If you are just now finding out about Tesletter, you can subscribe here!
If you already know Tesletter and want to support us, check out our Patreon page 🙏
Issue 29 - Google Maps adds EV charging station listings
jimmy_d shared another spectacular analysis of the NN on v9. His first post analyses the NN called AKNET_V9. Here are some details but you should go and read the original post on TMC:
- One unified camera network handles all 8 cameras vs. a sepate network per camera on previous versions
- Same weight file being used for all cameras (this has pretty interesting implications and previously v8 main / narrow seems to have had separate weights for each camera)
- Full resolution for the 3 front cameras and back camera (1280x960) and 1/2 the resolution for pillar and repeater cameras (640x480)
- All cameras analyze 2 frames at the same time (most likely so each camera can see motion)
- The size of the network makes jimmy_d wonder the amount of training data that Tesla is using in their backend and if they are manually tagging all the images since it seems a ton of manual labor, on his own words “there aren’t enough humans to label this much data”
- I really like his closing statement As a neural network dork I couldn’t be more pleased."
In a later post jimmy_d mentions that AKNET_V9 might not be the network that’s currently driving the car since he thinks it’s too big to run on HW2 or HW2.5. His best guess is that it can only run at 3 fps and that doesn’t seem fast enough to be usable. However, it seems posible that this is the network that FSD is using since in HW3 it could run at 30 fps. The current firmware includes a number of different NN and it doesn’t seem easy to understand what is use and what is not, a bunch of the NN that are there are at evolution of what we got in 8.1 but AKNET_V9 seems a completely different beast.
As always, it is delightful to read you jimmy_d, keep up the good work!
Read more: TMC Forum
Free Android app that lets you share navigation from your phone to your car from multiple sites. Didn’t V9 include this feature? Yes. Did this feature work on most apps other than Google Maps? Unfortunately not, so this is a great alternative!
See app in Google Play
Explained by beastpilot: “You get a normal notification, then in the app you can trigger it. Once you trigger it, you get ‘cancel update’ for 2 minutes, as well as a notification that a SW update is about to start. Once it starts the ‘cancel update’ goes away, and then you get a notification it finished once done.”
Read more: Reddit
Google Maps is working to improve the experience of finding charging stations for electric vehicles. Optimized search and listings surface more detailed information like ports availability, and supports Tesla, ChargePoint, and several other networks around the world.
Read more: 5to9Google
Plus rotation and TPMS programming. Winter is coming! And depending on where you live the need to swap your tires, here is how to do that for the Model 3
verygreen and DamianXVI have been able to visualize and decipher data to confirm that V9 has 3D bounding boxes. Proof below. As always, thanks for sharing!
Read more: TMC Forum
It clearly seems like an unintended feature (A.K.A. bug) because it goes away if you then start the navigation, of if you don’t have the camera up and start backing up. But one can only dream that Tesla would bring something like that to the Model S and Model X, this would be superior to the previous “split screen” if they let people re-arrange them.
What the acronym?
As Tesla fan I find myself sometimes having to remember what are all those acronyms, here is a list of the ones you can find in this issue.
- FSD - Full Self-Driving
- HW3 - Hardware 3
- NN - Neural Network
- V9 - Version 9
- M3 - Model 3
- SW - Software
- TPMS - Tire Pressure Monitoring System
Tesletter is put together by Ignacio
. We're really grateful for all the love and feedback we receive. Special thanks to our supporters
. If you like what we do, please consider supporting Tesletter. All the funds will go into making Tesletter more awesome.