5 ms·
> crashing on a rare edge case (i.e. not well represented in the training data) This isn't a case where it makes any sense to try to extrapolate from training
by thaumasiotes 6y ago
> crashing on a rare edge case (i.e. not well represented in the training data)
This isn't a case where it makes any sense to try to extrapolate from training data and call it a day. You don't need any training to avoid hitting stationary objects.
Compare this comment from 2018: https://news.ycombinator.com/item?id=17015528 https://news.ycombinator.com/item?id=17015528
> The first job of automated driving is to drive only on flat surfaces. Doesn't matter why it's not flat. Then you worry about where to go. That's how the off-road DARPA Grand Challenge vehicles had to work. "Flat road" is a pure geometry thing. Not much AI needed.
Why use "AI" to solve a problem you don't have in the first place? This is the wrong tool for the wrong job.
(Compare also GPT-2 chess, in which a model trained on transcripts of chess games isn't able to stop itself from making illegal moves. The solution is straightforward - give your chess model a model of the rules of chess! And don't attempt to drive into spaces you cannot physically occupy.)
- gpm 6y ago> You don't need any training to avoid hitting stationary objects. You do, because you need to correctly conclude that there is going to be a stationary object there with high probability. As it turns out, that's difficult to do perfectly.
- thaumasiotes 6y agoI'm having difficulty following your train of thought. It can be difficult to make conclusions about the future position of mobile objects. It can't be difficult to make conclusions about the current position of stationary objects.
- gpm 6y ago> It can't be difficult to make conclusions about the current position of stationary objects. You need to know that there is a stationary object in the first place, of a kind you care about (not a plastic bag). I.e. that it's not actually a moving object, an optical illusion, etc. You also need to know that you are going to end up moving through it (it's not a parked car that you will avoid by following your lane). That's... not trivial. (going to be was probably poor phrasing on my part, sorry)
- Daho0n 6y agoYet even microcars can do it. Like the Toyota Yaris and VW Up!.
- ClumsyPilot 6y agoYou can determinine a wall accurately without AI, we have for years.
- Arnt 6y agoIf one doesn't need any training to avoid hitting stationary objects, why do so many human drivers hit stationary objects? I can see arguments that the program shouldn't hit stationary arguments, but the ones I see apply to humans too with only minor changes. Is there a relevant difference between what's "training" for an CPU driver and what's training for a human driver, and if so, what?
- gpm 6y agoHumans have failure modes computers don't. Falling asleep. Distracting phone call. Distracting naked person dancing in the window. Computers have problems humans don't too of course... that's why we get these articles about "ridiculous crash happened".
- thaumasiotes 6y agoYes, computers have problems that humans don't. But this isn't one of them. This is a problem that humans don't have, and that computers also don't have. (Or rather, a problem that humans do not have trouble telling computers how to solve.)
- Arnt 6y agoYes, yes, humans have failure modes that computers don't, and those failure modes do lead to some problems. That's not relevant. I repeat myself: "I can see arguments that the program shouldn't hit stationary arguments, but the ones I see apply to humans too with only minor changes." That humans have extra problems doesn't really matter, so long as humans require training to cope with those problems that are similar to those computers have. For example, computers may use a vision device that sweeps across the surroundings and then use lots of postprocessing to build a complete scene. That's similar to what the human eye and brain do, except for the details of how the sweeping sweeps and how the postprocessing puts together a picture.