5 ms·
There are a lot of similar failure modes in humans. Sometimes they're are used for good, like making drivers slow down: https://www.insider.com/optical-illusio
by fyp 6y ago
There are a lot of similar failure modes in humans.
Sometimes they're are used for good, like making drivers slow down: https://www.insider.com/optical-illusions-3d-crosswalk-driving-2019-5 https://www.insider.com/optical-illusions-3d-crosswalk-drivi...
But other times they cause crashes: https://imgur.com/a/kYr94 https://imgur.com/a/kYr94
- Barrin92 6y ago>There are a lot of similar failure modes in humans. there are lot of similar failure modes in individual humans. The difference is individual driver errors are not correlated, a bad reflection tricks a few humans at once instead of tens of thousands of cars running the same model. There is significantly more brittleness in a fleet of cars than in a population of humans, because the human population has diversity in judgement and experience. This is not simple to fix because this uniformity is actually a feature of automated systems, it makes them explainable, produces expected results and conform and cheap, training a human for 20 years is more expensive than training all the cars once. However it also makes them collectively vulnerable, which is why heavy machines tend to be locked away on factory floors.
- newen 6y agoTrue. People need to stop comparing self driving cars favorably to the worst performing humans. Gives companies license to release these badly performing cars into the streets.
- hughisthis 6y agoThey are assistive systems. The only thing that's badly performing are the users who misuse them and get themselves into trouble.
- gmadsen 6y agohard braking or swerving for clear false positives is not very assistive... it seems to be the opposite of assistive. How can a person know all the weird things that will make a computer vision system badly fail. It shouldn't be used at all if that is the case
- Hamuko 6y agohttps://www.snopes.com/fact-check/road-runner-tunnel-crash-rumor/ https://www.snopes.com/fact-check/road-runner-tunnel-crash-r...
- rtx 6y agoThose examples in the first line are so bad, I hope they don't catchup.
- thdrdt 6y agoSome places in the Netherlands tested those. Imho they don't work as advertised. On a picture they might look 3d but when you driving towards them your brain knows the perspective doesn't change and doesn't trick you into thinking it is 3d. The effect is more like: What is this for strange art!?
- mcintyre1994 6y agoI wonder if these self driving cars would recognise those 3D crossings as the same as the normal thing. I could see it looking quite different to a neural network or something looking for features and a bit overfitted.