Slashdot Mirror
Google's Driverless Car and the Logic of Safety
mikejuk writes "Google's driverless car could save more than 1 million deaths per year and tens of millions of injuries. It is an impressive achievement, but will we allow it to take over the wheel? Sebastian Thrun puts the case for it in a persuasive TED Talk video. However it may be OK for human drivers to kill millions of people each year but one human fatality might be enough to finish the driverless car project — in fact it might not even take a death as an injury might cause the same backlash. Robot drivers might kill far fewer people than a human driver but it remains to be seen if we can be logical enough to accept the occasional failure of algorithm or hardware. Put simply we might have all seen too many 'evil robot' movies."
Put simply we might have all seen too many 'evil robot' movies.
I do not know what these movies you speak about are, but we have all read Sally.
We should be killing people, not saving them. This is more evil from Google.
Brings into the light the numbers on just how dangerous automobiles are. Few activities have these huge numbers of deaths, accidents, and property loss and damages.
Build your own energy sources from scratch. http://otherpower.com/
As I don't live in a country that's very sue-happy (yet, we're heading that way), yes! Please take the wheel! A snooze on the way to/from work would be excellent, thanks.
I am a viral sig. Please copy me and help me spread. Thank you.
"save more than 1 million deaths per year"
Wouldn't it be much better to save 1 million LIVES per year?
I've said it before, and I'll say it again. You can't take revenge against a computer. A human being killed is a-ok with most people as long as you can take revenge.
Join the Slashcott! Stay away entirely Feb 10 thru Feb 17! Close all tabs to prevent autorefresh!
People are obviously much more tolerant of human error than machine error. Machines in life safety areas are expected to be perfect.
Also who is liable in a fatal accident caused by a machine? People want a human scapegoat.
Put simply we might have all seen too many 'evil robot' movies.
Some people may have seen too many evil robot movies but I haven't seen enough. Maybe cinemas should institute some sort of test before people are let in?
Humans, while all built from the same base materials, rarely share the same OS and app software. The driverless cars [most likely, in the beginning at least] would. Which means that if widely deployed before the bug(s) is/are discovered, they're statistically more likely to kill a whole bunch of people in a short time. Which is why the robot driver scenario is so frightening.
At least, when people die in cars that they drive, the argument can be made that they were in control, and were therefore responsible for their own fate. The notion that you can die through no fault of your own is unsettling, to say the least. It's not a logical argument, of course, as people routinely place their fate in the hands of others on the road in mass transit, but the car has always been associated with independence, and by extension control over one's life.
Obviously the base programming of these cars will be to have them follow the local rules and being computers will be very good at this. Which means that government types will feel free to keep adding more and more rules to satisfy every voter. Thus these cars will quickly stop following the most efficient routes and going the fastest speed that is safe but will end up following routes that take them away from schools, parks, politicians' houses, and whatever whim they want. Even though these cars will soon be able to scream around at full speed safer than cars now they will end up going slower.
Also how are the morality police going to get their rocks off if now you can be passed out drunk in your car?
If the cars are all carefully following the rules and in theory you need far fewer traffic cops then who will catch people who jailbreak their cars into ignoring speed limits?
Lastly in this litigious society who will you sue if an empty car has an accident? The owner, the coder, or the local government who probably designed a crappy intersection or whatnot that induces the cars to crash at that spot.
I would venture to say the self driving car is simply inevitable, as the economic forces behind it are huge. Millions of people will buy additional cars, to replace theirs as well as to get extra ones to take their kids to work without them, create truck and taxi fleets with no drivers, etc. After cars become self-driving, they will become smaller, as they will really almost always carry one person and be used within city limits. That will be basically the same as PRT systems, which exist already. --- http://en.wikipedia.org/wiki/Personal_rapid_transit --- Personal rapid transit (PRT), also called personal automated transport (PAT) or podcar, is a public transportation mode featuring small automated vehicles operating on a network of specially-built guide ways. PRT is a type of automated guideway transit (AGT), which also includes systems with larger vehicles, all the way to small subway systems.
Build your own energy sources from scratch. http://otherpower.com/
Sadly, safety is something that is not handled rationally by the masses. It is mostly an emotional judgement.
Until it becomes mandated and I can't drive. I enjoy driving. I also understand most people would take the alternative to having to do it themselves if given the chance. Which is good, because a lot of them suck at driving. Of course, I'll die, and this generation will be fine with it because they grew up with it.
There is no -1 Disagree.
As someone who works for a-company-which-shall-remain-nameless-but-makes-a-self-balancing-two-wheeled-product, in a word, "Yes." That being said, the argument here is most closely aligned with the argument re: vaccines. Vaccines demonstrably save tens of thousands of lives, stateside alone, every year. And yet, one article by a (now-)discredited quack has thrown the whole program into a political and social quagmire. Why? Because people do *not* think as Spock did: "The needs of the many outweigh the needs of the few." This is, perhaps, most clearly shown by the gun lobbyists: guns kills thousands, every year. Home invasions in which guns could have prevented a victim's death are VASTLY fewer. While my sympathies lie with the thousands who die needlessly, this is not an easy question to answer based on body count alone.
I like the idea of a robot-driven car, but I think the difficult thing is that in the case of a death or an injury, people want to be able to hold a person responsible. It's difficult to know exactly how that would pan out with a robot car. However, I guess one advantage is that you would probably have a 'black box' that could give you a much better idea of exactly what happened.
To be honest, people probably worry about this more than they should. We already have the situation where injuring or killing people with a car is very lightly punished. It's exceptionally rare (at least in the UK) for anybody to do jail time for killing people. You can do all sorts of idiotic things in your car, kill someone and get away with a fine of a few hundred pounds.
A nice noisy petrol guzzler with manual transmission FTW.
Google can take their driverless car and shove it up their arse. No doubt they only want it driverless so the occupants can concentrate more on viewing google ads
I used to work for a company building a machine that screened pap smears. Of course it was not perfect - but it was much less imperfect that human screeners. But the FDA approval criteria seemed to be that the evil machines had to beat the best human screeners in every category of disease, no matter if the category was fished for post facto.
I for one welcome our robot overlords.
If a human with a net worth of negative $10^5 to positive $10^5 is behind the wheel when something happens, maybe one or two lawyers will take notice. But if a machine that was built by corporation X, which is worth $10^9, get out of the way of the lawyer stampede towards the courthouse that will look something like the running of the bulls in Pamplona. Just look at the unintended acceleration claims so far.
I think the real selling point for driverless cars isn't going to be safety, but efficiency. Road maintenance is very expensive. Adding more roads costs a lot of money, and widening existing roads often means tearing down whatever homes or businesses are built alongside them. Driverless cars could use cooperative algorithms to better handle things like lane closures and overall congestion. You wouldn't have free-rider problems (no pun intended) like people cutting in at the front of a line, slowing everyone else down. When a stoplight turns green, every car could start moving simultaneously, getting more people through the light. I bet a huge reduction in rush hour traffic would be a selling point for a lot of people (and regulators).
It would take a long time to implement. And there would be a backlash from people who want to do (possibly selfish) things the algorithms won't. But it's still a neat idea.
Visit the
Evil robots in movies is one thing in a world of fiction. Windows misbehaving, bluescreening and doing strange things, in the other hand, is something usual in this world. And the plenty of malware for it doesnt help exactly. Adding to that scenario the capability of harming people in big scale as isolated drunks do from time to time is not good.
Automobiles are not dangerous (unless, of course, they have inherent manufacturing defects that could cause injury or death under normal driving conditions). Humans behind the wheel of an automobile are dangerous.
To provide a useful comment on the article, I think self-driving vehicles would be invaluable in saving lives otherwise lost through human error. However, as with all things computer-based encroaching on normal life, it will take time and slow evolutionary progress for people to become comfortable with it.
We are already seeing the basic forms of such autonomous systems in cars that claim to automatically brake in dangerous situations or even parallel park on their own. As these systems become more and more reliable and commonplace, and people come to trust them, they will eventually trust them being able to perform more and more regular driving tasks.
Just randomly insert 90 year olds, drunk drivers or 16 year olds who think they're playing a video game when they're driving behind the wheel of the 'driverless' cars...maybe one out of a thousand. Then people will have someone they can potentially blame when the rare driverless car failure occurs.
Otherwise such a thing should fall straight into the ballpark of "tire blowout", "gun accident" or "surfer gets bitten by a shark". They'll all eventually happen, and we accept these as reasonable outcomes. Why not accept this?
In spite of how complex the notion of cause is in a situation like this, in spite of how completely illogical it is, without someone to blame or punish people will feel very cheated if a robot driver kills or injures a human. At least when a human driver does it we can punish them to appease our human desire for blame.
It's a simple matter to put a price on the cost of deaths. Insurance companies do it all the time. Give consumers a $50/month discount for using a robot car and they'll do it. Insurance companies will save money with fewer accidents and be able to pass savings on to consumers. Insurance companies will factor in their cost of liability and litigation and still save plenty of money. Consumers will eventually realize that they can do something else productive while they are being driven, and eventually the only active drivers will be doing it for fun.
What would be the life time of such a car?
Military grade electronics? Nowadays obsolescence is programmed in consummer goods. Electronics components fail on purpose so you buy a new car or get shafted by the mechanics.
The government is in bed with THEM because it forces people to buy new goods and they get taxes on those purchases.
a light bulb could last forever but not.
if that's the car of the future, car will not be owned, car will be a service paid on a monthly basis or according to the kilometers driven.
I don't care. I will still drive my old BMW from the 80s.
but to really foul things up beyond your worst nightmares, you need a computer driving a car.
I agree completely. People won't take the time to notice the statistics or numbers showing how many more live every year. They will get too caught up in having something to blame and many will rally behind it. The problem I see is if 1 drunk driver causes an accident he gets put in jail. If 1 robot causes an accident all the robot systems will take the heat.
Unlike the drunk , who is just one erroneous biological machine out of billions, the failure of software in one robotic car means it would most likely fail in any other such car in such circumstances.
I drive manual transmission cars, I ride motorcycles, and I love going to the racetrack and testing the limits of both myself and my vehicles. Never had an at fault accident, but in the interest of disclosure I was rear-ended while waiting at red lights TWICE.
So while I have a personal problem relinquishing control of my car to a computer because I enjoy driving it myself, I can see the benefits of computer aided driving especially on public roads. But I believe an in between system would vastly improve safety while leaving people in control. Instead of the computer having absolute control, have it perform the same analysis and assist in collision avoidance.
Approaching a red light at a speed beyond safety margins? Apply the brakes. Start fishtailing on the highway? Apply corrective steering measures. Changing lanes into another vehicle, cyclist or turning into the path of another vehicle? Sound warnings, apply brakes, etc.
The trick is setting the thresholds to a level where people are completely in control up to the point where they are somewhat close to having an accident. Because if you believe computer driven cars will remove ALL collisions, you're deluded. All it takes if for a child to run out between two parked cars in the path of another car, and all the computer systems in the world will not counter its kinetic energy.
And it would be VERY important for the vehicle to be usable with the computer systems disabled, for several reasons.
First, because many people enjoy driving. Short of banning every single existing car on the road, people like me will always be able to purchase and drive a non-computerized vehicle. Even today I can buy a functioning Ford Model T. Think about that for a second, and you'll realize it could take a hundred years before the last current car stops being available, short of outlawing them. But just like with cigarettes and alcohol, I doubt that will ever happen. Can you imagine the lobby all the wealthy car collectors will mount?
Second, because computer systems fail and sometimes they cannot be inexpensively repaired. A current car can still run with many of its electrical systems disabled (power seats, windows, navigation system, even alternator and starter) for a while. Having worked with cars and motorcycles for a long time, I can tell you I'd rather rebuild an engine than diagnose an electrical problem. A cold solder on a PCB can ruin a while weekend trying to figure out why your car will not start in hot weather, but works fine in cold (I'm looking at you Honda Main Relay!!!) The complexity of a computer that can drive a car is beyond anything we have available today ANYWHERE, and it has thousands of failure points. Sensors, cameras, gps, servo motors, switches, wires, PCBs and only lastly the main CPU. The fact it runs in testing is great, but these systems have to last 10+ years of abuse WITHOUT FAILURE.
Lastly, having fully computer driven cars will make people even more dependent on technology, which is NOT a good thing. I've had my GPS tell me to go down a railway track once. I looked at it, smiled, and found the real route myself. But people HAVE driven on railway tracks, into lakes or in remote areas where they died of hypothermia. Imagine if you program your car to drive you, without any input, and it makes such a mistake?
It's science fiction, until we can program a creative and reasoning mind.
Yes, we can build warning systems, or even systems that delivers fault free driving in most conditions,
but exceptions happens, and our technology is far from beeing able to handle the unknown.
The margins for errors when driving is frightfully small - we are travelling inches from death, and
even small errors are potentially fatal.
The human mind is excellent at doing fast intuitive reactions, and there is nothing that makes you gain respect
for the brain, than trying to program something that is dead simple for a human to do, like formatting a graph in a nice looking way.
Unfortunately, games that are just playing simple tricks are fooling us to believe that AI is simple and near.
I wont let anything drive me, unless it can also talk about something funny and relevant during the drive...
It's ok for coal to have killed and maimed thousands directly and more than a million indirectly, but a nuclear incident that gives a few workers a dose over limit.....
Here is my wish list for self driving cars:
DOs: Cars must follow all posted traffic signs.
Cars must be able to learn from previous experience and develop routes
Cars must respond to flaggers, police signals, electronic road signs..etc...etc...etc.
DO NOTs:
rely on GPS
rely on precomputed MAPS
require communication of any kind with external sensors or databases.
fooled by spraypaint over speed limit signs.
fooled by obviously false signs (200 MPG speedlimit)
Use of active sensors which can't scale.. If every car on the road is using radar continuously this is obviously a problem.
There are concerns have already stopped thinking for themselves but this "complaint" seem a bit overboard. One of the most monotonous, most error prone, and rarely deadly common activities people in the US do is drive to and from work. Its boring but requires our focused attention. This means the 30 to hour minute drive is often a lost time activity that we do twice a day. A repetitious activity that can easily bore a human and has to be done to time and safety tolerances? These are all of the hallmarks of something that a machine should be able to handle better than humans.
I'm not sure I'd want all cars to be self driving but as a "work car" then why not? Complaining how people abducted their choice to a nanny state because cars drive them to work belies the fact that most people don't seriously or rigorously plan their drive to work anyway.
It seems pretty obvious that the cost of this system will see it installed in high-end vehicles first: lorries and vans (and possibly luxury cars) before it trickles down to the ordinary domestic car. Personally I'd be far happier knowing that the articulated behind me was being controlled by a machine than by a sleep--deprived driver, who may or may not speak the language and is probably more concerned with finding the motorway exit sign, than observing the stopping distance to my vehicle - which is only 2% of its weight.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
To troll, asinine.
...they will have to be mandated and shielded from lawsuits by a gov't with the fortitude to accept the greater good. Look at the hysteria around nuclear power and toyota runaway acceleration. The existence of the Legion of the Stupid and the School of Land-shark Lawyers mean a strong federal shield is needed for the technology to exist for any length of time at all.
The vast majority of human drivers aren't creative or reasonable, so why do you think a robot driver would need to be? If robotic systems surpass humans, there's benefit to adopting them, even if they're not flawless.
Judging from the number of cars I see with drivers blabbing on cells phones while drifting around on the road, people stuffing their faces, digging around the passenger seat, etc I'd say we've had driverless cars for some time now.
It is by the juice of the coffee bean that thoughts acquire speed, the teeth acquire stains. The stains become a warning
I remember a plot point from the Shadowrun games being the GridGuide system, something like this but additionally using a "routing system" using transceivers placed in/by the road. I assume you could use a peer-to-peer system of some sort where the individual nodes sort out the shortest path for any given vehicle and then transmit that route to the vehicles, recalculating it every so often to make up for the changing conditions. People would probably accept a system with some measure of central control easier, especially if there where human operators monitoring the system from a birds-eye view.
It wouldn't do anything for the safety or control of the vehicle versus its immediate surroundings of course, but traffic routing of some sort would have to come into play if you wanted to, say, just hop into the car skunk drunk with an assault rifle and an ork and tell the car to take you to the local warzone-ghetto-clinic in order to put the ork's guts back into the ork.
Emotions! In your brain!
"I'm sorry, Dave, I can't do that. There's a 'No Left Turn' sign there. To do so could only be the result of human error."
Will computer steered cars be able to dodge other dingbats on the road who are: twittering, spilling their coffee on themselves and putting on makeup? That is the real danger on the road. And those are the types of folks who will refuse a computer chauffeur.
Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
As soon as it is proven that computers cause fewer accidents than people do, the rates for manual insurance will rocket. Just like it's now impossible for a teenage man (and when the non sex discrimination rules kick in, teenage women, too) to get any insured for less than several thousand £££'s, so it will be for drivers who wish to be in control, themselves. SO while the law may allow people to drive, it will soon be impractical for reasons of cost. Shortly after that it will become socially irresponsible and after that people will start to wonder why anyone would ever want to. It'll take a decade ot two, but sooner or later the only place people will be allowed to control cars themselves will be on private race-tracks next door to hospitals - provided you can afford the medical care.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
No, put simply, we all know only too well that the people at the top in this world are "race to the bottom" types, as opposed to race to the top. In their hands, there would be years of driving safety, and sparse traffic problems, relative stability, and then suddenly, *bam*, 100 million road accidents in one day, combined with a "terrorist" attack and a stock exchange melt-down.
It's never the technology or the scientists i have a problem with, it's the layer of scumbag sand castle destroyers at the top who inevitably get control of this stuff. Not that the EU is free of any of this but I prefer their proposals, no cars at all in city center by the year 2050 - f@#$kin A!
There are already robotic trains. Paris underground (line 16? can't remember the number) is robotic/AI. We need less cars, more bikes, better mass transit infrastructure, less bike manufacturers and more bike repair shops, availability of replacement parts....
Anyway, the people behind this Goggle/NSA ("do no evil") can't be trusted to take this guys genuinely heart-felt idea of road safety sincerely. Their plan would inevitably be to increase how people can be tracked via smart-car, and be able to take full control of personal transportation navigation. If that's a tax for living in their world, it's still better than being hunted down and ... oh, wait
You have to ask why Google would spend money to develop this. And the only answer is so they can feed more advertising to a "captive" audience.
A million deaths per year sounds inflated. Last year, the us had "only" 42k deaths. I can't believe the rest of the world accounts for 660k deaths, ESP when the US has a disproportionate amount of vehicles.
Stipulating "1m deaths" as fact makes me suspect the rest of this analysis.
--
$tar -xvf
Um, hello... The accident rates can't be compared like that. Human driving accidents stem from a very different fault model versus automated cars. Humans crash because they (tragically) drive unsafely (drunk, talking on cellphone, road rage), and of course this never happens to YOU in particular. A robot driver crashes, and for all intents and purposes, this will happen to any random driver, including YOU. This redistribution of risk the core reason for the rejection; there's a psychological/economical reasoning behind it.
Very right. However, there's one correction - i believe the corrective measures it would take would have to be somewhat passive... the problem i see is changing response to the same input ... the switch perhaps being another foot pedal.
A lot of what you know about driving is to do with the response of the car being predictable. I can imagine a situation where the car corrects the course of drive, such as autobraking or restricting your "gas" even though you are stepping at it full. Once it hands control back, you accelerate due to the above... it tries to counter it by braking, perhaps goes too much, and in effect you drive jerkily and the car behind rear-ends you
Especially if the back car is computer driven too... error propagation in control systems like that is something vicious
A safer way would be to sound a warning and ask for control
Modern cars could easily be programmed to never exceed 80, the current top speed limit in the USA, but there are no regulations forcing this and no cars do this. We could in theory get rid of a lot of invasive search laws if there were no DUI excuses. I don't own a car or drive, so it makes it very difficult for me to be unreasonably searched as I'm not capable of endangering those around me with 40,000kg*m/s of momentum.
In Hell when I trust my life and the lives of my loved ones to an algorithm, no matter how well written or secure. Humans may be flawed and dangerous operators, but unless this system can operate under all conditions and in all environments, human intuition will trump predetermined logic every time. I'd love to see how this system handles a one ton moose jumping out in front of your car while you're traveling at 100km/hour.
You can call it sheer ignorance, but honestly, if driving is such a drag, DON'T DRIVE. Walk. Bike. Take transit. Carpool. Telecommute. I know you want your own personal gas guzzling chariot - who doesn't - but there are already much more cost effective and safer ways to get from Point A to Point B in most urban centres.
Cars really kill a million people a year? That's almost 3000 people a day, you might want to check those statistics. No one in their right mind would trust a computer to drive something that can kill. Planes have autopilot but it doesn't mean pilots go to sleep and trust nothing will go wrong.
think about it. countries like Japan are full of trains. RIGHT NOW they are saving tens of thousands of people from death, per miles traveled, just by choosing to use trains instead of automobiles.
as for the 'million saved', its a bit misleading. there are 30,000 deaths in auto accidents in the US each year. even assuming google thinks it can prevent all of these with its magic car, which is crazy, that still leaves 970,000 other fatal crashes in other places on the planet. like, say, the middle of india, where rikshaws and 2 cylinder cars from the 60s and cows all share the roadway.
is google going to magically donate its technology to hundreds of millions of people so they can upgrade their cars? you realize of course that the auto-driving technology google wants to put in these cars will cost more than the cars themselves? and be instantly stolen and sold for scrap? there are many, many cars here in the US driving around that are worth, say, 500 to a thousand dollars, on a good day. is google going to stick $5,000 worth of technology into these cars? no.
they are not going to 'save a million people from dying in car crashes', because a million people are dying in car crashes in cars that are 30 years old and that google will never upgrade, ever.
what about new cars? fine. google will save one million people, thirty years from now, after all the old cars have been phased out. way to go google!
meanwhile if we had put that money into high speed passenger trains, we could have saved many more lives.
not only from preventing crashes, but from the massive amounts of pollution that are caused by automobiles, the massive amount of wasted public money that could go into health care that instead goes to transport, the obesity epidemic as 'walkable cities' are paved over in favor of the automobile, the increased rates of health problems caused by the stress of driving, the noise and the ugliness caused by the highway system, and so forth and so on.
See subject.
If a bug in your driving system causes an accident who is at fault? Is it you or the vehicle manufacturer? Currently if you a passenger in the car then the driver is at fault. In this case, you aren't driving. What insurance company is going to insure your car if you are not the driver? One possible model might be automated trains but that is a special form of transportation. In the past, if it is a manufacturing defect that causes an accident and it can be proven in court (such as sticking acceleration pedal) then the driver is at fault but those harmed (both the driver and the one hit) will presumably sue the manufacturing company for huge sums of money. If the manufacturer knew of problems but didn't disclose then that make the settlement value even higher. I don't see how this is going to be resolved without special liability limitations for the makers of these systems and that might not be in the best interested of anyone but the manufacturers. In addition, I don't see how insurance companies are going to be able to rate the risk of these systems as drivers but perhaps over time enough data will be accumulated. Insurance works with regards to insurable risks and probabilities based on past risk experiences. With a new technology like this there is going to be some insurance and liability issues to resolve.
Planes have already implemented "robot control" and people are ok with it. It has been challenged based on just a couple of crashes but by that time it had been used so many times that people actually listened to statistics.
We are already started to implement the car version and nobody has complained. Mercedes has a safety feature that will take control if you are about to hit a car or accidentally change lanes. As long as we don't call it a robot and add in the features slowly then it will be accepted. If we try to do it too quickly it will cause concern and be scrutinized to the shelf.
In order to understand our reactions to deaths attributable to machines/computers/robots, we should notice the different reactions in different cases. Several killings by industrial robots were reported in the news although they were in many ways more like accidents with chainsaws and snowblowers and other dangerous power tools than the "robot kills human" sort of headline suggests. On the other hand, the Therac 25 delivered therapeutic radiation controlled by very badly designed and constructed software, and it killed at least 6 patients in a rather gruesome way. These incidents seem much more like automated manslaughter than the accidents with industrial robots, but I never found any mention in the newspapers. The killings went on for more than a year, and early investigations focused on hardware problems, mostly ignoring the software. We need a few good dissertations and more careful studies of more incidents before drawing conclusions. But, it seems that our irrational reactions to machine/computer/robot killings are driven by something more subtle and complicated than a mere autophobia. I would hazard a guess that we prefer to get excited about relatively unimportant incidents with a nice dramatic presentation, while ignoring the rationally scarier incidents that expose real systematic safety problems.
Mike O'Donnell http://people.cs.uchicago.edu/~odonnell/
I agree. Just being able to relate to that driver who's trying to cut across 3 lanes of traffic to make a left turn during a traffic jam makes all the difference in the world (imagine trying to cross that line of perfect drivers; the robots initially would be like "wtf is that guy doing, well not to worry, i've got the right of way"). It's those kind of eye-contact interactions that probably save doubtless lives everyday that the robot has no insight into. And Thrun is definitely right about the court of public opinion. So it will probably be a long haul before this becomes mainstream.
Imagine a logic bomb, or rootkit which one morning, during rush hour, causes everyone to accelereate to top speed, and steers randomly. Entire productive population of a country wiped out just like that. It would take decades for that country to recover
Stupid people are dangerous. Not just to themselves, but to everyone else. They are the reason that sociopaths achieve political power. They are the reason that harmful and unjust laws get passed and enforced, to the detriment of people who have done nothing wrong. Their bad driving puts me at risk for an accident, as does their irrational thinking about what safety features their car should have.
The worse thing about stupid people is that they can be dangerous without meaning to. Sometimes their best intentions can drive them to take monumentally harmful actions.
I wish natural selection would work its magic on stupidity.
It is the driver behind the wheel which makes it dangerous.
And here's the problem with robotic drivers... They are all identical. Every one on a particular model will be byte for byte identical. Which means a fault in one is a fault in all.
Humans on the other hand are all different. Just because one causes an accident under certain circumstance doesn't mean another would.
Deleted
Problem is, this is more likely to kill pedestrians, OAPs and cyclists. Car drivers currently kill other drivers the most, not vulnerable people. So this would move the balance the other way.
Will computer steered cars be able to dodge other dingbats on the road who are: twittering, spilling their coffee on themselves and putting on makeup? That is the real danger on the road. And those are the types of folks who will refuse a computer chauffeur.
Will computers be able to handle the dingbats better or less good than a human driver? _That_ is the question.
BART = Bay Area Rapid Transit. BART is all driverless trains. Well they have drivers, but the drivers open and close the doors. The system is actually all operated from central control and controlled by computers. The ONLY time that accidents occur is when someone ends up on the tracks either jumping on or pushed on or when a driver decides to control the train. So driver less is possible and works better in most cases.
Also many airports have driverless trams that take people around the terminals.
Ok granted this is all on tracks with trains so it is easier. However with todays technology you can have enough cameras and processing power that a vehicle will be able to navigate a road. Now if you have mixed roads, people and driverless vehicles then it will be more complicated.
Just something to point out though, we do have cars TODAY that can parallel park themselves and you can buy one. ( http://blogs.cars.com/kickingtires/2006/09/ls_460_parking.html )
Yup, even educated people are really stupid about understanding probabilities and risk. But I do think that it's an analogous problem.
I know "open source" is not the precise term Bowie we're really need a robust community driven system for map corrections. That way, the jump from a hybrid system to a nearly fully driverless system can take place.
Corporations are not geared for doing this level of critical work without forcing governments to tax the hell out of us.
There will be people to monitor the roads in their communities.
Now think about "bad" neighborhoods
Nobody bothers to paint lane markers in such places in those places will not get mapped as effectively Everywhere else.
Code can keep people from accidentally driving through the wrong neighborhood.
Digital divide, sure. Just pointing out that it would be more efficient than the way it is already done.
I imagine the government would have to set up a strict requirements and tests, and then once licensed the vendor would be immune from suits unless gross incompetence or malice could be proven and that they would also then benefit from an insurance pool to compensate/benefit those injured due to the system with some sort of cap or limit to the payouts.
After that was in place, I'd think government and businesses would jump on it to avoid being sued for their employee's driving errors or for simply having their logo on the vehicle, not to mention long term cost savings of not having a human driver (no more wage, insurance, or tax costs) and therefor not needing a cabin and add 24/7 driving and better fuel economy (from no a/c and fuel saving driving patterns). Heck, UPS, FedEx, and the USPS could all redesign their vehicles to carry more while making it easier and even faster for the employee to deliver the mail or packages.
I'd also think those that can't or shouldn't be driving would also be quick to jump on it. For instance, once you had a system like this in place, I'm sure you'd see laws sprout up prohibiting anybody ever guilty of DUI from using a human-driven vehicle. Seniors that can't drive or would rather not will love these cars as they would give them freedom to go anywhere any time, and the people will love that they aren't driving, perhaps leading to lower maximum driving ages. I'd imagine minivans that had no "front seats", designed more like a limo, would be extremely popular with families. Make a sedan designed like that, with rear facing front seats with a built in table and power outlet, and I'm sure most commuters would jump on it. New drivers would simply be used to it by the time they could even apply for a license, so the number of people against the concept would naturally lower every year.
So those that wouldn't want to give up the control could hold out, at least until you end up with such a huge portion of cars driving themselves, that the population starts limiting human driving (more expensive and difficult to obtain licenses, automated-only rules for freeways, tolls, etc) or driving around them becomes too annoying (I enjoy driving, but if every car was going the posted speed limit, and I imagine those limits would only go down, it would get old very fast!).
drivers not kill each other first before going fully automated? It's like trying to run before learning to walk.
If slashdot keeps coming up with more car articles, someday we have to come up with our anti-car-analogy comments.
Democracy is for the people; you only vote once per season and we'll do the rest of the work for you don't have to.
What will happen to all of the Barney Fife's of the world if this is implemented. With no more speed traps to generate income for local governments, the cops will have to find something else to do or be laid off.
I'm saddened by how common this error is. We are constantly faced with "what's best on a population basis" Vs "what I want for me" questions. Logic has nothing to do with it. It's about values.
Otherwise thoughtful people seem to like to think that they have a "rational perspective", whatever that means. You might be better than "the masses" (whoever they are) at logically solving problems once you have a *goal*, but you can't reason your way to ends. You will end up chasing your ends to a plurality of mutually defensible and mutually assailable assumptions. If that weren't the case, moral philosophy would actually be useful, and used.
Worldwide, maybe, in the US, not so much.
http://www-fars.nhtsa.dot.gov/Main/index.aspx
http://en.wikipedia.org/wiki/Traffic_collision#Fatality
I love driving too, but I'd jump into an automated car without hesitation. I'd even pay _more_ to NOT have a wheel or driver fallback in the car, with the front two seats facing back like a small limo. I'd much rather have a fully automated car than one that includes a built-in backseat driver.
I don't agree about them having to work without failure or handle every single possible situation possible. If the car can't handle it or if something happens, then a secondary system would surely be in place for the car to pull over, while I use an OnStar style service to talk to somebody about it while they diagnose the problem and if needed wait for a replacement loaner car from my insurance company to drive up on its own to get me. I'd also enjoy paying those lower premiums because they essentially just cover roadside assistance and car replacement/repair, instead of liability (would resent still having to insure against uninsured human drivers though!).
And on top of all that, no more worries about my kids (or their friends) getting into accidents from joy riding, being too cocky, getting distracted (from passengers, phones, etc), or worse. Not to mention, once in college I'd know they always had a dedicated "driver" when out. I can't imagine many parents would buy (or allow their dependents to buy) human-driver cars anymore. Goes the opposite direction too, suddenly the grandparents have complete freedom to go anywhere whenever they want, safely and without the fear of all the aggressive drivers. Correct me if I'm wrong, but right there you have the two most dangerous driving age groups no longer driving, so even if I was driving manually, I'd be safer too.
Egad! Given the easy way that software is cracked, hacked and misdirected, we would have one day day when the "crypto" (based on the movie industry code no doubt) would be broken and all the cars would be told to hit those stupid bipedal creatures. The carnage would be incredible. The only safe place would be on small pathways through the parks.
This anonymous reader predicts 3.33 accidents per car, per day, each day, every year if these goggleless cars hit the roads enmasse. It reminds me of the first time I saw a drive up ATM machine with braille touch keys. I had a sinking feeling.
If a robot kills someone, then it is an industrial accident, not a murder. So the insurance companies and the courts already know how to handle the problem.
Hell coal generation in the US releases more radioactive minerals (mostly uranium and thorium) than is contained in all the nuclear plants in the US! If you live near a coal plant you get a higher average dose then living next to a nuke plant!
And the effects of coal related radiation is secondary to the respiratory illness caused by coal particulates released from mining and power generation.
If the coal industry was held to the same standards as the nuclear industry it wouldn't be profitable.
========
CINC, 4th Penguin Legion
the evil, ambulance-chasing lawyers and the politicians who accept corporate bribes (a.k.a. "campaign donations") to make ridiculous laws which have cultivated such a large disrespect for all laws in this country, even though the politicians attempt to bribe voters with "ear marks".
Running with Linux for over 20 years!
Funny. We seem to want robotic nannies to "save labor" (for what, one might ask?), but invade the grapefruit-sized domain of "pure thought" with a few AI functions and everyone goes ape from the sheer soul-less-ness of it all. When Skynet takes over, I hope they put high voltage collars on stupidity and install invisible fences around our anoetic living spaces.
``Tension, apprehension & dissension have begun!'' - Duffy Wyg&, in Alfred Bester's _The Demolished Man_
...which is exactly why it will be hard to accept them if they cause even one death or injury. If I am driving and I make a mistake it is my fault and I have to deal with the consequences. If a robot is driving I have no control over whether it makes a mistake and yet I will still have to deal with the consequences.
The problem is therefore one of trust. I trust (most) people I know to drive me safely - after all there lives are on the line too. However with a robot I have to trust that some random programmer has not made a mistake somewhere in the code and that the sensors and other hardware it relies on will not fail. We already have aeroplanes which can fly themselves and I see nobody is suggesting that we have computer controlled passenger jets...so why should cars be different?
What I really mean is, can we run a train before we drive a car in traffic? Trains have a pretty predictable path, and their only traffic problem is the pretty clear case of obstructions (moving or not) on the track.
In Calgary, we just had our local light rail "C-Train" system fire a driver because he was doing crosswords while driving: ...that suggests to me it is easier to automate. Why do you get a 3-car LRT or subway train every fifteen minutes (when all the cars are separately electrically-powered), rather than a single car every five minutes? Cost of drivers.
http://www.cbc.ca/news/canada/calgary/story/2011/03/14/calgary-transit-ctrain-driver-crossword-puzzle.html
Next up: buses. One "problem" with robot cars is that they would scrupulously obey every speed law and light and make conservative safety decisions (and save all those lives) but would cost time. People in a hurry cut those corners and would continue to do so. But bus drivers already drive like that NOW, so its not a "service reduction" like it would be perceived with a car. Automating them would again allow smaller buses that run much more frequently. The more-convenient mass transit systems would lure people away from cars - and thereby also save lives.
Lastly, only when those easier automations are proven successful, will you be able to build a market for driverless cars.
Taxi. Taxi. Taxitaxitaxitaxitaxitaxitaxitaxitaxitaxitaxitaxi. You are wrong and I'm sick of hearing this severely mistaken idea. I want a car to:
--Let me choose my own cost/benefits.
--Make a personal statement and stylistic preferences.
--Keep more stuff than I can carry with me, including from errand to errand, from day to day, and things for emergencies that others may not think is worth the cost of hauling around.
--Have immediately (not within 5 minutes) and closely (not within a block) available, almost certainly, almost all the time.
--Let me be familiar with its quirks and needs so caring for it is cheaper.
--Fit my individual needs whether it's low milage, the ability to haul a certain amount of gear, tow a certain sized trailer, or whatever.
I suppose you could have a fleet of taxis built to the most stringent requirements 99% of people will need, each pulling a cart of gear as needed (and kept by) the individual but the cost of that would more than outweigh the savings of having the cars as group property and it still wouldn't address individual preferences.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
...and before the grammar nazis strike - yes that should be 'their' not 'there' - sorry for the typo!
I imagine that the autopilot would be mostly for highway use, as that is the most predictable use-case.
And are you happy to be the one maimed or killed when I demonstrate that I'm no able to handle myself responsibly?
Why not? We let drunks, non-licensed, non-insured, aggressive, douchbags, etc drive. I'd rather have a robot control their car instead of them.
Uranium has to be mined, too.
Yes yes, China has a huge problem with safety, we've all seen the numbers, but they are far lower in the US and it makes up a much larger portion of our energy.
It's beside the point anyways, coal is not the only alternative to nuclear. It's not an A or B argument. Maybe there needs to be some space for nuclear in our future, but like any other source, it's not a magic bullet.
Robot driven cars will only exist on special roads that do not allow on human driven cars.
The machines will line up in a specials lane and enter one at at a time. There will be a fence to keep out other cars. See the movie, Minority Report for the idea.
It will replace the HOV lanes that exist today.
Mixing human driven cars and machine driven cars would be a disaster, but with all the cars talking to each other and the road bed, everything would be be great.
Imagine driving 6 hours and not having to watch the road. Turn your chair to talk to the passengers, check your emails, etc.
If it's not too expensive, I'd love to have a robot driving my car. But it has to be optional, I've got to be able to take over and drive how I want.
I *love* driving, when I'm not tired or drunk or on the phone.
To me, this is just like cruise control. I set it to the speed limit when I'm just cruising down an empty road. But I don't touch it when I'm enjoying my drive.
http://en.wikipedia.org/wiki/Traffic_collision
Complete with references.
I can't believe there's so much conversation above about "individual freedom" with respect to your cars: People willing to trade more deaths for the right to be irresponsible and people willing to actually try to have a measured conversation with them on that topic.
You can't drive a rocket car on a city street. Why not? Why can't you mount knives on your bumper? If you don't hit anything, it's safe, right?
Screw 'em. Nowhere in the constitution are they granted the right to risk the lives of others. If the autonomous cars are much, much safer than you driving, then hitting someone will quickly get you sued completely into oblivion for negligence and endangering others. The autonomous cars will dominate due to the insurance costs of non-autonomous ones.
And that's good. People in non-autonomous cars will cost us vast quantities of money due to gas costs, time delays and health issues due to exhausts and accidents. They deserve no support at all.
BTW; auto pilot does not use any of those things you mentioned. For an automated landing, the good old ILS is still used. I system which uses two separate tones riding on a carrier and phased just so that you get an equal balance of the two tones when you are on the glide slope. When you stray up for down, you get more 90Hz tone than 150 Hz and so the system know..oops, I am to high. Left and right uses different tones, but the same principle. Phasing that is hard though, 14 different antennas as opposed to 3 max for the up/down.
This is similar to the fate of automated trains in New York City during the 1960s. GE and Westinghouse rolled out a two-car consist and demoed the technology on the 42nd Street shuttle. It worked for about four years (with passengers not noticing much difference other than harder braking), but a fire near a switch on the line (that wasn't caused by the cars) caused a huge uproar against the project, forcing the MTA to scrap the idea completely. The closest concepts we have to this today are one-person train operation, which the elevated's in Chicago have been doing for years. See here for a better read.
I would imagine that the same issues apply here, even more so considering that cars cause way more deaths than trains/planes/etc do.
There are probably some parallels between the problems. Anyone who's planning to launch a driverless car had better study the rise and fall (sort of) of the nuclear industry.
I think the most important lesson is to not claim that your product is safe. Anything with big moving parts is inherently dangerous. Even if drivrless cars cut the death rate by a factor of a thousand (which seems unlikely) they will still kill thousands of people worldwide every years. If I was the CEO of that company I would say that our car is the most dangerous one on the market, except for all the other cars. I would be open about the fact that people are going to die as a consequence of mistakes and errors inherent to the driverlessness of the car. Sometimes chance will have it so that it will seem as if a car intentionally murdered a person. But you will be safer in and around my car compared to all other cars.
I would probably also get an insurance policy that would pay certain amounts of money to anyone killed or seriously injured in such an accident and make that policy and the amounts known publicly.
To me, it's all a question of liability. Who is liable when something goes wrong? How hard is going to be to hold them to it?
I look at the utter chaos of trying to debug and repair rare errors in existing car software or faulty sensors, and all I can see is a chain of denial, evasion and presure to cover up. Of course an accident may sink an automated driver project - it will be the test of the manufacturer's willingness to accept full liability for the damages. The moment they attempt to say "can't be my fault", it's over.
I like technology as much as the next guy, but do we always have to do something just because it's possible?
If we go with this system, we're totally dependent on Google's software to keep our cars on the straight and narrow, right?
Okay. What happens if somebody (Google, government or both) puts a backdoor in the system so that, at their whim, they could cause you and your car to drive off a cliff or into a tree? Or maybe somebody in power decides that you should be arrested for what you're thinking or posting online, and next time you try to go somewhere, your car takes you straight to the authorities?
Given that recently corporations and government seem to be merging into one entity, I'm not about to trust my life to them.
...aside from a computer's faster reflexes, direct control, and never being drunk, senile, or obnoxious:
A: You're about to be in my way!
B: Sorry, my breaks are shot. I'm going that way. (pointing)
A: Okay so I'll try to edge in behind you. Can you speed up a tad?
B: Okay I'll speed up. I don't have anyone in my back seat so you can fishtail into me on the way if you have to to get by.
A: Thanks.
C: Hey what's going on over there?
A: We're in a risky situation. You should try to avoid us. We think we'll be in these places at these times. (send chart)
C: Okay, I'm moving this way at this speed if I can. I'll keep you posted.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
Remarkably good.
In US the 2008 fatality rate was 1.25 per 100 million "vehicle miles of travel". That's not one fatal error in 80m miles; 2 deaths in one accident is being counted as 2 so the fatal error rate can only be lower.
A few tons of machinery whizzing around at up to (and sometimes over) 80mph, in all sorts of changeable weather conditions and innumerable other unpredictable variables, and still 1.25 per 100m miles.
1 death is one too many, but the main reason there are a lot of car-related deaths is there is a lot of car-related activity. Certainly, a computer is not prone to the human factors in car crashes, but that is to forget how astonishingly good we are when not being a complete asshat and driving when fatigued, drunk or using a cellphone.
All this brings back memories of some movies where the world one day would be taken over by machines, because some greedy business man wants us to believe that machines are better then humans, well i say a big no to any such attempt, our lives has already become slave to technology and in some cases its good but not every where, so for now google can try and find some new ideas and not this one.
The Copenhagen metro is fully automated and seems to be doing all right. It's probably easier to get people to accept automation on a line that has been built from scratch to be automated than it is to automate an existing line with it's history in the minds of the riders and it's legacy technical quirks.
Witness the current crisis in Japan.
Deaths from earthquake/tsunami: approaching 25000
Deaths from Fukushima nuclear accident: 0
What is getting the bulk of the media attention so far? What is everyone freaking out about?
The scary nucular radiation.
People are stupid.
The question is not will robot drivers be safer. The question that will stop this is: "Is a robot driver safer than me?" As long as I overestimate my driving ability, I'll never replace myself. And believe me, everyone thinks they are the safe driver.
From an old British Sci-Fi show.
In the modern world people as individuals are increasingly de-humanized. In many ways we really have become the drones that a lot of science fiction has depicted future humans to be.
This is called "progress."
Forcing people to buy cars that drive themselves would be yet another nail in the coffin that human individuality is being buried in and a sad thing.
If you want to be able to read the paper on the way into work ride the train or bus. Get into a carpool where you only have to drive once a week or so. Move to a place where you CAN take a train or a bus, whatever.
The world is NOT a safe place and the effort to make it 100% safe for everyone all the time is dehumanizing- and wrecking the planet. We build into nature reserves for more land to make ourselves more comfortable and displace wildlife which has lived there for thousands of years- and kill off the "dangerous" animals. And put a few in zoos so we can feel good about our species.
Yes, I believe strongly that auto-piloted cars are a bad thing and that the drawbacks outweigh the benefits immensely, incalculably.
Nobody is FORCING you to drive and don't spout that BS where if you don't drive you can't get where you need to go. That just means you live in the wrong place, fool.
I drive because I ENJOY driving. I LIKE to drive.
I am a huge fan of the idea posted earlier of having far tougher driver licensing requirements. In the US, for one, the exam is a joke. A reasonable and intelligent driver licensing program would likely remove at least 1/3 of the drivers from the roads and maybe even half of them.
Good- they drive like manure anyway.
And here is another argument against self-driven cars: there would be zero benefit to reducing oil/coal dependency: people who drive like crap would still have their own cars to waste fuel, just a computer program driving them.
Put the people who want to have a robot car on the bus whether they like it or not and we reduce reduce driving accidents and pollution in one fell swoop.
Now THERE is a win-win solution, folks!
Linux computers, watercooled, photography
This same dilemma exists with integrating Unmanned Air Vehicles into the NAS (National Air Space). Despite 1474 accidents and 474 fatalities in General Aviation in 2009 for everything from engine failure to pilots having hear attacks, and no fatalities for UAVs, FAA severely limits UAV access to the NAS for fear of one mishap.
If automated cars came about it goes without saying that people will moan about the smallest thing that goes wrong even if the automated cars are much safer. I think it's because people won't want a safe automated car. They want to take stupid risks and risk everyone's life to get home a few minutes earlier.
I rather see more trains and and a return of street cars in cities and other ways where we can move a lot of people through means that don't use petrol and instead use electricity. It will move any potential pollution out of the cities and it certainly has to be easier to make power generation more efficient if it's centralised. Likewise if the main mode of transportation is run by companies rather than individuals then we'll see upgraded more efficient transportation happen sooner.
Could this be seen as the natural progression past cruise control? As far as the human-machine interface is concerned, when I run on cruise control, I keep my foot hovered above the brake in the case of an emergency. Wouldn't automated driving also require constant user attention in the case of a rejection to manual control?
Sex sells, right?
So how to use sex to promote driverless cars?
One good thing driverless cars could be good for...sex on the motorway!
Yes and in the future when we are all using robot cars people people on this site will complain that their being harassed sued arrested etc. for trying to mod chip the car computers and why can't they run open source software ala Linux, FreeBSD, NetBSD etc. on the computer and still drive 65 on the Freeway.
Whether a robot drives the car for us in the future or not some things will never change.
Have my car drop me off, go park itself and when I'm ready I'll call it on the cell phone and it'll pick me up.
Now I could go for that!
I killed da wabbit -Elmer Fudd
"However it may be OK for human drivers to kill millions of people each year but..."
Could you please post a link to that piece of bullshit? Here's mine for "According to the World Health Organization, 45,100 people died in automobile accidents in the United States in 2002, which made the United States rank fourth in that category. China ranked first, with 250,000 automobile accident deaths" and "Traffic fatalities in 2009 were the lowest in 60 years".
Thank you.
Google's driverless car could save more than 1 million deaths per year
What curious wording. Most safety inventions would strive to "save more tan a million lives, but this one wants to save more than a million deaths. I guess now you can just use any words in a sentence and expect people to figure out what you intended.
I'm an American. I love this country and the freedoms that we used to have.
How about if we make all DUI drivers ride in driverless cars? That would get rid of the majority of accident victims right there, without killing anyone.
And if the car did crash, it would just be a drunk driver who was killed anyways. Who needs them.
The best thing google could do is drive these cars for lots of miles without an injury.
Does anyone know the accident rate per mile driven? The death rate per mile driven?
If google could better that by a factor of ten, that would help.
Better yet, they need to do driverless buses in all the big cities.
Bus drivers falling asleep, having heart attacks, texting would be a thing of the past.
wake up and hold your nose
THis Luddite tendency of americans is the self imposed glass ceiling that keep them teathered to yesterday's technology while the chineese and other forward thinking nations grow their technology divide and leave them as the science equivalency of a third world nation.
A few years ago when studying for my CS master aimed at "Intelligent Software Systems", I recall reading about a trial to replace doctors with so-called "expert systems".
The idea was to build a logic-database of all known diseases and symptoms, with a "20-questions" type frontend. A trained nurse would sit down with the patient and the software, the software would ask questions about symptoms, and the nurse would conduct tests and take readings. Finally the software concluded with one or a few possible diagnoses of the illness.
In the trial it was noted that the software+nurse combo produced significantly more accurate diagnoses than the doctor+nurse combo. I don't recall the exact numbers, but I think it was something like a 1/7 mistake for the doctors and a 1/20 mistake for the expert-system. Mainly, IIRC, it was much more thorough, and did not make faulty assumptions the way the doctors did.
Finally, the trial concluded that fully implementing the system would be close-to impossible, since very few patients could be convinced to "not see a real doctor."
The place to enjoy driving is on tracks and road courses. Much more fun than public roads, and much safer, both for the drivers and the public, too.
I'm sure that, whatever laws are passed around self-driving cars, you'll be able to take your car to the track and enjoy some hands-on driving in an environment that's designed for it.
I've been picturing the day when this becomes a reality. I see myself in a RV, get off work, hop in the RV, go to sleep, and wake up at the beach/mountains. When vacation is over, none of this drive all day/lose a day mess. you have a casual dinner, go to sleep, and wake up in the driveway ready to go to work.
Oh yeah, speeding would be eliminated, so there goes the only source of revenue for speed trap cesspools like Carlinville, IL.
Fuck your stance, mikejuk. And fuck your propagation, timothy.
Put simply we might have all seen too many 'evil robot' movies.
It is my opinion that you haven't seen enough. Fiction though they may be for the current, we have watched many fictitious plotlines and ancillary details become a reality.
Not only have sci-fi novels, movies and television shows produced ideas that have transformed our world, they continue to provoke powerful people to pursue ambitious goals. Military staff, look at sci-fi movies like Predator and Robocop and think to themselves "Oh boy! Lookie lookie, this is a such a neato keen 'DARPA-hard' problem! If I can get funding for some kind of awesome program, my career will be set for life and my wife will love me again.", with an astounding neglect for all the implicit oligarchy and misanthropy, not simply behind the villain's motivation in the sci-fi story, but behind the very idea or war itself.
You neglect to understand the idea of chain-reactions and feedback loops. Very real concerns, when considering that AI may find a way to control very powerful heavy machinery, never mind deliberately malicious weapons of war.
But I bet if we labeled it malware or someshit, you'd jump up and foist on us ideas like "herp-dee-derp! encryption will solve everything!"
Get. Bent.
Who's gonna save us from overpopulation? We need to kill people to avoid the Malthusian catastrophe, that's why cars were invented.
The biggest advantage for self-driving cars is for those who can't or aren't legally allowed to drive cars now. Large numbers of old people can't drive for a variety of health reasons. Youngsters can't drive for reasons of lack of good judgement and driving skills. This would greatly enhance the lives of oldsters and free up the time of people who now have to drive them around. A lot of ambulance and taxi drivers are going to have to find new lines of work.
Contribute to civilization: ari.aynrand.org/donate
It is highly unlikely that there is any car on the road today that has a fundamental flaw which would intentionally CAUSE an injury, death, or accident. Of course, that doesn't mean that the driver can't manipulate the vehicle into causing one, intentionally or accidentally, but if the vehicle is driverless, then ANY accident would be entirely the fault of the vehicle itself.
Fictional movies to the contrary, cars don't have a tendancy to just blow up, or otherwise perform in unexpected ways. The recent hoopla over toyota's acceleration problems were a big indication of just how unlikely these things really are. Extensive measures are taken to ensure that everything works as expected and nothing abnormal is going to happen. It's going to take a human driver to screw something up. The car won't be the problem.
However, one accident caused by the Google car will be entirely the fault of a faulty design of the vehicle or the artificial intelligence driving it. The sad fact is, no matter how many years...decades of testing we put such vehicles through, there will be no way to absolutely ensure that the vehicle as a whole will not cause an accident, even though, in theory, that SHOULD be possible. What makes this matter worse is the fact that the accidents that the vehicle is likely to be involved in won't be the dramatic 40+ car pileups that occur as the result of 3 fuel trucks suddenly colliding into each other at 90 mph on a busy highway... No, the car will handle THAT situation perfectly fine. However, while travelling at 2mph, a toddler will stroll out into the street and get run over and the car won't register the obstacle or the impact, thereby causing a severe injury where normally it would either be completely avoidable or severely minimized. Trust me, running over a toddler will definitely go a long way toward killing the project. Even the POSSIBILITY of running over the toddler will kill it.
The thing about the toddler example, if you, as a human who's paying attention to his surrounding, spots a yard with a toddler in it, you're likely to slow down a bit and be more cautious JUST BECAUSE you will be wary of the fact that the toddler might run out into the street, and you want to be ready and able to stop in time. The google car won't easily be able to distinguish between the toddler and a fire hydrant. One could jump out in the street, the other one likely won't. The car therefore needs to be observant of EVERY possible potential obstacle and take preemptive preventative measures to avoid them. It's going to have to be ready for that fire hydrant to jump out in front of it, and slow down for each and every one of them JUST IN CASE. This action will likely not make the vehicle desirable in any way, especially if you're the one driving behind it.
So, not only do you have to be able to show, in advance, that the vehicle will actively avoid all possible accidents as well as not cause any, it will also have to behave in a way that is efficient and not too annoying to humans it's sharing the road with.
-Restil
Play with my webcams and lights here
Driverless cars guided by artificial vision are totally new technology; nobody knows the implications of widespread use of such cars, how they can fail, etc. Furthermore, nobody understands legal and liability issues. It is rational to be reluctant to adopt them. We have centuries of experience with human-driven vehicles. We know what the failure modes are, who is at risk, and who is responsible when something goes wrong.
Furthermore, driverless cars don't solve the basic problems with automobiles: they are inefficient; we should instead invest in better city planning and better public transportation.
Getting out of bed each morning entails risk. Walking out your door entails risk. Even staying at home involves risk. In life, risks are inescapable but they are not necessarily intolerable. I take the risk of being killed on the road every day that I enter my vehicle and drive it to and from work. That is simply part of my reality right now and I have learned to accept the risks associated with it. Indeed, I was once very nearly killed in an automobile accident on the way home from work (no doubt that would have made my enemies here on Slashdot very happy). So yes, to answer your question, I am willing to risk being maimed or killed by your irresponsibility in exchange for living in a free society where I can choose to drive to and from work. Would I be happy if it happened? Probably not, but I can live with that.
And when the auto-pilot fails, the human can take over,
and make a problem into a bigger problem.
And then the computer tries to compensate, and the whole
thing spirals out of control.
And can you cite an example of a coal plant that has forced tens of thousands to leave their homes for decades due to high levels of radioactive isotopes? I can't. But I can certainly cite one for the nuclear industry and perhaps a second that is unfolding before our eyes. I am no fan of coal but comparing the risks of nuclear to coal isn't practical. A public company cannot survive the monetary losses they will incur with a large nuclear power plant disaster. The losses are so huge it will break them and then guess who has to step in and pay. That is what scares people about nuclear power plants - not dying of the radiation immediately but what it does to people's lives. Forcing thousands of people from their homes cannot be understated and scare the heck out of people with children particularly. If you don't have a child yet you really can't fully appreciate that aspect.
Until you're in a war.
This was a nice read... thanks! but did anyone here about that new +1 Ranking System that is being offered by Google? I have to admit, at first I thought it was a prank too! There usual April 1st nonsense... but it turned out to be legit, you can confirm it here I think it's pretty cool and it's definitely something that is far - overdue. This will allow Google to finally kind of "police" those sub pages a whole lot better without actually "policing" them, if that makes sense... there leaving it up to us! But I have to admit it seems very similar to facebooks +1 ranking system, or Digg's (digg system, but none the less I think google will make better use of the info. What do you guys think? Let me know your opinions.
Cheers guys,
Jeremiah
Chairman & CEO
The Nerd Blurb
Yep. People don't ever care about people dying. They only care about how they are killed, and who is doing the killing. We (humans) are much more interested in stories than numbers.
I think it could maybe fly though, as long as you still have the option to take the wheel and disable the autopilot. The freeway for instances is such a controlled environment, why not switch on autopilot when you get on the freeway?
But it's not going to happen. Or at least not soon. It's another silly futurist dream which also cares more about the story than about reality.
...with respect to aeroplanes the degree of automation these days compared to 20, 30 years ago is astounding, for precisely the same reason: it's been shown to save lives a few orders of magnitude more times than the ones they take.
This is, I think, the root of the problem. Using statistics works fine for aeroplanes because, every time you fly, you are essentially getting onto a random plane with a random pilot and so you want them to be very safe on average. However when you drive it is always you driving so, while you might on average improve the quality of driving with computers that is not what is important to each individual anymore. The question which needs to be answered is "is it better at driving than me?" and since it has been shown that we generally tend to think of ourselves as better drivers than we actually are convincing people will not be easy!
For example, supposing there is an idiot driving and they do something stupid will the computer be able to handle some crazy situation it might not have seen before? Will it speed if it needs to to avoid an accident: very unusual but I know one person who was about to overtake a lorry on a motorway and saw the load start to slip so they floored it to avoid being in the accident...would a computer be able to handle that especially since it will have speed limits pre-programmed? Statistically these are rare situations but convincing early adopters will be hard because you won't have enough statistics to know how often an unusual situation which the system cannot correctly handle will occur.
Nope, automated trains aren't dead.
https://secure.wikimedia.org/wikipedia/en/wiki/The_Plane_Train
Or even, it appears, dead in NYC.
http://articles.nydailynews.com/2009-02-24/local/17916708_1_nyc-transit-trains-subway-cars
We install seat-belts and make cars safer and smarter. How do people react? By driving faster while talking on cell phones, half asleep. So, maybe we can have a robot drive for us, because, you know, that'll solve our problems.
Five years after this technology is ubiquitous I can image many Darwin awards being attributed to clueless people falling out of their moving cars.
Technology can only defend stupid from his surroundings for so long.
What about motorcycles?
If we don't trust robot cars, we shouldn't trust robot elevators.
It bugs me every time some new technology is imposed on the public whether they like it or not. When I drive I like to steer the vehicle, decide when to change gear and when to apply the brake. There's a certain pleasure in driving. If all you want to do is get from A to B then fine, that's your choice, go ahead and use a vehicle which can get you there without your own intervention.
You can't really save anyone's life. At best all you can do is postpone their death. We're all going to die and one day, whether you like it or not, that is where you're headed. You might as well live a bit until your time comes, and if that means taking a bit of a risk, then so be it. We'll all be dead soon enough.
The idea of a driver less car, cars driven by robots to save lives is a great concept to explore and develop for the future. If we can save lives using the driver less
(robot) the cost is a good investment.
Doesnt this sound familiar? Its very similar to the panic over the japanese nuclear power plants
The problem is not with the computer. It is not with the hardware. The problem is this very complex hardware has to be maintained. I don't know if you've ever seen the state of repair of most peoples cars, but a lot of them are really scary.
I have no doubt that Google can make a brand new Prius, with brand new hardware, drive itself. I am not so sure they can do so for 200,000 miles over 10 to 15 years, being maintained by Edwin Luddite who doesn't have the cash or the desire to service the thing.
People will drive cars for years without ever changing it's oil. They wear the brake pads all the way through the pad until metal is grinding on metal. They run into things at all sorts of speeds. They ignore warning lights and run tires until they burst. They try and repair things with duct tape and bailing wire. Teen's chop off and bolt on anything that they think will make the car go faster.
If we can demonstrate that end users are incapable of maintaining the machines we have now, how can anything possibly good come from making the machines we don't maintain orders of magnitude more complex?
Maybe ASE will finally get it's dream of a union, and it will become illegal to work on your own car without being a qualified automotive engineer. Cars will have to disable themselves entirely if maintenance windows are missed, and it will add large amounts of costs to repairs.
Self driving cars would be great, but only if you can guarantee nobody will ever be able to mod them in their driveways.
"Running with Linux for over 12 years!"
Aren't you tired yet? :)
Vote monkeys into Congress. They are cheaper and more trustworthy.
I did some contract work for the MTA. They're a bunch of bureaucratic, money grubbing, egotistical idiots. Each little group was more concerned with making their little slice of the project more important than everybody elses'. For example, the line power guys (3rd rail) wanted to provide backup power to stations in the event of a power loss instead of battery backup. They kind of forgot that a power loss kills the power to the rail. When it was pointed out to them, they said, "meh, it'll never happen." This came just about a year after the big blackout in NYC.
Vote monkeys into Congress. They are cheaper and more trustworthy.
Funny no one ever talks about what happens at the world;s hundreds of nuclear power plants after a war. Forget the reactors, they'll just mostly melt into containment, but the spent fuel ponds have tens of megaCuries of contamination. The little fission pits of nuclear weapons are nothing next to those fuel pools.
Just saying, most of the time I would prefer the robot.
I'd prefer they have default options and Advanced Options
The tech crowd is an important and maligned segment - you absolutely need the ability to expand into the Advanced panel to do stuff. But it's okay if it takes two buttons to get there.
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
Which is why it will take a long long long time to ever have a driverless car in America. Your car is a part of your personality. I remember the day I got my license when I turned 16, it was my first taste of real freedom.
In a world with driverless cars will there be sports cars? Why would they even exist. Think of the classic american idea of the open road, driving down the PCH with the top down on a convertible. There is so much more to driving than the everyday commute.
There seems to be a geek tendency to hate driving that I see on slashdot all the time. Not sure where it comes from. But a lot of people love it.
Of course driverless cars do make a whole lot of sense. Even I will admit that, and I am someone who loves cars even more than I love computers (which is a lot). The way I see the transition happening is things like dedicated computer operated highways. You drive to the onramp, get into a queue, and the computer will take over control. It dumps you off at an offramp, stops the car, and you can regain control.
Install Windows Me.
Here. Lousy subs and editors linking to random shit-ass blog posts instead of direct to ted. I'm not turning on javascript for some regurgitator to earn ad views. Is this enough text to get past the lame filter? Posting anonymously to not karma whore. Not sure if the "no karma bonus" thing works. Let's try this.
Advanced options don't come for free. Not only is there a button to get you to them, there is the advanced options screen itself which novice users may find themselves in by mistake. They may stop the application working in the way they want, and not be able to correct it, leading to support implications. And finally, having more options leads to more code, with more defects. With a multiplicity of options it becomes impossible to test all combinations.
Usually, excessive options come from a broken development methodology, where UIs are not designed by designers nor tested on users. It happens when programmers have a choice of two or more ways of doing something, and they are unable/unwilling to make the choice themselves and so pass the responsibility on to the user.
With a decent UI designer on the team, the programmer will have to ask the designer to add the option. And the UI designer, knowing that the best UI is no UI, and the better UI is minimal UI, will ask what they can do in order not to need to have that option.
For sure there is a market for the tech crowd who want lots of options. But that's a subset of the tech crowd as a whole. There's also plenty of techies prefer minimal UIs. And as a whole the tech crowd is far smaller than the non-tech crowd.
It's good that there are multiple devices on the market so people can get what they want. But the mainstream devices should try to reduce options to a minimum.