31–40 of 106 posts
When are we going to admit that the current level of understanding and technical capabilities are not sufficient for safe autonomous driving?
I fully admit to having done things in the car far dumber and more dangerous than this when I was young. I'm not excusing the Waymo, rather I'm putting it in perspective.
"Waymo you're going the wrong way!" and the human uni-cyclist steers in front of Waymo, knowing that it will brake, in an attempt to force it back into the correct lane.
When are we going to admit that the current level of understanding and technical capabilities are not sufficient for safe autonomous driving?
To be fair, I've seen many human drivers doing equally stupid things.
Just yesterday I got yelled at by an Italian police officer who was directing traffic at a busy junction (I was in a rental car). [Full disclosure: I stopped, wound my window down, we had a polite conversation, everything turned out fine...]
The key difference is that if I screw up badly enough when driving, I risk a monetary fine and/or losing my licence and/or getting sent to prison. For most human drivers these represent clear incentives not to screw up when driving.
Questions:
Have we agreed on suitable penalties when an autonomous vehicle screws up, given there isn't a single human driver to sanction?
Could the fleet get grounded?
Could the company get fined?
Could relevant executives go to prison?
Earlier quoted context omitted.
That's why they still are in research and not widely available.
that's funny, because when I watched the video I've got the impression it was dangerously roaming around the roads of San Francisco with hundreds of non-consenting people in close proximity
I suspect this is human error rather than a computer error. Waymo is a supervised system -- from what I've heard while there's little remote driving by humans, there is occasional real-time human labelling and confirmation. So the car likely phoned home to ask "what the heck is this?", and a human clicked on the "overtake" button or something.
Ok. So what are the consequences for this human operator and/or Waymo?
Human drivers make mistakes all the time, it’s true. But there are consequences for those mistakes. If a human drives into the opposite lane that’s a big problem, especially if they end up causing property damage or bodily injury. Why are the same standards not being applied here?
Earlier quoted context omitted.
You would expect a captcha that shows you two yellow lines, and asks you on which side of the yellow lines a car should be driving. Wether there are unicycles present is totally irrelevant.
As always there is a relevant xkcd: https://xkcd.com/1897/
Earlier quoted context omitted.
To be fair, I've seen many human drivers doing equally stupid things.
Yes tired, drunk or raging people do stupid shit. But the goal is not exactly to have autonomous driving at their level ...
I don't think it's unfair to point out that humans, quite often, also suck quite badly at driving.