Uber grounded its self-driving cars because of an accident, where the algorithms failed to account for a real world condition. But...
-
-
My former company flew a reconfigurable neural net flight controller on the VISTA F-16 in the 90s/00s. That tech still isn't fielded.
-
It worked fantastically! But V&V requirements were too huge. And flight is a simpler case than driving.
-
Convolutional NN stuff is great for object detection and classification, but ultimately it's a controls problem to solve. And that's hard.
-
it's interesting as it's like human drivers in that regard. not really testable / reliable 100%. needs a mentality change?
-
We have a better understanding of how human learning works.
-
If we say "by the way in this circumstance you should do this" we know roughly how people's driving will change.
- 9 more replies
New conversation -
-
-
Fact is, no algorithm is going to account for 100% of all real-world situations. A line just has to be drawn somewhere.
-
Uber, like all self-driving programs, will have to decide when 99.998% of better-than-human driving is "good enough".
End of conversation
New conversation -
-
-
this is why I think they should start by creating self driving lanes. This would help weed out some of the unpredictability.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.