Gizmotoy
Active Member
Though that was a hardware project, which are always difficult, especially when you have to reverse engineer someone else's design. I'm sure someone knowledgable with the camera systems at Tesla could have done it much faster.Not so easy to achieve in software as you might think. WhiteP85 spent two months designing the circuit for the camera switch, and this was just to switch between a front and rear image on the touchscreen with two Tesla cameras mounted on the the car -- the stock Tesla camera on the rear and another mounted on the front.
I agree that, even still, it's probably quite difficult. It's one thing for you to look at the camera image and say "That's a car!", but it's much more difficult for a computer to process the image and come to the same conclusion. Think of all the false positives: we're moving but this section of the image where my blind spot is doesn't seem to be changing. Is that a car, or are we driving next to a wall? Is that a black car, or is it just kind of dark and it's hard to see very far?
I can see it helping inform data coming from some other sensor to improve reliability, but on it's own? I doubt kind of doubt it.