Submitted by Son of a Gunn on November 12, 2015
Tesla, there are thousands of people driving the Model S with Autopilot. Why can't we get a proper technical explanation of how Autopilot "learns"? We have educated hypotheses on how AP learns and Elon himself said it learns. We think we know how and some of us are probably spot on with speculation, but we want validation from JB or Elon. We can understand physics. Don't just explain like we're five years old. Communicate with us, your loyal user base.
Is it sampling? Is it real-time? How do new algorithms get back to the car? When are new algorithms received, (because I don't see any notifications for new updates)? Are you micro-updating transparently? Or is it simply a self-contained system on a chip independent of the fleet, doing it all alone, like geofencing, and there really is no fleet learning updated just yet?
This is the Google generation--we want genuine technical details. Sooner or later someone from Silicon Valley will put packet sniffers on the car and figure it out and they'll get credit for the explanation.