Categories
Autopilot (FSD) News Software Updates

Tesla starts rolling out FSD Beta 10.69 — “We may now control for slow-moving UFOs” (Release Notes)

Advertisement

Tesla has started rolling out the much-awaited FSD Beta software update release version 10.69 (firmware version: 2022.16.3.10). Elon Musk came up with this version name with the easter egg number 69 — which seems to be his favorite number. Tesla directly moved from 10.13 to 10.69, skipping numbers in between.

The v10.69 Full Self-Driving (FSD) Beta update was originally scheduled for release on 20th August (8/20) — the timeline Tesla seems to have caught up to at the last hour. This easter egg date was teased by Musk last week (8/20 = 2 x 4/20).

According to Technoking of Tesla Elon Musk, the FSD Beta 10.69 will initially be only rolled out to ~1,000 Early Access Program member-owners (except for internal testing employees). In the next phase, the rollout will expand to another ~10,000 Tesla customers.

– Sponsored –

In the FSD Beta 10.69 update, Tesla has upgraded the core architecture with an entirely new module for “deep lane guidance”. This was actually needed to handle more complex turns and lanes like the Chuck Cook-style ones that Tesla started working on in version 10.13.

Tesla claims that the addition of this new module for deep lane guidance reduces the error rate on lane topology by 44%.

The FSD Beta 10.69 has a 20-point long list of improvements that Tesla has detailed in the software update’s release notes (also applies to versions 10.69.1 & 10.69.2 / listed below).

Major improvements include an enhanced Trajectory Planner (point 2), taking unprotected left turns (point 3), architectural upgrades (points 1, 4, 5, 6, 9, 11, 14, & 16), false slowdowns/phantom braking (points 7 & 8), better forward creeping (points 10 & 12).

“We may now control for slow-moving UFOs,” Tesla added this line at the end of point 4 in the release notes. This line is particularly interesting because back in 2017, Elon Musk tweeted that Tesla vehicles will be able to brake for UFOs in the fog discussing the capabilities of radar-based braking with Hardware 1 (HW1).

Fast-forward 5 years, radar is eliminated from Tesla vehicles and the Autopilot algorithm works using only Tesla Visio to realize the Full Self-Driving dream.

The FSD Beta 10.69 rollout is in its very early stage as of this writing. The 1st car detected with this update was a Tesla Model 3 Performance from Oregon, United States. Other owners including some Tesla YouTubers have also started getting this update, check out the 10.69 test videos.

Let’s read the detailed release notes (in full text below) and let us know your opinions n the comments section below.

Advertisement

FSD Beta 10.69 Release Notes (2022.16.3.10)

These release notes are also valid for versions 10.69.1 & 10.69.2.

  1. Added a new “deep lane guidance” module to the Vector Lanes neural network which fuses features extracted from the video streams with coarse map data, i.e. lane counts and lane connectivities. This architecture achieves a 44% lower error rate on lane topology compared to the previous model, enabling smoother control before lanes and their connectivities become visually apparent. This provides a way to make every Autopilot drive as good as someone driving their own commute, yet in a sufficiently general way that adapts for road changes.
  2. Improved overall driving smoothness, without sacrificing latency, through better modeling of system and actuation latency in trajectory planning. Trajectory planner now independently accounts for latency from steering commands to actual steering actuation, as well as acceleration and brake commands to actuation. This results in a trajectory that is a more accurate model of how the vehicle would drive. This allows better downstream controller tracking and smoothness while also allowing a more accurate response during harsh maneuvers.
  3. Improved unprotected left turns with more appropriate speed profile when approaching and exiting median crossover regions, in the presence of high-speed cross traffic (“Chuck Cook style” unprotected left turns). This was done by allowing an optimizable initial jerk, to mimic the harsh pedal press by a human when required to go in front of high-speed objects. Also improved lateral profile approaching such safety regions to allow for a better pose that aligns well for exiting the region. Finally, improved interaction with objects that are entering or waiting inside the median crossover region with better modeling of their future intent.
  4. Added control for arbitrary low-speed moving volumes from Occupancy Network. This also enables finer control for more precise object shapes that cannot be easily represented by a cuboid primitive. This required predicting velocity at every 3D voxel. We may now control for slow-moving UFOs.
  5. Upgraded Occupancy Network to use video instead of images from single time step. This temporal context allows the network to be robust to temporary occlusions and enables the prediction of occupancy flow. Also, improved ground truth with semantics-driven outlier rejection, hard example mining, and increasing the dataset size by 2.4x.
  6. Upgraded to a new two-stage architecture to produce object kinematics (e.g. velocity, acceleration, yaw rate) where network compute is allocated O(objects) instead of O(space). This improved velocity estimates for far away crossing vehicles by 20%, while using one-tenth of the compute.
  7. Increased smoothness for protected right turns by improving the association of traffic lights with slip lanes vs yield signs with slip lanes. This reduces false slowdowns when there are no relevant objects present and also improves yielding position when they are present.
  8. Reduced false slowdowns near crosswalks. This was done with an improved understanding of pedestrian and bicyclist intent based on their motion.
  9. Improved geometry error of ego-relevant lanes by 34% and crossing lanes by 21% with a full Vector Lanes neural network update. Information bottlenecks in the network architecture were eliminated by increasing the size of the per-camera feature extractors, video modules, internals of the autoregressive decoder, and by adding a hard attention mechanism which greatly improved the fine position of lanes.
  10. Made speed profile more comfortable when creeping for visibility, to allow for smoother stops when protecting for potentially occluded objects.
  11. Improved recall of animals by 34% by doubling the size of the auto-labeled training set.
  12. Enabled creeping for visibility at any intersection where objects might cross ego’s path, regardless of the presence of traffic controls.
  13. Improved accuracy of stopping position in critical scenarios with crossing objects, by allowing dynamic resolution in trajectory optimization to focus more on areas where finer control is essential.
  14. Increased recall of forking lanes by 36% by having topological tokens participate in the attention operations of the autoregressive decoder and by increasing the loss applied to fork tokens during training.
  15. Improved velocity error for pedestrians and bicyclists by 17%, especially when the ego is making a turn, by improving the onboard trajectory estimation used as input to the neural network.
  16. Improved recall of object detection, eliminating 26% of missing detections for far away crossing vehicles by tuning the loss function used during training and improving label quality.
  17. Improved object future path prediction in scenarios with high yaw rate by incorporating yaw rate and lateral motion into the likelihood estimation. This helps with objects turning into or away from ego’s lane, especially in intersections or cut-in scenarios.
  18. Improved speed when entering highways by better handling of upcoming map speed changes, which increases the confidence of merging onto the highway.
  19. Reduced latency when starting from a stop by accounting for lead vehicle jerk.
  20. Enabled faster identification of red light runners by evaluating their current kinematic state against their expected braking profile.

Stay tuned for constant Tesla updates, follow us on: 
Google News | Flipboard | RSS (Feedly).

Related

Advertisement

By Iqtidar Ali

Iqtidar has been writing about Tesla, Elon Musk, and EVs for more than 3 years on XAutoWorld.com, many of his articles have been republished on CleanTechnica and InsideEVs, maintains a healthy relationship with the Tesla community across the Social Media sphere. You can reach him on Twitter @IqtidarAlii

Leave a Reply

Your email address will not be published.