Tesla has started pushing the Full Self-Driving Beta 10.11 update to users and the release notes include several critical improvements for the semi-autonomous driving system.
For now, FSD Beta 10.11 is only rolling out to Tesla employees, but if everything goes well, other users should get the update in the coming days. More importantly, Elon Musk tweeted on March 13 that Tesla will “probably” grant access to FSD Beta to users who scored 95 on the driver safety score.
Among the important improvements mentioned in the release notes are more accurate predictions of where vehicles are turning or merging, which should reduce unnecessary slowdowns, and an improved right-of-way understanding if the map is inaccurate or the car cannot follow the navigation.
Furthermore, FSD Beta 10.11 brings specific improvements for vulnerable road users (VRU), namely pedestrians and cyclists. VRU detection is improved by 44.9% in the latest update, allowing the system to dramatically reduce “spurious false positive pedestrians and bicycles” and decrease the incidence of VRU-related false slowdowns—a.k.a. phantom braking.
According to the release notes, Tesla was able to accomplish these VRU improvements by increasing the size of its next-generation labelers.
Another important update regards a 63.6% reduction in the predicted velocity error of very close-by motorcycles, scooters, wheelchairs, and pedestrians.
For more details, check out the release notes in full below.
- Upgraded modeling of lane geometry from dense rasters (“bag of points”) to an autoregressive decoder that directly predicts and connects “vector space” lanes point by point using a transformer neural network. This enables us to predict crossing lanes, allows computationally cheaper and less error prone post-processing, and paves the way for predicting many other signals and their relationships jointly and end-to-end. Use more accurate predictions of where vehicles are turning or merging to reduce unnecessary slowdowns for vehicles that will not cross our path.
- Improved right-of-way understanding if the map is inaccurate or the car cannot follow the navigation. In particular, modeling intersection extents is now entirely based on network predictions and no longer uses map-based heuristics.
- Improved the precision of VRU detections by 44.9%, dramatically reducing spurious false positive pedestrians and bicycles (especially around tar seams, skid marks, and rain drops). This was accomplished by increasing the data size of the next-gen autolabeler, training network parameters that were previously frozen, and modifying the network loss functions. We find that this decreases the incidence of VRU-related false slowdowns.
- Reduced the predicted velocity error of very close-by motorcycles, scooters, wheelchairs, and pedestrians by 63.6%. To do this, we introduced a new dataset of simulated adversarial high speed VRU interactions. This update improves autopilot control around fast-moving and cutting-in VRUs.
- Improved creeping profile with higher jerk when creeping starts and ends.
- Improved control for nearby obstacles by predicting continuous distance to static geometry with the general static obstacle network.
- Reduced vehicle “parked” attribute error rate by 17%, achieved by increasing the dataset size by 14%. Also improved brake light accuracy.
- Improved clear-to-go scenario velocity error by 5% and highway scenario velocity error by 10%, achieved by tuning loss function targeted at improving performance in difficult scenarios.
- Improved detection and control for open car doors.
- Improved smoothness through turns by using an optimization-based approach to decide which road lines are irrelevant for control given lateral and longitudinal acceleration and jerk limits as well as vehicle kinematics.
- Improved stability of the FSD Ul visualizations by optimizing the ethernet data transfer pipeline by 15%.
The previous significant update was FSD Beta 10.10 in early February, which removed the rolling stop feature that had drawn criticism from NHTSA. As of the fourth quarter of 2021, Tesla had almost 60,000 owners in the FSD Beta program.
- Share on Facebook
- Share on Twitter
- Share on LinkedIn
- Share on Flipboard
- Share on Reddit
- Share on WhatsApp
- Send to email
Source: Read Full Article