Driverless cars just got a whole lot harder. This technical paper by @amirrosenfeld raises some profound questions about the robustness of #DeepLearning as a perceptual mechanism. https://arxiv.org/abs/1808.03305
-
I would hope any serious self driving effort does not do single frame detection but rather integrates information across time (and uses Lidar). Also agree somewhat this is contrived, like inputting giberrish into Google translate - not too surprising you get weird outputs.
Also people always say these challenges are contrived, won’t work in real world etc and then it turns out that they can in fact be replicated in the real world. @AmirRosenfeld’s case particularly seem plausible in real situations. What happens to deep learning at a parade?
-
-
Certainly it is instructive, yet another sign that no self driving tech should be end-to-end; but such concepts are just marketing gimmicks anyway...
-
On that we agree :)
End of conversation
New conversation -
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.