I'm a happy and impressed FSD user who is curious about how Tesla gathers FSD data, filters and uses it to improve future versions.
My first question is re Teslas not using FSD. Musk has implied that the millions of Teslas currently on the road are somehow providing the company data that will improve FSD. What data might this be, and how is it gathered and utilized?
Second question is regarding cars like mine that are using FSD. I'm guessing that when my car is sitting in the garage at night, within range of my house wifi, video clips of FSD disengagements are uploaded to Tesla (tho I have no actual evidence of that). But if that is correct, what happens next? Is there a manual review of those clips to see which might be useful for training purposes? (Sounds prohibitively labor intensive.) If not then is there a software-based review of the uploaded clips that somehow sorts the useful clips and feeds them to the neural network?
Thanks for any help educating me.
My first question is re Teslas not using FSD. Musk has implied that the millions of Teslas currently on the road are somehow providing the company data that will improve FSD. What data might this be, and how is it gathered and utilized?
Second question is regarding cars like mine that are using FSD. I'm guessing that when my car is sitting in the garage at night, within range of my house wifi, video clips of FSD disengagements are uploaded to Tesla (tho I have no actual evidence of that). But if that is correct, what happens next? Is there a manual review of those clips to see which might be useful for training purposes? (Sounds prohibitively labor intensive.) If not then is there a software-based review of the uploaded clips that somehow sorts the useful clips and feeds them to the neural network?
Thanks for any help educating me.