory estimation in place of approximated kinematic models.
- Improved reliability and smoothness of merge control, by deprecating legacy merge region tasks in favor of merge topologies derived from vector lanes.
- Unlocked longer fleet telemetry clips (by up to 26%) by balancing compressed IPC buffers and optimized write scheduling across twin SOCs.
Knightshade
Well-Known Member
According to the Federal Motor Vehicle Safety Act, it is illegal to sell a new car with an open recall on it.... likewise the recall announcement itself points out it's ok for Tesla to keep selling cars specifically because new ones aren't getting this software
Cars that already had it queued/downloaded but not installed might be an exception... but new requests ought to be getting denied to my knowledge.
So again- I expect it's more than nobody in Tesla IT has bothered to actually do anything about it than it is NHTSA saying "Eh, keep distributing safety recalled items, it's fine"
Tesla is notoriously horrible at this internal IT kinda stuff. See the times they flagged FSD removal from a car that came into their possession on the back end, without actually pushing the removal to the vehicle until after they'd resold it for another great example of it.
DarkForest
10.69 made me a shill
Wow! Lots of amazing updates mentioned. I’m HYPED.
Haven’t you learned by now. Keep your expectations low…that way you at least won’t be disappointed lol
This one looks different. The updates seem to actually be quality-of-life improvements. Take, for example, “Improved handling through high speed and high curvature scenarios by offsetting towards inner lane lines,” this is actually something that will be noticed by the masses. I’m going to not try and be overly optimistic, but I think this one might actually be
(I hope). Then again, we haven’t gotten an update in literally forever so I could just be excited.The big question is have they solved lane selection when there is bad map data…. If they could fix this, then at least on my drives that I do I would be close to be willing to call my car a full self driving vehicle. Close- not there yet. Nobody panic.
Will be interesting to see who all gets the “voice report” feature. Probably just the shills and OG testers.
With all the stuff going on with HW4, the recall, and now these crazy release notes, much is happening in the Tesla world.
DanCar
Active Member
- Improved handling through high speed and high curvature scenarios by offsetting towards inner lane lines.
Now it will drive more normally on exits and right turns?
Now it will drive more normally on exits and right turns?
Yes, exactly. They cannot sell or deliver new FSDb, as I read it anyway. If they in fact are still doing so, then that's Strike 1:
Please be reminded that under 49 U.S.C. § 30112(a)(3), it is illegal for a manufacturer, to sell, offer for sale, import, or
introduce or deliver into interstate commerce, a motor vehicle or item of motor vehicle equipment that contains a safety defect once the manufacturer has notified NHTSA about that safety defect. This prohibition does not apply once the motor vehicle or motor vehicle equipment has been remedied according to the manufacturer's instructions. {PDF}
Well, based on the testimony of Tesla's director of Autopilot software, Ashok Elluswamy, after the AP crash death of Walter Huang. Ashok said he did not recall being at any formal meeting to discuss the high-profile crash death, when one can hardly imagine how they COULD NOT have had serious meetings about this major media story. So one can guess how seriously they take safety-related issues at Tesla. If we ask him in a few weeks, can he recall holding any meetings to discuss this recall?
A combination of Autopilot following a bolder line on the highway plus see below...but this I didn't know ---> Tesla had no response?
Driver in 2018 fatal Mountain View Tesla crash was playing video game, NTSB says
"Cell phone records for Huang's iPhone 8-plus show he was active on the "Three Kingdoms" video game, his hands off the steering wheel in the six seconds before impact, and that he was on the game driving to work each day that week."
"Huang's family said he complained that the autopilot steered toward that barrier on several occasions, as recently as four days before the crash."
Driver in 2018 fatal Mountain View Tesla crash was playing video game, NTSB says
"Cell phone records for Huang's iPhone 8-plus show he was active on the "Three Kingdoms" video game, his hands off the steering wheel in the six seconds before impact, and that he was on the game driving to work each day that week."
"Huang's family said he complained that the autopilot steered toward that barrier on several occasions, as recently as four days before the crash."
Or maybe it won’t hug the centerline on twistier roads like it does now?
Must be employee only Notes. NO WAY this:
- Added voice drive-notes. After an intervention, you can now send Tesla an anonymous voice message describing your experience to help improve Autopilot.
Makes it into the consumer version. Can you imagine Tesla hiring >50k people to sit and listen to a >million audio recordings every day?
- Added voice drive-notes. After an intervention, you can now send Tesla an anonymous voice message describing your experience to help improve Autopilot.
Makes it into the consumer version. Can you imagine Tesla hiring >50k people to sit and listen to a >million audio recordings every day?
My phone does a very good job of listening to voicemail messages and converting them to text messages. Whatever the current version is, it's better than my voice dictation keyboard feature, as it makes pretty good sense even when people are not speaking carefully.
I'm sure Tesla has access to the very latest versions of such software. Once transcribed, then they can mine it for keywords and presumably match it up with camera and telemetry info if they want.
But as you suggest, this feature might not be released widely.
Similar threads
