California To License Self-Driving Cars
DevotedSkeptic writes "Californian senators have passed a bill that looks set to make the state the second in the US to approve self-driving cars on its roads. The bill was passed unanimously by state senators, and now hits the desk of governor Jerry Brown, who's expected to sign it into law. It calls on the California Department of Motor Vehicles to start developing standards and licensing procedures for autonomous vehicles. 'This bill would require the department to adopt safety standards and performance requirements to ensure the safe operation and testing of 'autonomous vehicles', as defined, on the public roads in this state,' it reads."
Considering half the drivers there don't seem to be paying attention to their driving, self-driving cars would probably be a huge improvement.
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
...and they'll work the security into it after the first major hacker-caused pile-up.
Not safe right now... the difference being is that we can make continually make self driving cars more safe, since driving only requires a set of rules and environmental awareness. Humans will never become more safe, in general, because they are inherently mistake prone due to fatigue, poor judgement, distractions, intoxication, and many other factors.
Just look at the wonders of automated flight. Most airline accidents that aren't due to terrorism or mechanical malfunction are due to pilots overriding the autopilots.
while(1) attack(People.Sandy);
Oh come on. Both articles you link to are full of nothing but conjecture and opinion, and both are about the same accident. Plus, anecdotal evidence tells us nothing. What I want to know is: how many accidents on average do Google autonomous cars have per mile, and how does that relate to the average for human-driven cars?
My Photography - http://ian-x.com
The Deathlings (comic) - http://thedeathlings.com
And your driver's license lets you vote in CA, does that mean these cars get to vote? Can they vote themselves "car friendly" politicians? Will we be talking about "vehicle rights" in the next election?
In a panic, will we try and pull the plug?
I want to delete my account but Slashdot doesn't allow it.
If a self-driving car crashes while not self-driving, I would blame the driver, AI which isn't running.
And either way, would you rather have more accidents, which you could "do something about".... or less accidents, which you can't really do anything about? Although humans like to have full control, they always end up in more accidents.
what about stuff like code review and liability?
Now there are 2 big liability parts criminal liability and civil liability.
and no who makes the car and or the software coders who make the code can't hide behind a mandatory arbitration or an eula.
Even more so if say the car hit's some thing out side of the car.
I'm glad I didn't have to be the one to Catch that both articles were talking about the same accident, and the first article even links to the second.
Two articles referencing the same incident! My god these aren't safe!
ITT, Microsoft shill tries to discredit Google yet again.
This was a single incident caused by human error. Go shill elsewhere.
There are a lot of interesting legal implications for these self driving cars but all that a side I dream of the day when a drunk can stumble out of the bar and fall into the back of his car and wake up in the drive way of his home the next morning.
Anyone who seriously moves to prevent the self driving car from becoming reality regardless of how safe they are is simply against saving lives. I'm sure most people will wonder how anyone could be flat out against self driving cars but people like that do exist and at some point this will move from a legal issue to a political issue when it starts looking like mass adoption might happen and these people will come out.
Left, maybe. Right, maybe not!
Don't put any ethanol in the tank! Or you'll see a lot more DUIs...
“He’s not deformed, he’s just drunk!”
Liability won't be an issue. If (when) these are truly safer than humans driving, the insurance industry will be falling over themselves to insure the cars. It'll be pure profit for them, and the incredibly rare incidents that pop up will be more than covered by all the other people driving problem-free.
With reports of Google's self-driving car crashing left and right how could anyone want to be in one of these vehicles? They just aren't safe. When something happens when you're driving then it's at least your fault and you could do something about it, but not in self-driving cars.
Was this meant to be sarcastic? Both of those posts referred to the same accident. These cars have logged hundreds of thousands of miles, with ONE accident(which may well have been human error). That's far, far safer than the average human driver. If you're in the drivers seat of the self driving car, you CAN take control of it should you feel the need, too.
However, realistically that's not going to be useful. The car will be better at accident avoidance than you are - it's not that big a programming challenge to achieve that. People don't like to admit it - it bruises their delicate little egos - but the car knows *exactly* how fast every car around them is moving, their acceleration, and can put itself exactly where it wants to be every time. No delayed reactions due to inattention, no slight overreaction due to panic.
Yes, self driving cars will be involved in accidents, and will be at fault, from time to time. This does not make them less safe - it's inevitable, particularly when human drivers are involved as well. Human drivers, on the other hand, are extremely unsafe. Everyone wants to think that they are special, and unlike everyone else they're awesome drivers, but the reality remains that human drivers are in accidents extremely regularly.
Don't get me wrong. I'd hate to be in a robotically driven car. Logically, I know I'd be much safer than with a human driver, but I'd be enormously squirrelly about the whole process. And, of course, I love driving - I'd never be comfortable giving that up to a machine. I consider myself a good driver, too (like everyone else), and I've never been in an accident for which I'm at fault, but I can acknowledge that there have definitely been times I've driven with far less than ideal circumstances. Distraction, emotional distress, tiredness, ill health, the list goes on an on. In all those cases, I'm less than 100%.
Meh.
I assume that in order to actually have one of these things drive on public roads, insurance is required? And which insurance company will insure this relatively incalculable risk, and at what price?
Are you kidding? Maybe we should have given up on everything that just wasn't safe at the time.
Strapping a combustion engine to a frame with wheels?
Rockets used to launch astronauts into space that have a chance of blowing up?
Medicine and procedures that had higher failure rates once upon a time?
We shouldn't have even tried! I am sure if I had time I could name two dozen more.
When this technology is mature it will save lives. Human distraction and lapses in reason kill, and this will eliminate that. If these can have 10% less, and I think that is a vest underestimate, crashes than their human counterparts isn't that reason enough?
autopilots acting on bad data or coding issues??? had lead to crashes.
What about that air show crash where you had stuff like
Thus he may not have heard these warnings (and thus any other warning or alarm as they sound in cockpit and not always in the headset).
that black boxes had been tampered with. (maybe to cover up the airbus issues with it's autopilot)
In the month prior to the accident, Airbus had posted two Operational Engineering Bulletins (OEBs) indicating possibilities of anomalous behavior in the A320 aircraft. These bulletins were received by Air France, but were not sent out to pilots until after the accident:
A320 crashes
http://www.airdisaster.com/cgi-bin/view_details.cgi?date=03221998®=RP-C3222&airline=Philippine+Airlines
The aircraft overran runway 4 while landing. A malfunction of the onboard flight computers prevented power from being reduced to idle, which inhibited thrust reverse and spoilers from being used. The offending engine was shut down, and brakes applied, but the aircraft was unable to stop before the end of the runway
It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
If a self-driving car crashes...
Make sure to get the breathalyzer sample from the tailpipe... that corn alcohol is powerful stuff.
“He’s not deformed, he’s just drunk!”
I'm expect a lot of political trouble from trucking unions etc. Driving is many peoples livelihoods.
Left and Right you say? It's the same incident in both links (indeed the C|net article is based on and points to the Jalopnik post).
Also, from your own source:
Updated 3:51 p.m. PST: Google would only give me a further one-line statement. A spokesman said: "The car was in manual mode at the time. We have confirmed it in our logs."
That's 3:51 pm August 5th, so it was cleared up by the time you got the link.
I'm just as concerned about this, but your post is downright deceptful.
This post contains no rudeness or derision of any kind. All arguments are friendly. Terms and exclusions may apply.
It's clearly just a matter of time until automomous cars are head and shoulders safer than those driven by people. Once this happens, adoption will be driven by the insurance companies. It will become prohibitively expensive to drive your own car.
I actually look forward to this, and wonder how it will change the interior design of cars. Will we turn the front seat around and go for a more social living room style arrangement? Will we dispense with the view from the front windshield in favour of an immersive large-screen TV? Beds for those long drives? Will we have refrigerators and microwaves so we can get breakfast on the morning commute? The possibilities are awesome.
Your theory isn't holding up in the face of the data. Googles Cars have logged hundreds of thousands of miles and have one accident caused by human error.
300k miles.
1 (non at-fault) accident.
A driving recod like that would probably rank in the top decile of all drivers in America. Stop spreading FUD (seriously, what did autonomous cars ever do to you?).
I have left slashdot and am now on Soylent News. FUCK YOU DICE.
They have well over 300k miles.
Source
I have left slashdot and am now on Soylent News. FUCK YOU DICE.
It's not as if cars don't already contain software today that affects how it drives, from anti-lock braking to engine control to powered steering to... It's simply the next (admittedly, big) step, not going from complete manual control to complete automatic control.
Donate free food here
Where are you getting that the average human driver has an at least one accident every few hundred thousand miles? I wouldn't call this "far, far safer" yet. It has the potential to be.
Also, most of the tests have been in still fairly controlled environments. Meaning, the car wasn't woken up in the middle of the night to get a pregnant woman to the hospital quickly over dirt roads, past nighttime street-racers, etc... Loads of "special cases" exist in the world of cars. It will be quite a long time before we have a really solid understanding of their viability. Right now, a "typical commute" is probably the safest use, or even for standard-route delivery vehicles without a high time sensitivity. Even better if certain roads / routes / lanes get set aside for autonomous vehicles only, which would make them even safer and more efficient.
Come play Moral Decay!
Car drives you!
You know you were all thinking it.
what about criminal liability the insurance industry can't cover that.
And who will go to jail if say a auto car some how things a small kid on the street is a bird or road kill and runs it over?
Ew, that Cnet article reads like a gossip magazine, or a script from Glenn Beck. Isn't there anything more... credible?
What?
...street view "live" !
It will be quite a long time before we have a really solid understanding of their viability.
Not at all. Airline statistics tell a big story. They are incredibly safe now. Not too many computers are flying the plane into a mountain.
“He’s not deformed, he’s just drunk!”
Google has not proven it does not tinker (read - assist via local or remote operator) with the software to pass qualifications. NHTSA should have more controls over this, namely up to the level of FDA regulations. Considering NHTSA's record with Toyota "unintended accelerations" where they had to recruit NASA engineers, the necessary qualifications are not there yet.
the self diving cars better have the same level of code review that autoplot software get's.
and even if that you can still get errors like this
http://www.airdisaster.com/cgi-bin/view_details.cgi?date=01201992®=F-WWDP&airline=Air+Inter
While on approach into Strasbourg the aircraft impacted the side of a mountain. The cause of the crash was found to be a faulty design in an autopilot mode selector switch which led the flight crew to inadvertently select a 3,300 foot per minute descent rate on the approach instead of the desired 3.3 flight path angle.
or this
http://www.airdisaster.com/cgi-bin/view_details.cgi?date=09141993®=D-AIPN&airline=Lufthansa
The aircraft skidded off the end of the runway during landing. The aircraft touched down with sink rate low enough that the onboard flight computers did not consider it to be "landing," which inhibited thrust reverse and brake application for nine seconds.
http://www.airdisaster.com/cgi-bin/view_details.cgi?date=03101997®=A40-EM&airline=Gulf+Air
A flight control failure at V1 caused the crew to abandon the takeoff, with deceleration beginning at V1+8 knots. The aircraft overran the runway, causing the nosegear to collapse. The flight control problem was traced to a faulty microchip in the aircraft's Fly-By-Wire system.
how many accidents on average do Google autonomous cars have per mile, and how does that relate to the average for human-driven cars?
not enough data. Sample size of Google autonomous cars too small.
Nice trolling, don't know why you're at 2, as you reference the same article twice and when looking at its contents, it brings no meaningful facts to the table, only conjecture and opinion, ergo, they do not support your conclusion that these cars aren't safe.
You might be writing history, joining a certain individual who proclaimed: 'The horse is here to stay but the automobile is only a novelty—a fad.'
The fact that an AI doesn't fatigue, doesn't text, doesn't lean over to grab a water bottle and cause a head-on collision, doesn't fall asleep, doesn't drive drunk, doesn't run red lights on purpose, doesn't forget to signal, doesn't speed and has 360 degrees of vision and laser-radar object detection and processing all this information at once where a human has to rely on his eyes and brain and reaction speed, all of which are affected by his physical condition and which deteriorate when he gets older. The AI will sooner or later replace a human driver. If you will, you could consider the current form a very sophisticated version of cruise control, where a human supervisor is still required.
But I could see driverless taxis in Vegas taking you from your hotel to a casino (and back), by just stating your destination, confirming it and paying with your NFC enabled phone.
Hell, I have driven well over 300K miles without an accident....
Perfectly safe is a stupid metric to even consider given it's ill-defined, and would be a bad metric even if it were well defined.. They're safer than human drivers.
Who is doing the code review on your brain? Serious question. People crash cars all the time and the automated cars have already been demonstrated to be at least as safe as the best human drivers. Are automated cars perfect? No; but so far their record is.
-1 overrated isn't the same thing as "I disagree".
yet EVERY airplane has a trained pilot waiting to take control....
Can't wait to jailbreak my Google WheelDroid car to tinker with its firmware and make it go faster by compiling -O9.
why have driverless taxis when they can build the monorail to the airport and have a downtown link.
OTOH, I don't have race conditions and can reboot myself if needed. My MTBF is about 50 years - better than pretty much any computer based device I've ever used.
Upgrades are a bitch though, I'll give you that.
Faster! Faster! Faster would be better!
Yes, waiting... But the numbers speak for themselves. Pilot error still is and always will be the biggest killer. The driverless car will make riding in them almost as safe as the airliner. I would wager that the death toll will be less than 1000, maybe even 100, per year when these vehicles become ubiquitous.
“He’s not deformed, he’s just drunk!”
Nobody. Same as now.
Faster! Faster! Faster would be better!
Self-driving cars will eventually be the majority.
Driving 20 over the speed limit may make you get there more quickly, but not having to focus on the road for the whole trip will make the trip more enjoyable and will make it feel like you get there more quickly.
Who is doing the code review on your brain?
From what I've been able to determine, most people seem to be written in an early variant of Visual Basic.
Wake me up when they've been refactored.
Faster! Faster! Faster would be better!
If all cars are self-driving, then we can happily increase the speed limit -- and probably by a lot!! We might even get a scenario where one speed limit applies to humans, and another (higher) one applies to computer-controlled vehicles.
char*f="char*f=%c%s%c;main(){printf(f,34,f,34);}";main(){printf(f,34,f,34);}
It used to be the trained pilots who drove the planes into mountains, while these days they do it with much less frequency, thanks largely to the autopilot technology.
I would think that normal people without such rigorous training will be benefited more from the autopilot technology if they drive airplanes or cars.
What did you think the "Motor Voter" bill was about a couple of years ago?
Meanwhile, the DMV seems to have decided that the robots don't speak Spanish, so it's ok to let them drive.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Eventually sure. It's gonna be a trainwreck getting there, though.
Even if you were to combine accidents from software bugs, driving the speed limit, or some other factor, I'd absolutely bet that they would total far fewer than accidents by drunk drivers, falling asleep at the wheel, using cell phones, talking to passengers in the car, highway hypnosis, misunderstanding street signs, or lack of knowledge about right-of-way. Pick one.
They don't have to be safe, as nothing, not even laying in bed, is completely safe. They just have to be safer than what exists now. That is a pretty low bar to reach.
while(1) attack(People.Sandy);
So you are an outlier, among the thousands here on slashdot that didn't respond the same. I've read the average person gets into an accident every 6.5 to 10 years depending on location, or 78-120k miles.
Google cars have had over 480k miles accidents-due-to-system free. There was one accident when one was rear ended while stopped at a redlight, and another when a human tester decided to override the automatic driving and drive it himself.
while(1) attack(People.Sandy);
The problem is, all cars are NOT self driving, and probably won't be in our lifetimes.
Somewhat off topic, I know. But if we're going to have auto-driving/piloting, then wouldn't self-navigating ships be more important, from a practical perspective? (Though I can see the fun and technological offshoots in designing self-driving cars.)
Self navigating cargo ships might need to be be piloted manually when leaving and entering docks (at least to start with), but in the open oceans they could auto-navigate and be centrally monitored.
Open water piracy would take a dent as there would be no crew to kidnap, and there would be no incentive for ship owners to follow pirates' demands to reroute ships. After all, if you're going to lose a ship and its cargo either way, then might as well do it by not appeasing pirates.
It would also mean that ships would not be piloted by crews who try to navigate tricky waters to cut corners.
Most airline accidents that aren't due to terrorism or mechanical malfunction are due to pilots overriding the autopilots.
[citation needed]
Hint - there is a difference between "pilot error" and "overriding the autopilots". There innumerable factors, input and output, that contribute to a safe flight, only a small fraction of those are handled by an autopilot. Arguably, the same holds true for automobile travel. Then again, I live in a city where the humans regularly demonstrate that machines are better drivers.
... since driving only requires a set of rules and environmental awareness.
If you replace the word driving with any verb your statement is true. I can't imagine the mayhem this will mean - in part because I don't think anyone knows how 'autonomous cars' will be implemented. In the simplest since we are talking about putting our safety into the hands of something engineered. Fair enough - we do that every day of our life - most of it unwittingly. However, engineering it seems to me, is best when solving a specific problem we're having with a conflict between our sets of rules and the environment. Homes and bridges were probably the first things experienced, if not hunger or thirst, solved by engineering. I can't for the life of me think of a problem solved by 'autonomous cars' that won't create more problems requiring more engineering.
The Luddites were ahead of their time.
They still are better than a woman driver.
I already have an autopilot in my car which is constantly giving audio prompts and attempting to take control, but enough about the wife. I for one welcome our new autonomous car overlords - as at least I can kick the wife out... heh
Saftey critical systems require less then 1 fatality in 10^9 hours of continual operation. That's 114000 years of non stop operation. Good Luck.
I disagree - human driven cars have become more, and less, safe with engineering, and other, advances over the years.
Collapsible steering column, steel belted radial tires, seat belts, crumple zones: more safe
Crowded roads, mobile phones, multi-hundred horsepower engines: less safe
Anti-lock brakes, air bags, tougher DUI laws, better roads: more safe
Texting, e-mail, twitter and Facebook on mobile phones: less safe
It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
That's pretty short sighted. As much as I enjoy driving my car myself, I imagine automatic cars getting their own lanes and higher speed limits within the next 10-20 years, it's one of the few things I like about their coming (that, and less idiots to contend with on the road.)
We're seeing the birth of the San Angeles Police Department. By the way, you are fined 1 demerit for violation of the verbal morality statute.
It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
That already happens,example here !!
You are incredibly right about the drivers egos. There are plenty of psychological studies made for various reason, but they all show this, that most drivers think they're better than they really are.
These cars WILL have accidents, but I guarantee, 99% of the time, it will be caused by a driver in a normal car.
I don't drive, I don't even own a car, hell, I don't even have a license. But that's my choice. I'd rather switch from bus to subway and back again going to work, than waste even more time driving and getting myself stressed out.
When those smart cars start having a solid presence though, that will change. It will be like having your own personal driver. No, better than that, a personal driver that doesn't cost you anything and doesn't get tired or makes mistakes.
That means the hours spent in transit can be turned into useful time, when I can read, browse the news, or just work.
Right now, for me, driving is scary in the big cities, or just stresful and tiring everywhere else.
Well, 300,000 miles, one non-fatal accident (with, again, a human at the wheel - but we'll ignore that for now).
Now, I pulled these numbers of a set of google searches. There was a fairly wide range of stats, so I took a bit of an average:
Insurance industry assumes one accident CLAIM per 17.9 years (lots of minor accidents go unclaimed, but we'll ignore them too). Average of 15,000 miles per year per driver. Thus, an average of one accident per 268,500 miles per human driver.
Of course, while the human driver stats are numerous (and this is why insurance is expensive!) the self driving car stats are not. Only one accident with new, unrefined technology in 300,000 miles... and that with a human in control of the car.
That said, your example? That's where a self driving car is much, much better than a human. A human driver with a pregnant woman giving birth, woken up in the middle of the night is going to be tired, highly agitated and distracted and definitely not at his best. The self driving car isn't tired. It doesn't care what time it is. The self driving car will be aware of the speeding racers - and know their exact speed, trajectory, and likely path - sooner than the human driver will, as these are very simple computations to make. The self driving car is indifferent to the passenger; which is also important. It's not distracted, worried, or anxious.
Of course, there certainly are cases where that's just not good enough, extreme emergency cases. That's why all these self driving cars can be driven in manual mode. You've always got that option if need be.
Obviously, routes being set aside for autonomous vehicles will be safer, but routes mixed will be safer than pure-human routes, because autonomous cars are simply safer than human driven cars overall.
I've been rearended while stopped at traffic lights six times in the last twenty years; every time due to an inattentive driver. None of those would happen with an autonomous car.
Finally, yes, mechanical/electronic failure can result in crashes. Just like it can with human drivers - sticking accelerators, for example, failing steering linkages, brakes, etc. Software problems? No different than a human driver having a heart attack, stroke, seizure, getting stung by a bee, etc - those all happen all the time. There's no real difference there.
Meh.
And what evidence is there that proves the safety of those cars, other than taking Google's word for it?
Crashing left and right? Both of those stories talk about the same incident where a human was driving.
"I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)
Well then, people should be driving the speed limit and not an arbitrary speed over because they feel entitled to break the law.
I fail to see the problem.
Eat the rich.
Your theory isn't holding up in the face of the data. Googles Cars have logged hundreds of thousands of miles and have one accident caused by human error.
Slow vehicle driving significantly black the prevailing speed cause accidents for other vehicles, while seldom getting hit themselves. They cause chain reaction fender benders two or three cars back, which they are seldom even aware of, and drive away, never to show up in accident statistics.
At least that's the theory put forth by those who perpetually drive over the speed limit.
Sig Battery depleted. Reverting to safe mode.
You fail to see reality. You talk about what should be, and willfully ignore what is.
I don't think so. I think self-driving cars will catch on very quickly.
Imagine when you get up for work in the morning if you could safely do things like comb your hair or eat breakfast while driving to work. You could save 15 minutes to maybe a half-hour. And a computerized system where the cars drive as close as possible to one another and optimize their placement, and where everyone is allowed to merge when they want to will make rush hour not completely suck.
Very quickly the number of accidents will go down, the number of people getting pulled over for speeding would drop to almost zero, and road rage will evaporate.
Not only that, but because of the cooperation aspect of automating cars(meaning since they all think in almost the same ways), speed limits may actually be increased. Because a computer program can consider multiple variables much more reliably than humans can, their ability to stop or avoid accidents should be drastically better than ours. And because of that, there's a real possibility that automated cars could be legally allowed to drive at significantly faster speeds than human-operated cars are legally allowed to drive.
A320 crashes http://www.airdisaster.com/cgi-bin/view_details.cgi?date=03221998®=RP-C3222&airline=Philippine+Airlines
The aircraft overran runway 4 while landing. A malfunction of the onboard flight computers prevented power from being reduced to idle, which inhibited thrust reverse and spoilers from being used. The offending engine was shut down, and brakes applied, but the aircraft was unable to stop before the end of the runway
I couldn't read the article you referenced ("server not responding"), but the accident report states this was caused by pilot error, not malfunctioning computers. From Wikipedia: "A selection by the pilot of the wrong mode on the onboard flight computers prevented power from being reduced to idle, which inhibited thrust reverse and spoilers from being used. The offending engine was shut down, and brakes applied, but the aircraft was unable to stop before the end of the runway".
There's a surprising number of people who believe that the high level of automation on Airbus is intrinsically more dangerous, but the figures show that the Airbus A320 is the safest narrow-body jet you can fly on. It's true that automated stuff can go wrong, but this can be more than compensated for by the ways it makes flying (or driving) safer.
but I don't think many taxi drivers would.
self-driving firetrucks with fire-detecting robots?
self-driving delivery with a robo-caller to let you know your pizza/laundry/package is waiting?
I'm not sure. One other problem that has been brought up elsewhere is maintenance. If you change the diameter of the wheels, as a simple example, A human driver will have an idea the speedometer is wrong. A computer, not so much.
This simple example could probably be avoided by using gps, but there are places where gps doesn't work, such as tunnels and concrete canyons.
Some redneck hacker will try to modify his own car to disastrous effect.
That's great! That means that the statistics for those cars will clearly show how unsafe they are compared to the automated cars.
Wow, I did not realize drivers were so bad. maybe that is the problem. A drivers license should not be that easy to get...
Unfortunately, there are places where a car is a necessity,
Now, where are you getting those statistics from? Mechanical malfunction is constantly causing airline accidents, and is often one of the causes of the accidents attributed to human error.
In fact, everybody seems to be fast attributing airline accidents to human errors nowadays, even when there are little the crew could do to save the plane. The reality is a bit more complex, since there are several safeguards preventing a problem in a plane, when an accident happens it is because all those safeguards failed. Even the act of attributing the accident to a single thing is already a lie, but the press and several governamental bodies (often in the Judiciary) will make damn sure that somebody creates that lie to them, so they can act upon it.
Rethinking email
Our lifetimes? I'm a Singularitarian, you insensitive clod!
Space game using normal deck of cards: http://BattleCards.org
with license plate recognition, driving the speed limit, this might be a huge windfall for cash-strapped states that have to rely on expensive CHP officers.
"In fact, everybody seems to be fast attributing airline accidents to human errors nowadays, even when there are little the crew could do to save the plane"
What in the hell are you talking about?!? Find me ONE case.
Also, I would like to see those cars drive in harsh winter conditions, on a mix of snow and ice...
Tomorrow is another day...
and the Wright Flyer wasn't exactly up to FAA standards either.
i totally see your point. /dumbass
The majority of airline landings are still done manually, and even the best "auto-land" still requires substantial human input. This is another variant of the "airliners all land themselves these days" myth that seems to be so prevalent, and if it's used as a basis for policy ("Sure we're ready for self-driving cars, just look at airplanes!") a particularly dangerous one.
The correlation between ignorance of statistics and using "correlation is not causation" as an argument is close to 1.
"A flight control failure at V1 caused the crew to abandon the takeoff, with deceleration beginning at V1+8 knots."
And there you have it. The failure that triggered the abandonment was in the fly-by-wire, but the abort was at V1+8. The ENTIRE POINT of V1 is that you CANNOT abort above it without an overrun. A failure at V1 is bad for exactly this reason, but ultimately the overrun itself was the crews fault (though without any description of what kind of control failure there are scenarios in which an overrun can be preferable to trying to get the thing back on the ground).
300k miles. 1 (non at-fault) accident.
Sure, but my own record is better (zero accidents in 4X that many miles), and that car has not driven in anywhere NEAR the variety of situations that I have. In the more carefully controlled environments it has driven in it has done "OK", but it remains to be seen that this will be true when it meets the diversity of situations I encounter in real life.
There are roads near me where the traffic flow is typically 10 or 15 MPH over the posted limit, because the posted limit is ridiculous and everyone knows it is there only for revenue generation, not safety. Well in excess of 95% of the drivers drive at least 10 over, and if you rigorously follow the limit, you'll gum up the traffic flow something terrible, not to mention piss off all the drivers stuck behind you who will then tailgate each other in frustration.
That's fine if you suppose all the cars are auto-driving, but if you mix one with normal humans in such situations, I foresee a lot of anger and road rage.
Will be the police. When these go online, a HUGE revenue stream for them will dry up when they can't really get drunk drivers that much anymore and next to no one will be breaking the rules since they will not be the ones driving and the computer will not break them. I see this as a great thing. Now maybe they can get off their bum ass, do their jobs and actually start chasing crimes more instead of padding their resumes with petty stuff so much. They shouldn't be chasing low hanging fruit or treating it as a for profit venture.
Also, I am wondering how this will effected the trucking companies. I am betting they will have the truck on auto-pilot everyone and have 1 person in the vehicles at all times as a just in case for when it breaks down, needs gas or legal reasons. The trucks will never have to stop really except for gas and the trucker really is just a babysitter who will spend most of his time sleeping or reading or maybe going back to school since he no longer has anything to do. But I am expecting the wages to drop steeply as they also cut the number of drivers and will probably double up on the size of the loads as much as possible with a trucker driving with 2 rigs chained together to cut costs further. Not even going into how they are getting a tax break on the highway wear and tear compared to the damage they do.
Even with the increased cost of gas, I can also see this leading to more families going on long distance vacations as they will no longer have to wait months in advance for plain tickets at decent prices and they can sleep on the way.
With the exception of the truckers (say hello to increased unemployment and even lower wages and even fewer job opportunities, they may have used the bait and switch during the industrial revolution but no need to hide it this time) I see this as nothing be a giant win in everything else.
When police start riding these, we're all in trouble. Imagine a police car which at any given moment knows the exact speeds of the cars around it, and can read license plates of those ahead. Heck, just drive on the highway in an unmarked car, and have it automatically issue speeding tickets to everyone. Neat.
Drive, er ride, these silly things in rush hour commuting to and from a major city like NY, LA, Boston for a year and we'll talk.
Simply getting to where you need to be in a non-ludicrous amount of time requires breaking laws and safety rules minute-to-minute.
Primary examples:
On and off Ramps
Merging
Following Distance
Oh, and BTW, we only hear of the car's success, not any of the failures. There must have been numerous failures, as in any engineering project. Without knowing the failures and weaknesses (for example - what situations do they avoid driving in?) we have no objective framework to evaluate these car's performance in.
Well sir, how else are you going to explain the very dramatic reduction in airline accidents over the last few years? The computer leads the pilot much closer to the runway than before, and reduces the distraction of having to monitor the systems. The pilots might not use it, but the autolanders* are fully functional, and can land the airplane in zero-zero conditions. It's just not legal yet. But the aircraft are certified, at least as far back as 40 years ago with the L-1011. And we're are finding out, due to Air France 447, that some pilots are hardly being trained to fly the plane when the computer goes on the fritz. When I was taught to fly (small plane, but the principle still stands), we would cover up some instruments to simulate "partial panel" conditions very similar to this one. It's difficult, but not impossible. Driverless cars will be at least 100 times safer than what we presently have.
*Reminds me of the old joke:
The squawk-list entry,
"Autolander touches down extremely hard"
The reply by the mechanic,
"This aircraft not equipped with autolander"
“He’s not deformed, he’s just drunk!”
Under what conditions? I've poked around a lot but I can't find evidence thy tossed their car into the conditions I see on my daily commute.
I love tech, but this utter lack skepticism some people have here is really sad. One guy above guarantees any accident between an auto car and human controlled car is "99%" the human's fault. How does someone even make that claim outside some pseudo religious context?
And if you dare raise any question at all, you're called a Luddite, or it's explained how asking prudent questions would have kept men from landing on the moon, or other hyperbolic bullshit he geekverse can't seem to he'd itself of.
Some redneck hacker will try to modify his own car to disastrous effect.
As if morons don't do that already and somehow this is the fault of the car.
-1 overrated isn't the same thing as "I disagree".
It may be surprising, but the icy/rainy slippery roads are a walk in the park for a computer controlled vehicle. Current anti-lock/stability controls do show excellent reliability already. It's reacting to unknown/unexpected conditions which can be tricky: road constructions with messed up road markings, cities with unpredictable pets/small children. Those I'm far more worried about
I think you underestimate the CPU load of driving, or overestimate the computing resources that will be made available.
OTOH, perhaps a car could accurately estimate how fast it's CPU could let it drive, and higher priced cars could have fancier computers. That, of course, would increase the need for inter-car communication, which would put even more load on the low end models, so they'd need to slow down even more.
There's all sorts of possible futures out there. Some of them look quite familiar, from a distance. And those are the ones that people tend to head for. When they get there they are often rudely surprised.
I think we've pushed this "anyone can grow up to be president" thing too far.
IIRC, turn signals became mandatory within about a decade after they were allowed. This isn't quite the same, so I give it 20-30 years to become mandatory on local roads, as well as on freeways. Whether that's within your lifetime or not I wouldn't choose to guess.
I think we've pushed this "anyone can grow up to be president" thing too far.
They've taught drones to land themselves on aircraft carriers, though, so maybe it will be tomorrow that airliners fly themselves? Of course these systems still cost millions, so we'll probably have to wait a while to buy a self-driving car.
Wow, I lol'd at this... How could you possibly compare these two? Airspace has an incredibly consistent, standardized and mostly centralized air-traffic control system. you have ~7,000 aircraft simultaneously in the entire US Airspace. We have over 242 Million registered vehicles in the United States. I couldn't find data on how many are in operation simultaneously, but I think it is safe to say you can find over 7000 in operation simultaneously during rush hour in any average city on the interstates there alone.
Add to that the room/flexibility to maneuver in a vehicle on a road system, parking lot, parking garage, shoulder, dirt road, etc. compared to "air space".
Both have weather hazards, granted. Except that often when weather is rough, planes don't fly there (route around it). Motor vehicles don't work that way.
I'm sure I have BARELY scratched the surface here. Maybe we should instead be debating what I meant by "quite a long time". I'd say a significant number of autonomous vehicles in operation in the U.S. is at least a decade away, maybe two.
Come play Moral Decay!
After all, if corporations can be people, why shouldn't cars be people, too?
And speeding tickets? Ha! When cars drive themselves they will be unable to do wrong, again, just like corporations. And if one ever should, well, the owner will get the ticket and have to pay, won't he? It will be just like today when your car gets stolen and the thief gets caught speeding or blowing a light by a traffic-cam, or parks its remains illegally after he strips and abandons it, when you, the owner, get the traffic and parking tickets and the bill for towing and storage to the impound.
Your theory isn't holding up in the face of the data. Googles Cars have logged hundreds of thousands of miles and have one accident caused by human error.
Hundreds of thousands of miles? That's nothing, Americans drive billions of miles each year, and I don't even need a source since that's pretty obvious, with 310+ million Americans each would only need to drive 3 miles a year to reach a billion.
my karma will be here long after I'm gone
Your theory isn't holding up in the face of the data. Googles Cars have logged hundreds of thousands of miles and have one accident caused by human error.
Slow vehicle driving significantly black the prevailing speed cause accidents for other vehicles, while seldom getting hit themselves. They cause chain reaction fender benders two or three cars back, which they are seldom even aware of, and drive away, never to show up in accident statistics.
At least that's the theory put forth by those who perpetually drive over the speed limit.
Awww that's sucks! Then the drivers who are speeding are blamed! And the people driving the speed limit aren't even given a ticket. I'll remember this next time im driving, that speeding saves lives
my karma will be here long after I'm gone
To a computer, traffic density and room to maneuver shouldn't matter. I'm just pointing out that flying has become much much safer with automation. The machine is extremely safe and reliable, and with very few exceptions*, almost to the point of being statistically insignificant, pilot error remains virtually the only outstanding issue now And the same will apply to automobiles.
* which to me are not being handled properly, but such is the nature of human economics.
“He’s not deformed, he’s just drunk!”
Even if they were slightly less safe, society could probably accept the MASSIVE advantages self-driving cars would have over manual.
Why OpalCalc is the best Windows calc
That, pardon the pun, is a two way street. If speed limits were set by engineers using scientific principles like they are supposed to be and like we pretend they are I would agree. However, they are too often tampered with by revenue seeking politicians and law enforcement.
If the accident was caused by a programming error, do you think Google would have admitted it and risked killing all th new legislation that's being passed, and thereby also the project?
While self-driving vehicles may be acceptable for some highways, I will not even consider them as acceptable for my driving until AI has grown by leaps and bounds. Cars may be able to get the data and respond to it more quickly than humans... but without a good AI, it cannot interpret it well. For example: After a winter ice/snow-storm, if road x has a stoplight at the top of a hill, then I know I cannot approach the hill in the same manner as I could during other conditions because my tires are much more likely to break loose if I come to a complete stop and have to start accelerating when the light turns green. I have seen inexperienced drivers do this and slide backwards. Here is another example: I do a lot of driving on country roads near Amish farms. If I am on a road with a lot of small hills, and I see fresh horse dropping (aka road apples) in the middle of my lane, then I know that a horse-drawn buggy has driven down here recently and I need to take extra care with the hills blocking my view. These are just a couple examples.... I have learned many other things over the years and have become a better driver with the experience. Can you tell me that somebody in Silicon Valley that has never even seen an Amish farm is going to think of programming the AI to interpret horse droppings as a sign of an horse buggy up the road? Do you think that someone that has little experience driving in ice will be able to create an AI the can actually plan how to do things so it will minimize problems that will occur later/further down the road? Self driving cars for anything other than highways will be nothing but a pipe dream for a long time.
I see reality, and part of that is that a large segment of the population has collectively decided to not give a shit about traffic laws. This needs to change though better education, stricter licensing requirements, less cameras everywhere and more actual police officers being highly visible and helpful in public (another can of worms right there).
The best thing I can do is to set a good example, so that's what I do.
Eat the rich.
I think you overestimate the average human. For example, please tell me exactly how far it would take your vehicle to stop from 60mph in current weather conditions with the current brake wear? Don't know? A computer would, and it could adjust speed accordingly, given condition of the brakes, weather, road conditions, even traffic information, because if all vehicles are reporting their gps location then your vehicle knows if there is another vehicle close by or not.
my karma will be here long after I'm gone
They are to be forced to give a shit? What is this, a shit dairy?
Hell no, I'm very much against self-driving cars. The very reason people seem to clamor for them is because the average driver doesn't give a shit about driving and thus constantly breaks the speed limit while driving distracted. This is a societal problem that needs to be solved by a change of mindset, not the introduction of even more technology to take responsibility away from us like children.
I used to call myself a geek proudly. No more. Now I just like tech because it's fascinating, not because I think it'll solve all the world's problems.
Eat the rich.
Not to mention, the number of people who cannot drive, but would now be able to have a car.
Which is a chicken and egg problem solved by self driving cars.
Not safe right now... the difference being is that we can make continually make self driving cars more safe, since driving only requires a set of rules and environmental awareness. Humans will never become more safe, in general, because they are inherently mistake prone due to fatigue, poor judgement, distractions, intoxication, and many other factors.
With that I'll agree. But standards have to be in place to ensure that whatever machine drivers you put on the roads, they are at least no worse than human beings.
Just look at the wonders of automated flight. Most airline accidents that aren't due to terrorism or mechanical malfunction are due to pilots overriding the autopilots.
Do you have a cite for that? It's not that I doubt your word so much as I think you're making shit up. Because my understanding is that most accidents happen during takeoff and landing where autopilots aren't routinely used and that weather is a big factor in many accidents: http://www.faa.gov/data_research/accident_incident/
Self-driving cars will eventually be the majority.
I don't see why they would be. I think manually operated cars will continue to be much cheaper.
We can keep increasing the limit until they reach the safety levels we enjoy today! But at least your car will automatically slow down when weather conditions are bad and might even refuse to take you out on the road if they're bad enough and there's no emergency.
Even if they were slightly less safe, society could probably accept the MASSIVE advantages self-driving cars would have over manual.
I doubt it. People seem to want to be in control, even when their control is inexpert. I hope I'm wrong, though.
Sure I'm paranoid, but am I paranoid enough?
They don't have to be safe, as nothing, not even laying in bed, is completely safe. They just have to be safer than what exists now. That is a pretty low bar to reach.
That makes me wonder - how well do they handle low bars?
Sure I'm paranoid, but am I paranoid enough?
> with 310+ million Americans each would only need to drive 3 miles a year to reach a billion.
With actual numbers:
"There were 190,625,023 licensed drivers in the United States in 2000." [1]
190625023 * 3 = 571875069
So not only are you wrong, but I don't see your point either. Americans drive a lot, but they also have a lot of car accidents. (Feel feel to provide more recent numbers, but you won't get a billion even if you count the whole population.)
To be fair, I think that computer controlled car should be granted the right to drive, if it can pass the driving test, which human drivers need to pass. Should there be an accident, the company that provided the car should pay. That is unfair for the company, but it is to earn the trust of the population and to ensure that cars have as little defects as possible.
1) http://www.fhwa.dot.gov/ohim/onh00/onh2p4.htm
It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
Even today's higher end cars have cruise control that is set to keep a minimum distance X to the car in front of it. When that car slows, so does the cruise control, right down to a full/emergency stop if necessary.
What is it with Slasdotians pointing out the most obvious of possible shortcomings and then shouting out "Hah, you didn't think of that, did you?" You think the folks developing these are that stupid? These cars have sensors and reaction times that literally dwarf anything you or I could do behind the wheel.
People replying to my sig annoy me. That's why I change it all the time.
multi-hundred horsepower engines: less safe
You know what I find one of the scariest things on the road? Underpowered cars that can't get up to speed fast enough and hence have to enter the highway going 20 less than what everybody else is doing. When it's relatively quiet on the road you can change lanes so he has more room, but when it's busy that's not always an option. Then you see one car hitting the brakes (seriously folks, using the brakes on a highway = fail) and a whole cascade happens of people seeing brakelights, not knowing what is going on and hitting the brakes themselves. All because 1 little jackass in a tiny tin box couldn't be bothered to shell out for an extra 20 horses and the tires to get them on the road properly.
Now you really don't need more than about 90 bhp in a small car, but the big saloons and SUVs, especially the ridiculously large monsters that pass for cars in the US, need a lot of power to accelerate.
People replying to my sig annoy me. That's why I change it all the time.
Lenina Huxley: I thought your life force had been prematurely terminated!
John Spartan: Yeah, I thought I was history too. What the hell happened? All of a sudden, this car turned into a cannoli.
[End Of Line]
I'm not talking about the robot cars crashing. I'm talking about the people that come up behind them suddenly and have to react to a car going 20mph slower than them in heavy traffic.
Try driving the speed limit on the 101 or the 5 in LA when traffic is moving freely, and observe the effect on other cars.
Google's self driving car has never crashed when it was being computer driven. It only crashed in that one single incident because it was being driven by a person.
The story you linked describes perfectly why autonomous cars need to replace human drivers. They are much safer.
Within 10 years.
And it will take very little time after that for self driven cars to be made illegal in most metropolitan areas.
Your kids won't need to get a drivers license.
It won't be possible for your grand kids to get a drivers license.
Hundreds of thousands of miles isn't shit. In Detroit it's typical for a new model to have 100 cars built and the fleet log nearly a million miles before production starts. And that's not doing any fancy self-driving. When you build 100,000 of them and keep them in service for 100,000 miles each then we can talk about "data".
The only reason they need safety standards is because they're going to exempt the manufacturers from liability. For when an accident occurs with 2 of these cars (or a single car accident with one) there will be no question who is to blame.
Actually no, insurance companies make money as a percentage of their sales. More accidents, higher costs, higher premiums, higher profits.
Less accidents, less costs, lower premiums, lower profits.
The insurance industry will not be very happy overall. Although some companies may take targeted approaches to make money as the market changes over. Hoping to take a larger percentage of the overall reduced pot once things sort themselves out.
Slashdot. Ask a fair question of tech, get modded down. Yeah, it's not form of religion.
"unions aren't designed to protect peoples jobs from automation"
glad you awoke from what I assume was the 50s...
just to cite one example I remember being in Hawaii 10 yrs ago when pacific longshoremen striked to protect clerical jobs from wireless computers (& this was yrs before iPad)
A necessity only if you are a lazy slob. Oh wait, we're talking about Americans, aren't we? A car is necessary for you people to go down to the corner shop.
...the number of people getting pulled over for speeding would drop to almost zero, and road rage will evaporate.
But rage at the government for increasing taxes will go up several fold. Gigantic swaths of the USA would have to figure out how to deal with the loss in revenue if drivers lost the ability to volunteer to pay these unofficial taxes.
They'll scoot about in gleaming alloy aircars, two lanes wide, no doubt.
Until then, the robot car is going to impede the monkey cars, and it's going to cause monkey car crashes.
Maybe the thing to do is to let the robot car speed along with traffic.
Or stick a streetcam on it, declare it a pace car, and ticket anybody that flouted it, by mail, with panoramic video of the offense. You could probably manage the LA area with a few thousand of those, a couple miles apart.
Streaming footage out of those might be pretty entertaining to watch, actually...
This has a certain apples to oranges component to it in the first place (self-driving cars vs truck drivers). But if they end up at cross-purposes, then trucking unions etc. would get a political response from spileptics, and we can scream. If I had one of these cars I'd be so happy I'd have to start using Google for searches again.
Epilepsy is a varied disability, but on average driving with epilepsy increases the risk of an accident approximately 11% per mile over a comparable driver lacking it. There are other chronic diseases found to generate more dangerous drivers with higher rates, but of course a catastrophe from epilepsy is easily imaginable even without a damn car.
CA is one of a bunch of states that revokes your license if you have a seizure. I've lived in California as a software engineer more than ten years,
living alternately between having fistfuls of cash and brushing the edge of homelessness, mostly from inadequate job markets closer than driving distance, and from the risk of leaving the house for very long at all. After a seizure at home, I just have a hangover. If I spend any time outside, every so often I'll wake up inside an ambulance taking me to a local ER so they can strap me face up with post-ictal nausea and migraine to a board for six hours while some lab somewhere tests blood samples for a spectrum of illegal drugs (medical bracelets, ER history, etc. etc. etc. notwithstanding). These stupid trips cost more than a grand each. Being outside has a certain financial risk per hour, ultimately, thanks to the density of heroes around here with cellphones and driver's licenses.
Although in fact I have to confess (in another state, in my twenties) I actually did have seizures while driving. Usually I saw auras before then, and not being a total idiot, I would arrange a ride or call a taxi. But two or three seizures had auras that must have been too brief to pull over.
In no way advisable (I mean duh, so shut up), it's actually possible to still drive a car around during an epileptic fugue where you have lost consciousness, especially on familiar trips. I've seen it happen to me on a bike too. Frankly I'm surprised to still be alive.
During recovery from a fugue one is slowly regaining consciousness. At first simple automatisms can take over your actions- you steer a wheel to stay between lines, you see a red light and you get in line, you might make right turns, maybe four or five in a row. You'll feel like going faster and push the right pedal; etc. Even if the route is familiar, you go straight and miss lots of turns. I had a seizure at work once and even walked up to a guy's cubicle and started typing gibberish into the keyboard- that's how sophisticated they get. Once a familiar commute enters unfamiliar neighborhoods, the avalanche of unexpected visual stimuli, and the sense of being lost, will present itself as a problem. That may trigger the emergence of consciousness since these are first moments that form any long term memories- of encountering some legitimate frustration beyond the ability of automatisms to control. Then the suspicion dawns of possibly just having had a seizure. They hide themselves from recent memory fairly well- I can end up sending emails to people asking if there was just a seizure after a sudden inability to type in a single line of code. (Skills return on varied schedules.) In a car, once I regained full consciousness, I always realized I had gotten lost. (this was pre-GPS). But surprisingly it also dawned on me that I hadn't crashed into anything, and frankly, still can't figure out why. But of course the State of California isn't going to care about stuff like that.
So once I find out that I can afford a self-driving car, you can bet your ass I'm getting one with tinted windows.
Your theory ignores teh wake of carnage each Google car leaves behind it. Sure, ,Google cars have no AI-caused accidents...
Redneck hacker? Interesting mental image...
With reports of Google's self-driving car crashing left and right how could anyone want to be in one of these vehicles? They just aren't safe. When something happens when you're driving then it's at least your fault and you could do something about it, but not in self-driving cars.
In both of the cases you cited, a human was driving. The only accident the google cars have been in are ones where a human driven car smashed into the google car while it was performing legally and properly.
Of course, both stories tried to cast question on whether the car was human piloted or auto piloted, because who would want to read a story about a car driven by an idiot crashing into a driverless car?
Your theory isn't holding up in the face of the data. Googles Cars have logged hundreds of thousands of miles and have one accident caused by human error.
Slow vehicle driving significantly black the prevailing speed cause accidents for other vehicles, while seldom getting hit themselves. They cause chain reaction fender benders two or three cars back, which they are seldom even aware of, and drive away, never to show up in accident statistics.
At least that's the theory put forth by those who perpetually drive over the speed limit.
That would seem to indicate that other drivers aren't following at a safe speed and/or distance. Which is a good reason to get people out from behind the wheel.
I would assume that none of these laws allow vehicles to actually drive themselves without human supervision, so I would say that the human behind the wheel has the final responsibility to override the software in case it tries to do something crazy.
Air France 441 - While the pilots could technicaly make the plane not fall, they had absolutely no information available telling them WTF was happening, so they choosed wrong. Mainly attributed to crew mistake. Keep in mind that the mistake wouldn't have happened if the windspeed sensor actualy worked, if the crew was not pressed into following a known dangerous path, if the designers of the automatic pilot tought about the ergonomics of the emergency system used, and a ton of other factors.
TAM 3054 - There was little chance of either the pilot or the co-pilot discovering what was wrong on the plane even if you gave them several minutes to test things, but they only had a few seconds. Mainly attributed to crew mistake, their mistake was not noticing that a control was stuck a few millimeters before it should be. Again, that mistake wouldn't have happened if there weren't mechanical and ergonomical flaws on the plane, if the airline did follow other maintance procedures, etc. If they were landing in an airport with a bigger runway they could put the plane on air again, but not there.
Well, out of my mind, that's all, for any other example I'd have to research. Hope that's enough. Keep in mind that in both cases there the crew did make a mistake, but attributing the accident on it is a bit too much.
Rethinking email
Meaning, the car wasn't woken up in the middle of the night to get a pregnant woman to the hospital quickly over dirt roads, past nighttime street-racers, etc...
A human in the exact same conditions will be dozens of times more likely to get into an accident as well. Maybe that's why, despite specialized training and flashing lights and sirens, ambulances have a much higher accident rate per mile than cars.
While, as other users have pointed out, the articles the parent links to may be anecdotal, do you really think he was trolling? Modding down should not be a proxy for disagreement.
Plus, they aren't throwing in a desktop CPU. They would more than likely have multiple processors handling different aspects of the car. Additionally, the processors would be specially built for the task. You don't need as much power with specially made processors. After all, just look at how much computing power it took to get to the moon or how much faster encryption is with the specialized AES instructions.
You're naive if you think the insurance companies will lower rates in line with their costs. Sure, they'll lower them SOME, just to appease everyone, but costs will be so much lower that they'll be more profitable in the end run.
So they can waste more energy? Why should self-driving cars get a higher speed limit?
That depends. In Nevada a driver is required to be behind the wheel at all times. The same is expected in California. While it would be nice if self-driving cars allowed disabled people (such as myself) to drive, it seems more likely that--at least for a long time--a licensed and non-disabled driver will be required just in case an override is needed.
Remote controlled car bombs are perfectly safe for the operator.
I would be surprised if it took a decade for that requirement to be removed, and if it wasn't removed within two decades I'd be astounded.
Computers' ability to multitask grows exponentially as time goes on. At this point in time, autonomous cars don't have a track record. But in ten years they will. I think the requirement for a human in the driver's seat to be able to override is based on the fear that the computer will fail and the car will be out of control, but given enough time it will be obvious that the computers' ability to maintain safety will be greater than humans' ability to do so, and then I expect that requirement to be removed.
But if all the other cars are automatic, no one will be going over the speed limit.
You give some great examples where human experience can't be matched by a computer. I could add such things as eye contact with the other driver(s) at a four-way stop sign; noticing the tag number on a car and predicting where it may exit off the interstate; recognizing a car you've seen day after day on your commute and knowing where it's going; and avoiding potholes on a rainy day (unless you've seen it on a dry day, there's no way of knowing how deep it is when filled with rainwater).
I've driven more than two million miles and there's no way a computer can know what I know about driving. There's no substitute for experience.
I'm a little uneasy at what will happen when most people are driving these and how they will interact with the police. I suppose there wouldn't be any point in trying to ticket one, but I still would expect they'll eventually do something like on a signal from a police car, the autonomous car will pull over and stop itself. Things just get more ominous, if more unlikely from there. What if it was set up that, if the Government wants you, any autonomous car you get into will automatically drive you to the nearest police station nonstop?
I don't reply to ACs
Lenny: Hey look! Homer has one of those new robot cars.
Carl: Yeah, one of those AMERICAN robot cars.
same as if you built a microwave that killed children.
Between the two, I've seen far more 300+hp RWD cars spin and crash than I have seen underpowered cars actually cause collisions from merge difficulties.
Mostly the slow to accelerate actual crashes I have seen have resulted from drivers not applying the power they had, pulling out into 50+mph traffic and doing 20, because they're not going very far or whatever. I actually spoke with a guy that got rear-ended this way, his attitude was that just because everybody on the road was doing 50 in a 45 was no reason they had to, the traffic was supposed to slow down for them. Legally correct, sir. Too bad your pregnant wife miscarried as a result of your knowledge of the law + ignorance of how things really work.
Yeah but if everyone else is also in automated cars there will be no one to fine. We could drastically lower the police force numbers.
Here is a scenario where if a self-driving car can pass 100% of the time, then I would deem it safe to get into.
Driving on a mountain road around a sharp corner where there is a steep cliff on the right side. Auto-car is passed on the left by some *sshole "manual" driver, but then the *sshat driver cuts in short because of oncoming traffic at the last second. Robo-driver identifies there is suddenly a car intruding into its safe-T-zone (TM) and does what its programming tells it to do, avoid hitting other vehicles. So the self-driving wonder swerves right to avoid the other car and zooms off the cliff.
A human driver would recognize that hitting the other car in this instance is the safer solution then to go careening off the steep cliff.
I agree that a self-driving car can work, and 99% of the time will perform adequately to protect its occupants from disaster. But since we have not mastered true AI yet, all self-driven cars will be built with flaws in their logic that will fail catastrophically. "Avoid hitting all cars", for instance, is not a good enough directive to ensure the safety of the occupants in 100% of all situations.
Someone mentioned that the deaths caused by self-driven cars would be far less then manual drivers, but then I would disagree that any technology introduced on the highways would be adequate to allow any fatality, especially in scenarios where a human driver may have been able to avoid death.
Basically what I am waiting for is the inevitable 100 car pile up with massive fatalities that WILL occur at some point in time where investigation will identify that a self-driven car, or cars, was the cause of it. Any company involved in programming or manufacturing that self-driven car will be sued out of existence and the "love affair" everyone seems to have about auto-driving cars will end quickly.
I am amazed at how delusional governments are into so quickly allowing this technology on the roads, sounds to me like there is some massive lobbying going on to short-cut the necessary amount of time to test auto-driven cars under all senarios, not just ones in controlled and predictable setups like we have seen. 5 years ago robo-cars could not drive around a dirt track, now they are quickly being allowed on our highways. That just is irresponsible.
I haven't thought of anything clever to put here, but then again most of you haven't either.
Even so, softwares and hardwares can be fixed.
You cannot fix people.
Because a computer can navigate safely at those speeds. A human cannot.
And will there not be someone who can operate the vehicle waiting to take control?
Just as soon as they make a CPU faster than my brain and AI smarter than my reasoning, they can rubber stamp that all they want. Until then, get robocar out of my way.
You do, however, have attention, reaction time, and perception limitations that the machine doesn't. And while the machine may have redundant systems, if your brains shuts off you don't have a backup.
> It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
Is this really true, or is it just an Urban Myth invented to justify breaking the law and endangering other road users lives ?
Citation ?
I already use one of these self driving cars. It's more commonly called "a bus."
-- Experience is a wonderful thing. It enables you to recognize a mistake when you make it again.
Motorcycles?
-- Braden's law of data: All data spends some of its lifetime in an excel spreadsheet.
New Scientist ran an article earlier this year on a study which indicated that if something like 1 in 10 vehicles were autonomous and were linked to information about traffic lights etc, it would order traffic and improve everyone's time to work.
The fact is that our cars have been getting more automated over time. I have cruise control, intelligent braking and anti-skid technology in my car, and it ain't anything special.
I can't wait for more automation. Especially something that can spot kanga-fucking-roos. I would like something to predict their movements too, but I suspect they are quantum events that are only resolved once they are under the wheels of your car.
Oh, and 1.2 million people die every year in road accidents due to human error. That isn't a particular hard target to improve on. People are stupid. And stupid people often drive very fast. Just this morning I was held up by an idiot who put his high performance car under a ute (sure it could have been the utes fault, but my money is on the rev-head dickhead). I have to say that his day started (and probably ended) worse than mine - it took them over an hour to get him out.
Long before that, as soon as the stats show that accidents mostly happen when drivers override the computer, you will find that insurance companies will not cover overrides, and hey presto, no more free driving except for the very rich and uninsured.
Not when you factor in insurance.
It's not safe for the simple reason that the automatic cars will drive the speed limit, and cause accidents because everybody else is going 20 over.
If you stick to the speed limit, it is not you that is driving dangerously and causing accidents. But no doubt you are one of those superior, race-quality drivers who has a right to travel at whatever spped you think fit, and expect anyone going slower than you to move out of your way.
To have a right to do a thing is not at all the same as to be right in doing it
About 95% of the people I saw on the Freeway this morning (and I traveled 300km, so not a bad sample) were driving 5-15k over. Most seemed to be on cruise from what I could tell. You could spot the police speed traps a mile off because people slowed down.
In a democracy when the majority flout laws you should change them, not re-educate them. Or are you in favour of Big Brother?
Fuck a citation. This sounds like something for Myth Busters! High speeds and carnage. Right up their alley.
" I can't for the life of me think of a problem solved by..."
Oh, well, that finishes the argument then. Thanks for clearing it up for us.
If you have to brake suddenly to avoid hitting someone in front was obeying a speed limit, you were driving too fast and/or without due care and attention. Public roads are not racetracks, and so however good a driver you are you have to share the road with slow drivers, old drivers, cyclists, trucks, motorbikes, pedestrians, horsesor whatever comes your way.
To have a right to do a thing is not at all the same as to be right in doing it
Road deaths (at least in Oz, and we do a fuck of a lot of driving) have declined by 26% in the last 10 years, and I am guessing that total distance traveled has increased quite substantially.
http://www.minister.infrastructure.gov.au/ck/releases/2012/january/ck001_2012.aspx
One in 10 of those deaths is a pedestrian. They have also benefited with the removal of hood ornaments and softer shelled cars, but are probably worse off by the % of 4WD vehicles and digital media/phones...
With black out windows you could even have a shower and crafty J Arthur Rank on the way to work.
To have a right to do a thing is not at all the same as to be right in doing it
+1
Big dog could walk on ice four years ago: http://www.youtube.com/watch?v=W1czBcnX1Ww
People on the other hand don't seem to have a good grip (pun intended): http://www.youtube.com/watch?v=Fbb3631ew_4
Even when they have four wheels to help them: http://www.youtube.com/watch?v=8hemAjudq4g
If all cars are self-driving, then we can happily increase the speed limit -- and probably by a lot!! We might even get a scenario where one speed limit applies to humans, and another (higher) one applies to computer-controlled vehicles.
Unless you have different roads for human- and computer-driven cars, it would be a recipe for disaster to have vehicles travelling at wildly different rates next to each other.
To have a right to do a thing is not at all the same as to be right in doing it
Because a computer can navigate safely at those speeds. A human cannot.
Yes, we know, you seem to have completely ignored the point about the waste of energy if suddenly cars are going twice as fast.
To have a right to do a thing is not at all the same as to be right in doing it
I used to call myself a geek proudly. No more. Now I just like tech because it's fascinating, not because I think it'll solve all the world's problems.
In other words, you have grown up and become selfish and cynical, like most people do.
To have a right to do a thing is not at all the same as to be right in doing it
But a computer doesn't have to be better than you the way you do it. It just has to be better than you.
Night vision, radar, knowledge of traffic light sequences, knowledge of prevailing traffic conditions, always knowing the speed limit, being able to check that it is in the optimum stopping distance from the car in front AND behind, sensing the road temperature and moisture. Hell it may even have access to the driving patterns of all other drivers on the road, and be in communication with them. And it can update its knowledge thousands of times a second.
You really don't know what you are up against.
Incidentally, the local authority could be given feedback on road conditions and get to that pothole long before it became a serious problem. Whether the local authority would act on it is another question entirely...
You fail to see reality. You talk about what should be, and willfully ignore what is.
The fact that a lot of people do something doesn't automatically make it morally right. As a society, we have decided to have speed limits on roads, therefore people should stick to them and not whine when they get caught breaking an agreed law.
To have a right to do a thing is not at all the same as to be right in doing it
I'm not talking about the robot cars crashing. I'm talking about the people that come up behind them suddenly and have to react to a car going 20mph slower than them in heavy traffic.
Try driving the speed limit on the 101 or the 5 in LA when traffic is moving freely, and observe the effect on other cars.
Then the fucking traffic cops should start pulling a few random speeders over, give them a severe public beating and a ten thousand dollar fine and confiscate their car. Pour encourager les autres.
To have a right to do a thing is not at all the same as to be right in doing it
It's called "not living in a naive dreamworld".
Eat the rich.
WOOOOSSHHH!
That was the sound of a computer-controlled vehicle driving past you. :)
Agreed, but I think "rev-blooded dickhead" would have been funnier.
In other news, I really can't wait for the day that I can sit there and read slashdot instead of having to drive when going to work/on holiday.
New car designs where everybody sits around a central table and plays boardgames or something.
And it won't cost 10 bajillion pounds like the train (Going 136 miles on Saturday, that's £117 for 2 adult returns coming back a week later)
Roll on self driving pods.
Not when you factor in insurance.
I don't know how much YOU pay for insurance, but at my rates, I don't think so.
Google cars have had over 480k miles accidents-due-to-system free. There was one accident when one was rear ended while stopped at a redlight, and another when a human tester decided to override the automatic driving and drive it himself.
who prays for Satan? Who in 18 centuries has had the humanity to pray for the 1 sinner that needed it most? ~Mark Twain
According to Google. There were no independent tests.
I don't understand UK train prices. I grew up there, and still don't understand them. But it is probably down to Maggies privatisation of the railways.
Here in Sydney, last time I traveled up to the far side of the Blue Mountains, about 120kms away, it cost me about 10 pounds (AUD$15) one way. I can only assume they are as expensive as they are because the alternative, driving, is also very expensive due to high tax on fuel (we are about 2/3 of your price, and that is with the AUD at unnatural highs).
And, I can see it now, every spoiled 8 year old would have their own car. It will be like today with smart phones.
And what evidence is there that proves the safety of those cars, other than taking Google's word for it?
You haven't yet proven that you are unbiased. Please prove that you do not work for Microsoft. Then, please prove that you are an American. We cannot have foreign nations influencing our laws. Then prove that you didn't steal someone else's Slashdot credentials to make that post. Please also prove that you regularly make useful contributions to the world (as Google has) so that we will have some reason to consider your suggestions. Then, perhaps we will consider what you have to say. ...or maybe we should all just try move forward in good faith without requiring proof at every step along the way.
Instead of building this: http://www.et3.com/
Humans would rather build automated cars....
It's the same logic as having an individual elevator for every guest at a hotel.
Such a waste of resources, when we could be building objects of such greatness such as ET3.
Automatic car lane.
Got anymore easily solvable problems to throw at us, Luddite?
True, but they won't be doing it in the left lane with the rest of the morons. I don't care if somebody is going 60 on the freeway, as long as they are doing it in the correct spot.
I think the reality is that many, many driver suck. Drinking, texting, applying makeup and eating are just part of an ever growing list of distractions. Let the people who don't care about driving hand it off to a robot. Bus drivers aren't perfect, and they crash on occasion, but we still let them drive.
I wonder how autonomous cars will handle motorcycle lane sharing?
Slow vehicle driving significantly black the prevailing speed cause accidents for other vehicles, while seldom getting hit themselves. They cause chain reaction fender benders two or three cars back, which they are seldom even aware of, and drive away, never to show up in accident statistics.
The California Highway Patrol agrees with you. Here in the Bay Area you don't see police on the interstates during rush hour. They know that everyone is speeding and everyone is tailgating. They know that when they get on the freeway people in their immediate vicinity hit their brakes often causing accidents. The only time I see the highway patrol is when there is already an accident. Of course this means that the interstates have the most appalling bad behavior from some drivers. It's a law free zone. I'm not saying I think this is a good thing, it is just the way it is.
I see biggest problems being with the interaction of human and AI cars. If the AI cars drive the limit and stay on the right, it might work. If an AI car will move to the right and yield right of way to a faster car or someone who flashes their lights (as is required in CA) then great. It might even be fun.
Here is a benefit I haven't hear yet. What if we could re-task our police away from traffic duty and get them working solving crime! In a place like Oakland where there are over 100 murders per year and most go unsolved, that would be a huge benefit. In a place like Marin where violent street crime is less common, you could actually reduce the number police!! I've always thought that the presence of heavily armed public employees who think they are above the law was detrimental to our society. Wouldn't it be nice to give half of the the pink slip?
-- QED
And you are an obtuse asshole that went right for the ad hominem rather than troubling to understanding the post.
Not believing what a company claims about its own product without proof is not bad faith but rational behavior. Otherwise I might as well order a box of these pills that claim to enhance my penis length by two inches. I don't have to prove anything, because I didn't claim anything. My post was a (admittedly rhetoric) question.
Thing is, the guy going 60 forces the guy going 62 into the passing lane, and that blocks the people coming up at 85. You've got the initial event, and then the building road rage of those trapped behind, which carries on down the road even after the 62mph guy gets done passing.
I saw it all the time with trucks in california... speed for them was 10mph below everyone else.
Thing is, the guy going 60 forces the guy going 62 into the passing lane, and that blocks the people coming up at 85. You've got the initial event, and then the building road rage of those trapped behind, which carries on down the road even after the 62mph guy gets done passing.
I saw it all the time with trucks in california... speed for them was 10mph below everyone else.
I understand that completely. However, it's far better than the drooling moron going 50 in the left lane. I'm in the Bay Area where the freeways are 5 to 10 lanes wide (usually 3 to 5) and people drive slowly in all the lanes. Getting more of the slow folks into one or two lanes will leave three or four lanes for everybody else.
Human drivers or not, there will always be cars going below the speed of traffic.
Your comment "I saw it all the time with trucks in california... speed for them was 10mph below everyone else" makes me think two things.
1) California is a big state, so what you see in one area can be entirely different than another area. I was on a motorcycle trip a few weeks ago, and spent the entire morning on narrow roads with no center stripe. By the afternoon, I was almost home on a 6 lane highway.
2) You don't live in California, so you probably don't see LA or SF traffic on a regular basis. I do. The roads are full, and a large percentage of the drivers aren't really aware of their surroundings. I don't think robot cars are the ultimate answer, but they will probably be better than a lot of the drivers out there.
Actually, I don't overestimate the average human. I consider all human drivers unsafe. I consider myself unsafe enough that I yanked my own license. But I expect that we will hold the machines to a higher standard than we hold ourselves. That's how it usually works.
I think we've pushed this "anyone can grow up to be president" thing too far.
It's not unfair to the company. If there is a manufacturing defect that is the cause of an accident, then the manufacturer should be liable. They will have insurance for such an occurance. If it was operator error (such as the human intervenes and causes the accident), then they will be accountable and will have insurance for that circumstance.
All my liberal friends think I'm a conservative, all my conservative friends think I'm a liberal.
The summary (and the article) identify California as the second state to approve self-driving cars. They don't tell us that Nevada was the first.
Example source: http://www.examiner.com/article/google-s-autonomous-cars-clear-california-hurdle
Slashdot link: http://tech.slashdot.org/story/12/02/17/1320206/nevada-approves-rules-for-self-driving-cars
True: there is no substitute for experience. But, how many of the other drivers on the road have your level of experience? How about if we take your experience and put it into *all* the "autocars" out there (contrary to your post, it is possible to do all the things that you mention), and therefore have a bunch of "you"s out there sharing the road? And not just in your geographical area of expertise: do that in each and every area in the country (because there are a bunch of "you"s out there that could be tapped as a resource) to make all areas just as well-known, to *all* the "autocars", just like navigation systems. Doesn't that make sense?
Personally, I would prefer a bunch of "you"s rather than a bunch of distracted, angry drivers to share the road with, and with computer-driven cars, it's at least a possibility.
Last accident: 1997??
Apparently it's partially because our railways weren't totally destroyed in the war so they are only now replacing the Victorian lines.
Which we're paying for, through the nose.
APPARENTLY we'll see returns on this investment. But do we really think the private railway companies are going to reduce their fares by 90 odd percent when all the track is replaced?
Maybe they'll nationalize it again! 8D
Then we wouldn't have umpty billion different rail providers for a country the size of a postage stamp.
LOL, good job at posting two different reports of the same incident and implying that they are multiple incidents.