Originally Posted by
flangehead
Agreed. Figured it wouldn't hurt to try. It did seem to delay the onset of normal form, but ultimately a fail.
I've been riding more at night lately (seasonal) and I came to the same conclusion. Nothing special about this radar-less Tesla. I had gotten a bit lazy about relying on yellow jersey, so this motivated me to make sure I've got a reflective vest on. I use reflective ankle bands which would help (bio-motion) but my panniers block them from rear view.
Alas, I think the problem in terms of this as a topic is we can't do anything but guess as to what the Tesla system is going to use as cues based only on lighting--I don't think even Tesla knows that. For example, I haven't got a clue as to whether the AI is going to be able to figure out that reflective ankle bands are even on the same vehicle as each other, or their relationship to the rear red light or whatever. I agree with the video that the removal of the radar really can't be justified as anything but a cost-saving measure. Isn't the whole point of AI that the machine is capable of synthesizing more data faster than a human being? As the video points out, sure removing a stream of data reduces the conflicts between streams of data, but those conflicts might be the things stopping the AI from deciding it's ok to roll over a motorcyclist.
This might not have been the ideal topic for it, but I really like the idea of stating the constraints of the topic in the OP, and hope you won't give up on it.