Science!! Fucking magnets, how do they work?
- Thread starter Flunklesnarkin_sl
- Start date
Abefroman
Naxxramas 1.0 Raider
- 12,588
- 11,904
I can see this having military uses and for Trains or Boats. For everday driving, not a fucking chance. That car gonna be able to pull up to a drive thru window and not fuck it up? I don't want to have to reboot my car in the middle of the intersection. People trolling the fuck out of those cars by faking breaking and cutting them off. So many variables to driving in the real world. Also as someone said, the first time one of these cars kills someone the news will have a fucking field day with it. Lumie will then come here and tell us how the cars are being programmed by the lizard people as hit cars.
Tuco
I got Tuco'd!
- 45,474
- 73,558
1m20s
Unless google is lying about their results, an autonomous vehicle in normal conditions is a solved problem. Right now the major barrier is cost. Taxi drivers are not expensive, but the sensor payload for the google car is. I did hear they produced their own lidar that is 1/8th the cost of the velodyne they're using, but who knows if they're bullshitting or not.
- 16,428
- 7,439
Lithose
Buzzfeed Editor
- 25,946
- 113,035
They don't work off of Sat Maps, as far as I know. Mercedes/BMW work off of image recognition from cameras; it's essentially just like the human eye, it detects the edge of the road by the way it looks. They've tested it on every kind of road. Googles cars use laser sensors that bounce off objects for instant navigation (Which, from what I've read, is different from the camera recognition.)--In any case, again only from what I've read, GPS/Sat maps are only used for long range course plotting, not moment to moment driving; if it plotted a course into the river, the actual driving computer would stop it when it arrives at the river.
Apparently, Google's already lobbying for the car manufacturer to be at fault. They're pretty bold with saying that the car should never have a self caused accident--ever, and if it does, it's their fault, not the drivers. Reading over their publications; I must admit, I'm impressed. Their cars are more reliable than professional drivers at avoiding "high risk" zones for collision.
Like Tuco said; within the 70k miles driven thus far--the cars are significantly better than even professional drivers that have been data collecting to test them against. I mean, the reality here is that within the next 10 years, all cars are going to have auto-collision systems no matter what; right now, the high end cars already have it. Frankly, these systems (The ones in use in use today) are alreadydramaticallybetter than a human at avoiding cut offs and brake checks--just far superior. They work the brakes better, and react MUCH, much faster; and aren't affected by environmental sensor loss nearly as much. Even if we do continue to manually drive for the next 30 years? Our driving is goingHEAVILYsubsidized by computer control; meaning no matter what, in the next couple decades it's going to be a computer avoiding that guy cutting you off, not you.
iannis
Musty Nester
- 31,351
- 17,656
You know, not to be a dick, but that's about a credible as RJR claiming that tobacco is not linked to cancer. And it should probably be taken just about as seriously.
It will happen. And google knows it'll happen. And the smart thing to do is get way out in front of it and assert that it cannot possibly be their fault. Because until it does happen they're right -- it cannot possibly be their fault. And at the point where it happens that lie has already served its purpose to push the tech and to sell gomobiles.
I do believe the claim that they're safer. It's this 100% bit that's laughably self serving.
Ambiturner
Ssraeszha Raider
- 16,040
- 19,501
While it may happen at some point, it's pretty far from denying cigarettes are related to cancer.
iannis
Musty Nester
- 31,351
- 17,656
not that far really. They're both willfully false marketing claims aimed at avoiding legal obligations for defects in their product. That claim isn't bold. It's straight up hyperbolic.
I mean I know what you're saying. The defect in tobacco is intentional and avoidable, the unavoidable (and so far theoretical) defect in a gomobile will be unintentional. So yeah tobacco wins on the scale of evil but the strategy to deal with the defect is not that far different.
It seems further removed than it is because they're smart enough to start doing it before they have to.
I mean I know what you're saying. The defect in tobacco is intentional and avoidable, the unavoidable (and so far theoretical) defect in a gomobile will be unintentional. So yeah tobacco wins on the scale of evil but the strategy to deal with the defect is not that far different.
It seems further removed than it is because they're smart enough to start doing it before they have to.
Tuco
I got Tuco'd!
- 45,474
- 73,558
I think this is the big takeaway from demonstrations like this for the near future. Autonomous vehicles are a reality, but mass-market consumer-level autonomous vehicles are decades away. However the prevalence of safety systems and driver assistance is going to continue to increase.
Re: Camera vs lidar, google has a variety of prototypes but I'd expect they have both cameras and lidar onboard most of them. Virtually all high end autonomous vehicles use lidar and camera.
Eomer
Trakanon Raider
- 5,472
- 272
Out of curiosity, why do you say decades? Not that I agree or disagree to any great extent (I personally have no idea), but for what reasons do you think that they're going to take 20+ years to become mass market as opposed to say 5-10 years? Seems to me that the biggest thing holding them back is the legal framework and societal acceptance, not the actual technology or implementation.Tuco_sl said:
Tuco
I got Tuco'd!
- 45,474
- 73,558
Mostly societal acceptance (specifically demand) and cost of technology.
iirc from the above article it sounds like google wants to produce 100 of those vehicles. No idea what they want to do with them exactly since they said they aren't planning on selling them. If they start mass producing these specialized sensors it'll drive cost down, but for now most of the sensors are too expensive for mass-market consumer-level vehicles.
The other challenge I don't know if Google is attempting is creating a vehicle that doesn't require a team of engineers to actively manage. This is a big issue that's pretty boring so I haven't head anything about it. Does that blind guy borrow the google car for weeks at a time without contact from the google engineers? I doubt it.
Why would big auto give a shit? Self driving vehicles allows formorecars on the road.
Tuco
I got Tuco'd!
- 45,474
- 73,558
The only pushback you'd see is if someone outside the big three were to mass market an autonomous vehicle before the big three were close on one. Ex if Tesla Motors were to team up with Google and make a $150,000 autonomous Tesla S I think you'd start seeing some lobbyists from the big three push for some very stupid safety standards.
Lithose
Buzzfeed Editor
- 25,946
- 113,035
The difference is, the Cigarette companies didn't send their lobbyist out to say "If you get cancer; we're at fault, and we will be paying". But that's precisely what Google is doing--as they are working with law makers, they are showing that their cars have so much recorded information, they can detail out any accident--and if it's their fault, they should be liable, no one else. (In their 70k test, their car has one accident--but it was the other drivers fault. And they could show that clearly because of how much video/Lidar information they had on velocity ect.)
I trust people a lot more when they put their money where their mouth is. I have no doubt there WILL be accidents, I think Google does to. BUT the fact, again, that they are so readily lobbying that they be held accountable? In my opinion, shows they trust their product. (Unlike Tobacco; which spent millions trying to distance themselves from liability.)
Yep, this is how I actually see us eventually swapping to full automation. It won't happen in "one generation". Rather, it will slowly become more and more automated. This decade will be auto parking, emergency breaking and assisted collision avoidance. I bet ten years after that, we'll replace cruise control with "True Cruise" or some other stupid name--but it will essentially just be computer operated velocity, while the driver still stears (Computer does the brakes, acceleration). Then cars will come out with auto pilots that are voluntary to use. It will go from being switched on to look your cell phone quickly, to watching TV and then to napping (When people get really comfortable); and after that you'll start seeing cars without steering wheels become the norm.
But it's all going to come piecemeal, I think. As each sub-element of driving is replaced, people will become more and more accustomed to simply being passengers--until they won't think it's a big deal that their cars have become mobile offices or living rooms.
Tuco
I got Tuco'd!
- 45,474
- 73,558
It's important to note that Google isn't planning on producing consumer vehicles, so while they're making themselves more liable if their research vehicles collide with someone, they aren't setting themselves up for a consumer-product liability burden in the future.
Lithose
Buzzfeed Editor
- 25,946
- 113,035
True, but the way the laws will be written now; puts manufacturers on the hook. So I still feel like Google is pretty confident in their product, because they are pumping money into it knowing they are going to be able to sell it despite manufacturer taking on that liability. (Of course, laws can change. I'd become a lot more dubious if they did down the line. But for now, I'm pretty optimistic.)