I've done a bit of system integration with bill acceptor machines, and they should be fine. They're not looking for visual spectrum stuff, or comparing a bitmap, they're checking for a finite number of specific features. Usually, it is 9 or 11 small spots that are each checked for one thing. None of them are the face visuals.
Slashdot videos: Now with more Slashdot!
We've improved Slashdot's video section; now you can view our video interviews, product close-ups and site visits with all the usual Slashdot options to comment, share, etc. No more walled garden! It's a work in progress -- we hope you'll check it out (Learn more about the recent updates).
The smart cruise control already is being installed in regular cars, and isn't even called self-driving. Since the human is still in the seat, and does small parts of the work, they just call it "intelligent cruise control." Cars that have it, it is almost impossible to read-end the car ahead. Current models have lane-assist during cornering, too.
Life passed up your prognostication... A-priori!
Existing self-driving cars have trouble with snow, and heavy rain, but work fine in most weather conditions, including most precipitation.
The problem with the design of current cars is that a deadly defect is built in; a steering wheel allowing direct, manual human control!
This design flaw causes the vast majority of all automobile accidents.
The google stuff is fully autonomous already. They have steering wheels for 2 reasons: they're required to by the State of California, and they're prototypes and they need to be able to steer them around with some of the equipment turned off.
A consumer model doesn't need to be able to still be operated after a malfunction. It can just shut down and call a tow truck.
You have no idea how long I'll live, you insensitive clod!
Nonsense, the insurance would never get shifted onto the manufacturer, because maintenance happens after that, and is part of the accident risk. The vehicle owner will be required to pay for ongoing insurance, and will be required to be licensed (for the purpose of purchasing insurance!) until the car insurance requirements get shifted so that non-licensed drivers can buy insurance on self-driving cars.
But I like the taxi analogy. And it keeps working too; just like in that situation, the most likely thing an idiot does is grab the wheel and crash into something, or run over some pedestrians while trying to save their own ass; even though if there are modern airbags and collision detection, the taxi is likely to stop in time anyways. Even the human-driven new cars are getting safety features... where it is the computer that takes over to stop the vehicle when the human screws up, rather than the other way around.
The computer, having detected the loss of brakes, will not ask the magical human to save it, it will just activate the emergency brake and flashers, maintain control of the vehicle, stop in the lane, and call roadside assistance.
Then the car will ask you to do a manual chore. Pushing the car off the road. And there is absolutely nothing about the brakes going out that makes the human suddenly better at controlling the steering wheel, avoiding accidents, or obeying traffic laws.
Humans are too stupid to operate cars now that they don't have to. They actually even think they're better drivers... than the existing self-driving cars! Only in the snow, Bub. Only in the snow.
... car drives license. License drives you.
You don't have to be licensed so that you can pay attention "in case something goes wrong," though you'll probably be expected to push the car out of the roadway if physically able.
The reason you have to be licensed is that if the car malfunctions and creates an insurance claim, there is lots of existing legal precedent related to insurance liability that means the insurance company will require a licensed driver, until the laws are changed by people not scared of self-driving cars. That will take up to 50 years after they're first legalized.
Also, in most states there will still be a requirement to "exchange driver info" after any accident; including one that is not the fault of the self-driving car. There are probably lots more examples. The first steps will only say, "yes you can do this new thing but you also have to follow all the old rules too."
Human reaction time is slower than computer reaction time, so it is unlikely that you'll have more than an emergency shutoff to deal with mechanical problems. The idea that the human will be interrupting the computer in real time while in traffic to correct some mistake, that is a total joke. The humans would cause a wreck almost 100% of the times they tried it! I don't drive very much, but I'm out there enough to know how awful human drivers are. And that's when they're giving their legally-required full attention to the task.
By that theory, nobody ever drives anywhere, because there could be an unexpected road closure. I go lots of places where there is only one road, and if it is closed (which happens) then you can either try the next day, or drive an extra 250 miles. I've never once heard of it as a reason people don't go to those places. Even a doctor isn't going to stay in town and never go to the beach on a day off because of some small percent chance the road would be closed.
If the car is leased with a service agreement (likely for early versions) then you probably just call roadside assistance if it strands you, and they send a tow truck, same as AAA.
Gosh, nobody would even play golf, because of the lightning risk.
I know, right? Like, how can you drive your license around if you're not driving. Oh wait, but it says the car will be driving. Wait, I don't even drive my license as it is, I drive my car!
If it doesn't drive, I'll agree it isn't self-driving. But if it isn't licensed, then I can only agree it is not self-licensing.
It is a bit of a "no-brainer" that at first a licensed driver will be required, for the purpose of integrating normally into the existing insurance law and regulation. Only after they're common will the laws be streamlined.
Really? You spend time to reply and reply and reply, and even call me "stupid," you must feel really smart to magically know how people feel. You seem to think I "wasted" my time, or that you personally were who I was writing for. Guess what, I've been online since before the internet was public, and I've never written you even one personal message.
As for feeling "stupid" or not, lets put it this way: I stand by my analysis, and further, you didn't even succeed in addressing it. None of the intended readers need your name-calling to read the ideas and claims and decide what has value. To take it to another level, I can point out that my analysis is actually mainstream, past-tense stuff about an already-shrunken industry.
Some hand-waving about "anecdotal shit," well, like you say, we could look up the numbers.
So, your point is that... because something exists and is important to niche users, it must not be true that it isn't used "a lot?" Huh?
You seem really hung up on absolutes. But I didn't use any. I didn't (and wouldn't) say that trains aren't used, I said we don't use them very much. And indeed, most things that were moved by train in the US in the past are now moved by trucks. If you ever visit the USA, I recommend you take a side trip through a few random, 30+ year old industrial parks. What you'll find is railroad crossing on the roads, but weeds on the unused tracks. Even places where the tracks are still used, they'll usually have weeds on the business access sections, because nobody cares. You'll find loading bays next to the tracks converted for loading trucks. Are there still trains? Yes, of course.
Outside of certain regional commuter routes, passenger trains are used as a luxury alternative to... buses. They do not often arrive more than 10% faster than a bus. And nobody cares, because if they were in a hurry, they would have flown.
"A couple of recent accidents" only tells you that trains exist. It tells you nothing about the relative number, or what is usually used for shipping.
And if you spent time on US roads, you'd quickly realize that there are a large number of oil tanker trucks. You can stand on the side of a major freeway and count them. On the west coast there are numerous places where there is only 1 major north-south freeway, and 1 major north-south railway, side by side. You can stand there and count oil cars. You'll see a few on the tracks; maybe even 1 train car for every 100 trucks. Presumably if you live right next to an oil distribution facility you'll see more rail cars than that. We have various petroleum distribution facilities in my town, because we're at a rail junction that connects to other regions. So they built them here, many years ago. But guess what? They removed all but 1 track of rail access! And that one has weeds. I don't think I've seen a train parked there for over 20 years. But if you have to drive in that neighborhood, you encounter a large number of oil and gas trucks. Someday they'll build a new facility... closer to the freeway! But for now they only lose 10 minutes by being stuck out by the railroad tracks, and they already have the storage tanks.
Oh, gosh, you saw a train on tee vee, they must be falling from the sky! Golly gee!
It should be noted that our electric light rail are almost all modern, except for BART in the Bay Area. (No faulting BART, they pioneered the field) Trains are well used... at the municipal level, for moving people short distances. We're not going to build expensive high speed crap for that. They're faster than driving, and have bike racks. They connect the `burbs.
Your theory about the markets represents considerable original research, I don't think it is useful to just assert those ideas as facts or as changes that have happened in the markets. You're obviously aware of specific pieces of equipment that used to have a capability gap that no longer does; you mentioned CNC mills, for example. But just waving your hands and asserting the US no longer has manufacturing sector equipment exports that are difficult to replace, well that just shows ignorance of US exports in that sector. It isn't magic, so it doesn't apply to random things like CNC mills, which are somewhat trivial.
If you ever visit the US you'll find out how funny the trains comment is; we don't really use many trains. It would be somewhat predictable that modes of transport we don't use don't have much investment. It has nothing to do with complacency. I doubt the existing horse-drawn carriages are state-of-the-art, either. Maybe you can sell the sleigh industry on upgrades.
US has 14% of the global machinery equipment market. Export leaders included: construction machinery, engine equipment, turbines and turbine generator sets, and agricultural equipment.
Nonsense, it is absolutely do-able to have a realistic understanding of your actual security. The impossibility of secrecy does not refuse the usefulness of true information.
And I agree, there are few things more secure than the best available open offerings. But well financed law enforcement and security agencies are outside of that security. That the attack vectors are not revealed as such in the media is meaningless when the necessary capabilities are know to be possessed by them, and where their tactics are considered secret.
Luckily once somebody being honest about the security situation understands all that, they can just get on with locking out black hats, which is what the software can do; protect you from those without legal recourse to tell your ISP what to do. You just can't protect your privacy from government actors based solely on technology. They're in the position to MITM anything, to keylog anything, to anything anything.