Tesla FSD(Full Self Driving) Beta V10.12 Update out to employees, it’s a Big One, On 19 May 2022.
Tesla FSD beta software update V10.12 with some improvement based on over the 250,000 video clips from the fleet, Tesla has released notes for the FSD beta update on Twitter to aware people of the extensive work on its Tesla Full self-driving (FSD) Software. This update is a big update with a lot of improvement has been done by Tesla.
As per the TeslaScope, the FSD beta V10.12 update is now going out to the employee but the release notes for the update beta have not been released so far.
Last week, the CEO of Tesla, Elon Musk said that the upcoming TESLA Beta version contains a lot of updates in its software including additional features such as new tendencies and behaviors these are in the testing phase, and Tesla employees notify customers about the FSD beta update program before participating.
Right now, The company has up to 100,000 owners in the FSD program, it is expecting that beta has more data to train its neural nets.
Musk also predicted that if everything will be done well with beta update 10.12, Tesla may lower the required safety score to join the data program to 95+.
Certainly, This is first-time video clips pulled from the fleet and the carmaker has expected that around 250,000 new clips use in the training set for this beta program.
Musk said, replying to a question on Twitter, “Team is working the weekend to get 10.12 in limited release tomorrow. Then we evaluate, do a point update, and broaden the release. 10.12.2 is probably where we can expand to a safety score of 95+,”
Elon also announced that it has removed three neural nets from the system and enabled 18 frames per second(FPS) to improve the system frame rate.
There are a lot of improvements in the FSD beta program below:
- An upgraded decision-making framework for an unprotected left turn with better modeling of objects’ response to ego’s actions by adding more features that shape the go/no-go decision. This increases robustness to noisy measurements while being more sticky to decisions within a safety margin. The framework also leverages median safe regions when necessary to maneuver across large turns and accelerate harder through maneuvers when required to safely exit the intersection.
- Improved creeping for visibility using more accurate lane geometry and higher resolution occlusion detection.
- Reduced instances of attempting uncomfortable turns through better integration with object future predictions during lane selection.
- Upgraded planner to rely less on lanes to enable maneuvering smoothly out of restricted space.
- Increased safety of turns with crossing traffic by improving the architecture of the lanes neural network which greatly boosted recall and geometric accuracy of crossing lanes.
- Improved the recall and geometric accuracy of all lane predictions by adding 180k video clips to the training set.
- Reduced traffic control-related false slowdowns through better integration with lane structure and improved behavior with respect to yellow lights.
- Improved the geometric accuracy of road edge and line predictions by adding a mixing/coupling layer with the generalized static obstacle network.
- Improved geometric accuracy and understanding of visibility by retraining the generalized static obstacle network with improved data from the auto labeler and by adding 30k more video clips.
- Improved recall of motorcycles, reduced velocity error of close-by pedestrians and bicyclists, and reduced heading error of pedestrians by adding new sim and auto labeled data to the training set.
- Improved precision of the “js parked” attribute on vehicles by adding 41k clips to the training set. Solved 48% of failure cases captured by our telemetry of 10.11.
- Improved detection recall of far-away crossing objects by regenerating the dataset with improved versions of the neural networks used in the auto labeler which increased data quality.
- Improved offsetting behavior when maneuvering around cars with open doors.
- Improved angular velocity and lane-centric velocity for non-VRU objects by upgrading them into network-predicted tasks.
- Improved comfort when lane changing behind vehicles with harsh deceleration by tighter integration between the lead vehicle’s future motion estimate and planned lane change profile.
- Increased reliance on network-predicted acceleration for all moving objects, previously only longitudinally relevant objects.
- Updated nearby vehicle assets with visualization indicating when a vehicle has a door open.
- Improved system frame rate +1.8 frames per second by removing three legacy neural networks.
Here are the Official release notes below: