Another Tesla on autopilot steers towards a barrier

Post by dingo_bat » Sun, Apr 08, 2018, 6:10 AM UTC

kwhitefoot wrote: post #409

m4x wrote: post #326

I don't understand why people use Autopilot, or why Tesla has released it. To use it safely and according to Tesla's instructions, you have to remain 100% vigilant at all times - so you might as well just drive yourself. And if you fail to remain vigilant, which is likely since you are sitting passively in the drivers seat, you might kill somebody. Where's the upside? Why on Earth would I want to use such a product?

> Where's the upside? Why on Earth would I want to use such a product? I use it because it makes my driving less stressful. The car and I work together which means I don't have to work so hard to keep to the speed limit, to keep a sensible distance from the car in front and stay in lane.

Post by tonyquart » Sun, Apr 08, 2018, 7:14 AM UTC

Post by w_t_payne » Mon, Apr 09, 2018, 4:55 PM UTC

Piskvorrr wrote: post #474

w_t_payne wrote: post #464

Earlier quoted context omitted.

A lot (but not all) of the costs are due to additional process steps that can be automated. I think we can dramatically increase safety and quality by selectively adopting some of the concepts from aerospace and aggressively automating them. Traceability; NLP for requirements etc..

That is probably true, and might help - but that only gets you so far, can't program your way out of everything . For high reliability, you'll need redundancy (and thus fallback and voting and whatnot), which will a) drive up HW costs, while also b) increasing complexity. I do agree there are some LHF opportunities to learn from aviation, but doing that is insufficient.

Post by Piskvorrr » Mon, Apr 09, 2018, 10:25 PM UTC

freerobby wrote: post #451

Piskvorrr wrote: post #430

Earlier quoted context omitted.

In such conditions, cameras are not unreliable, they're downright useless. Heck, even people have trouble driving during snowfall. I have a feeling that this is a massive set of scenarios that aren't even edge cases, yet SDVs are completely unprepared for them. ("No snow in coastal California - it doesn't exist at all!")

I think you may be imagining what a human sees in a foggy camera image, rather than what a neural network can see in it. In fact, atmospheric obstructions degrade Lidar's signal much faster than a camera's. If you check out NVidia's DriveNet demo from a couple years back, you'll see that they already have a vision-based NN that outperforms humans in fog and snow[1]. We can debate whether people should drive in those…

Post by ucaetano » Tue, Apr 10, 2018, 9:01 PM UTC

dingo_bat wrote: post #478

ucaetano wrote: post #476

Earlier quoted context omitted.

> Can you point me to a website or store where I can buy my fully autonomous waymo car? Not sure what are you mentioning, self-driving cars only exist when you can buy them personally? Let me know where I can buy a M1 Abrams, if you don't, I'll claim they don't exist! But actually, you can get a ride in a Waymo self-driving car already, with nobody on the driver's seat.

> Not sure what are you mentioning, self-driving cars only exist when you can buy them personally? No, but some body should be able to buy. Waymo is just vapor right now. It's an experiment, not a product.