Tesla FSD Beta 10.5 Release Notes, Improvements, & Videos

Below is a portion of an article originally published on Tesla Oracle.

Tesla has rolled out the latest installment of its Full Self-Driving software to the early access Tesla employees and owners — FSD Beta 10.5 (firmware version: 2021.36.8.8).

Just like the previous update V10.4, Tesla has again presented a list of improved features, new capabilities enabled, and the scaling of Autopilot imagery and video labeling to the next level.

Looking at the release notes (below), it’s obvious that Tesla is exponentially improving the safety of vulnerable road users (VRU). With each new update, Tesla is explicitly mentioning how much (in %) the algorithm and data processing has improved for the safety of pedestrians, bicyclists, motorcyclists, and such susceptible road users in the surroundings of a Tesla vehicle equipped with FSD Beta.

One new interesting development is that the “emergency collision avoidance maneuvering” is now enabled in “shadow mode.” This means that the entire Tesla fleet now will be monitoring collision events and what would have happened if the vehicle had used the Autopilot emergency collision avoidance feature. This will improve the AI dataset and labeling of crash avoidance maneuvering data inside the Tesla Neural Net.

FSD Beta 10.5 Release Notes (2021.36.8.8)

Improved VRU (pedestrians, bicyclists, motorcycles) crossing velocity error by 20% from improved quality in our auto-labeling.
Improved static world predictions (road lines, edges, and lane connectivity) by up to 13% using a new static world auto-labeler and adding 165K auto-labeled videos.
Improved cone and sign detections by upreving the generalized static object network with 15K more video clips and adjusting oversampling and overweighting strategies (+4.5% precision, +10.4% recall).
Improved cut-in detection network by 5.5% to help reduce false slowdowns.
Enabled “emergency collision avoidance maneuvering” in shadow mode.
Enabled behavior to lane change away from merges when safe to do so.
Improved merge object detection recall by using multi-modal object prediction at intersection.
Improved control for merges by increasing smoothness of arrival time constraints and considering possible merging objects beyond visibility.
Improved land changes by allowing larger deceleration limit in short-deadline situations.
Improved lateral control for creeping forward to get more visibility.
Improved modeling of road boundaries on high curvature roads for finer maneuvers.
Improved logic to stay on-route and avoid unnecessary detours/rerouting.

FSD Beta 10.5 test videos

The latest version of FSD Beta is better than the previous one but all software need constant improvement over its lifetime. And this is just in the beta phase as of now, so still a long way to go.

FSD Beta 10.5 did make some mistakes during tests, but the overall experience by beta testers was that it has actually improved a bit over 10.4 as Tesla has claimed. Let’s watch how it did on city streets.

[embedded content]