Leading EV maker Tesla finally began the rollout of the FSD V11.3 to a wider employee fleet on February 19.
So long as Monday is considered the first day of the week, the rollout fulfills CEO Elon Musk‘s promise to release the update at the “end of the week.”
It is worth noting that the EV giant has been assessing the unified-stack FSD Beta software for a long time. Still, the update has yet to be considered suitable for widespread deployment.
Tougher than expected software update
As mentioned, the Tesla boss promised it would be available “in two weeks” or “by the end of the week.” However, the software update has proven “tougher than expected.”
On Valentine’s Day, Tesla distributed the FSD Beta V11.3 to select employees. Notably, this standard phase of the development process allows Tesla to detect significant issues before sending it to beta testers.
Nobody knew what the new update would bring. However, it was expected to include a unified software stack for highway and city driving. Furthermore, the latest software version would use neural networks for much more programming, providing overall improvements.
Tesla released the FSD V11.3 to a wider internal fleet
Things are improving this week, as Tesla deployed the V11.3 software version to the bigger employee fleet late on Sunday.
It means that rather than 50-60 cars, the latest FSD version now runs on 350-500 cars. Everybody at Tesla can directly test it and help find bugs, a significant step toward a larger customer deployment.
The highlight of this update is FSD Beta on highway driving, which unifies the vision and planning stack on and off-highway.
The FSD Beta V11.3 now employs complex multi-camera streams to reduce reliance on lanes and enable the addition of more intelligent behaviors, sleeker control, and better decision-making. The software can scout the road ahead of time, allowing for a quicker response to blocked lanes and high curvature.
Improved AEB feature
The Automatic Emergency Brake (AEB) feature has undergone significant changes to handle vehicles that may be in the path of a Tesla. In these situations, cars might run a red light or turn in front of another vehicle without giving it the right of way.
By relying more on the object’s instantaneous kinematics and trajectory estimates, the response time to red-light and stop-sign runners has also been advanced by 500 milliseconds.
Significant advancements have been made in trajectory prediction and handling unusual situations on highways and in city driving.
Tesla also claims that moving to more densely supervised auto-labeled datasets enhanced the detection of rare objects by 18% and lesser depth error in large trucks by 9%.
It is also true for school buses and vehicles shifting from stationary to driving. According to Tesla, the new software also improves occupancy network alerts by oversampling on 180K challenging videos, including rain reflections, road debris, and high curvature.
“Voice drive notes” are an intriguing addition to the OTA update 2022.45.5, which includes the FSD Beta V11.3.
It permits Tesla owners to send an unidentified voice message to Tesla describing their experience whenever FSD misbehaves for the software to be improved. Hopefully, the new update will reach more beta testers this week and be widely available “by the end of the week.”
