Upgraded the Reinforcement Learning (RL) stage of training the FSD neural network, resulting in improvements in a wide variety of driving scenarios.
Upgraded the neural network vision encoder, improving understanding in rare and low-visibility scenarios, strengthening 3D geometry understanding, and expanding traffic sign understanding.
Rewrote the Al compiler and runtime from the ground up with MLIR, resulting in 20% faster reaction time and improving model iteration speed.
Mitigated unnecessary lane biasing and minor tailgating behaviors.
Increased decisiveness of parking spot selection and maneuvering.
Improved parking location pin prediction, now shown on a map with a P icon.
Enhanced response to emergency vehicles, school buses, right-of-way violators, and other rare vehicles.
Improved handling of small animals by focusing RL training on harder examples and adding rewards for better proactive safety.
Improved traffic light handling at complex intersections with compound lights, curved roads, and yellow light stopping — driven by training on hard RL examples sourced from the Tesla fleet.
Improved handling for rare and unusual objects extending, hanging, or leaning into the vehicle path by sourcing infrequent events from the fleet.
Improved handling of temporary system degradations by maintaining control and automatically recovering without driver intervention, reducing unnecessary disengagements.
Unified the model between Actually Smart Summon, FSD, and Robotaxi for more capable and reliable behavior.
Help Tesla improve Self-Driving by selecting an intervention reason on the main screen after taking over.
Upcoming Improvements
Expand reasoning to all behaviors beyond destination handling.
Add pothole avoidance.
Improve driver monitoring system sensitivity with better eye gaze tracking, eye wear handling, and higher accuracy in variable lighting conditions.