Nice share, interesting that mobileye is commenting.
If I can classify the person 1 second before the vehicle hits( which watching the video I can ) then a computer potentially can. I don't think this means Uber should have been able to avoid this collision. I say this because guessing a proper reaction time is very difficult to do. When I read that they were able to classify the person 1 second out, my question was immediately "How does that relate to humans and other AI systems?". Here is what I found for humans,
PRT means perception-reaction time.
1. A "standard" or "generally accepted" PRT cannot and does not exist; (
http://www.visualexpert.com/Resources/reactiontime.html )
The first point is that it's very difficult to guess what the PRT is because situations vary, A LOT.
"[PRT] can range from .15 second to many seconds. It is also highly variable. In many cases, the very concept of perception-reaction time simply doesn't apply2. "
The other interesting point is this:
5. PRT generally does not explain why a collision occurred. It is not a cause, but rather a symptom to be explained. The real cause lies in the answer to the question, "
Why was the PRT insufficient?"
Why did the AV have to respond to this lady in under 2 seconds of possibly being able to detect her? Because she was walking her bike across the street where she shouldn't have. Add to that she was doing this at night and was with a bicycle, makes it that much harder to respond to. Could a person have braked in time? Maybe. Could an AV have braked in time? Maybe.
For reference here is a screen shot from the video taken about 3.5 seconds before collision. Can you spot the lady and her bike? There is a lot more than detecting visually (LIDAR ect...). However this is the only data we have so far.
It doesn't seem as obvious as some of you seem to feel. It's a very complicated problem/investigation. These types of incidents will inevitably increase as AVs become more widespread. It's important that as a society we react rationally and not emotionally.
Final point:
This little exhibition by Mobileye, while it should be taken with a grain of salt, at least gives a hint at what should have been happening inside that car’s brain.
I think the first half of this sentence should be highlighted. Mobileye is a pitching their product. This is going to be a trillion dollar industry and there is blood in the water. There is very clear bias from the source and I don't think they paint a complete picture. Their insight does help to begin to paint a complete picture though, as the bolded highlights. Good read.
