I'm sceptical the camera resolution and frame rate will suffice to tell the difference between a drunk bike rider about to weave and one that will keep course. Or a rear rack passenger about to hop off. A weave from someone just getting going along the sidewalk. Also with young children, I can tell from their finer posture nuances whether they're about to leap right in front of my bike. I see what their eyes are focused on, interaction with people, animals and objects. Give me 100 such situation and I will beat 2025 FSD at it. What could help FSD is the response time but they'd likely be too late already.other things too, for instance having seen more than my fair share of roadside weaves, how is the software at differentiating between 'just a weave flying by' and not an accidentally-tossed-into-traffic new-born baby?
gotta get out there and get all that juicy data mmm mmm good
I warn people about to step in front of my bike (voice, bell) all the time. In a car I'd have been slowing down before the point I'd like to alert the pedestrian. How can FSD ever be 10x better than me at seeing such accidents coming? An infant under a wheels is just game over.