Tesla’s autonomous software suite known as Full Self-Driving has a Beta branch that features the very best technology the company has on offer. While we’ve seen incremental improvements through recent version updates, the Tesla community has been awaiting a large milestone release V11 for a very good reason.
Version 11 of the FSD Beta is expected to have a ‘single stack’. Currently, in V10.x there are multiple software stacks in one release, one that controls the car at low speed (smart summon), another (the best one) during city driving, that takes corners, roundabouts etc and another that handles highway driving (navigate on autopilot) which is the same as the production release.
Moving to a single stack, effectively controlled by AI, rather than human-written code has been on the cards for almost a year, but seems to be finally ready for prime time.
Update
Some partial release notes have now been posted online.
Partial release notes have become available and are available below. More to come soon.
– Enabled FSD Beta on highway. This unifies the vision and planning stack on and off-highway and replaces the legacy highway stack, which is over four years old. The legacy highway stack still relies on several single-camera and single-frame networks, and was setup to handle simple lane-specific maneuvers. FSD Beta’s multi-camera video networks and next-gen planner, that allows for more complex agent interactions with less reliance on lanes, make way for adding more intelligent behaviors, smoother control and better decision making.
Improved Occupancy Network‘s recall for close by obstacles and precision in severe weather conditions with 4x increase in transformer spatial resolution, 20% increase in image featurizer capacity, improved side camera calibration, and 260k more video training clips (real-world and simulation).
– 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 c[unreadable] given lateral and longitudinal acceleration and jerk limits as we[unreadable] vehicle kinematics.
– Improved stability of the FSD UI visualizations by optimizing ethernet data transfer pipeline by 15%.
– Improved recall for vehicles directly behind ego, and improv precision for vehicle detection network.
Version 11 should also include many of the enhancements we saw during AI Day 2022, with the occupancy network finally coming to highway driving, which detects where objects are and routes a path through the driveable space. This differs from the way Tesla had approached the problem for some time, which was to recognise and respond to items based on what they were and their predicted trajectories.
During AI Day 2022, one of the Tesla staff members also shared that the updated summon stack is coming too before the end of the year and the timing would certainly align if that were to show up in V11.
Elon Musk took a break from his appearance on Twitter Spaces tonight to confirm to WholeMars that V11 has rolled out at 11:11pm For those not paying attention, it’s the 11th of November in the US right now, making it.. 11/11/22 at 11:11pm.
According to TeslaScope, which doesn’t yet have release notes for V11, V11 will be 2022.40.5, so those FSD owners or subscribers should look for this release. It is likely deployed to employee cars first, then if things go well in the next couple of days, it’ll reach customer cars.
Naturally here in Australia, we are keen to get the first FSD Beta. The positive side of having to wait a couple of years is that by the time it arrives, it’ll be far more capable than when it first showed up around October 2020.
