Self-driving cars projected to reduce injuries by 90%, save $450B annually
techspot.com
techspot.com
Probably in the hundreds of thousands globally.
Self-driving cars can't come fast enough.
I'd explain how there was a time when we were free to drive our own cars, and make our own decisions. And simply hop in a car and drive.
Not like now where you get in the car and it asks you to download the latest software update, then it monitors your every move, sending back data to the car makers/NSA and showing you ads as you drive along at ridiculously slow speeds. It'll probably also drive you past a certain shop if they pay the car maker enough money! Oh no, the car has crashed. Better turn it off and on again and hope for the best. Nothing like putting my life in some low paid software engineers hands! Oh dear, the car maker has gone bust. Now you can't get any software updates, and the software has stopped working as it can't dial home anymore. Oh well, tough luck! You didn't really think you could buy a car to own and do what you want with did you? You can only pay subscription fees for cars now.
Freedom is certainly a double edged sword. With freedom comes responsibility. But taking away freedom isn't the answer. Making people more responsible is.
It's going to be pretty impossible to take away the peoples right to drive cars (Even harder than taking away the right to bear arms in the US!) Stop drinking the kool-aid. It's not going to happen.
I'm rooting for some massive disasters/scandals with self driven cars, to put the idea to bed. And you can be pretty sure there will be some big disasters if they ever get anywhere with it.
I also agree with you about trusting my life to a disgruntled software/firmware engineer: I'd rather not. I know people like self driving cars because 'ZOMG! Computers react so much quicker!' but I'm not convinced that they will react better, or at least, that they will react better in critical circumstances. I don't want a software bug, a flaw in an algorithm, or a dust covered sensor to kill me.
So the better question is whether you trust the median driver more than you trust the process used to approve the self driving system. This is a much different test than your disgruntled engineer.
In an ideal world I would not mind self driving vehicles nearly so much if everyone had one. Advocates of self driving too often forget that there is an adoption period for technology and that for a vehicle it will span a decade. During that time, why I should I trust the firmware/software to react appropriately to other people's insane/suicidal driving maneuvers? For that matter, when those people are a minority on the road, I'm not sure that they wouldn't become worse, engaging in ever more stupid maneuvers because they can assume the other car is 'smart enough' to move out of the way.
It's a great idea to remove the human element from driving completely in 99.99% of scenarios. The problem is that its the 00.01% of scenarios that kill you. The other problem is that adoption doesn't happen overnight and during that stage I see no logical reason to put myself in more danger by removing my ability to react to my surroundings and instead trusting that the software of my car can handle the suicidal driving of angry high-way drivers.
If you keep that rule in mind at all times, you'll be pretty safe.
I would absolutely trust a random driver far more than a self driving car.
If you want to keep driving as a hobby, I'm sure there will be some accommodations made. But just as you can't tote a rifle in a downtown mall, there will be some restrictions to make sure you don't endanger anyone.
The younger generations I know can't wait to drive, and have the freedom to travel where they want to.
I think it's simply the ability to be able to go wherever that lures teenagers to driving. I know that's why I got a job and saved up for my first car. You can still do this when cars are that don't require a driver.
No, they will simply phase out manual cars. No big deal, it just means it will be almost impossible to crash, and that's a good thing.
It doesn't mean you won't be able to drive, but you'll be driving with a computer keeping watch over the road ahead and obstacles around you.
Put the car in auto mode and sit back... Most people will choose that. Those who prefer to drive themselves can do, but within limits. You couldn't for example deliberately crash into the back of someone. There won't be a setting to turn off to achieve an outcome whereby the car touches any other object. It just won't happen while you're driving on public roads. For example the computer will not let you cross over double lines at certain speeds, or accelerate too fast at busy intersections, and so on. But it may let you accelerate fast if it determines that doing so won't put anyone else at risk, and conditions are suitable etc. The algorithms controlling when the computer takes over will be clever enough to almost be invisible, and that's what will win people over.
Regulations will prevent car manufactures from making fully manual operated vehicles. Every car will be required to have automatic safety overrides for any action deemed to put the car at risk of impact with another object.
Everyone thought it would be people who override the computer, but it will be the computer who has the final say in the direction and speed of travel in a hazardous situation. I think car manufacturers and the public will LOVE the fact that driving is safer.
Any driverless car accidents will be statistically unlikely, and everyone will see this. Even when just one car crashes, it will be headline news each time, like commercial airline crashes are now. People continue to fly because accidents are rare.
So, the "right to drive" won't actually be taken away. Your right to put yourself and others in danger on public roads WILL be taken away. Have a cry, but then you'll need to get over it because 450 billion is significant.
Can you not see the idiotic flaw in your plan.
You're driving along. You're coming to a junction. A bike isn't looking and is going full speed and will hit into the front side of your car. You do not have time to brake. Your only option is to swerve into the next lane, which will mean you crash into a fellow car.
You have crashed into a car, but you have saved the cyclists life.
There is absolutely no way in the world you'll be able to get regulations in place to prevent making fully manually operated vehicles. You're high. It's not going to happen.
There's also no evidence to suggest that driverless cars will make roads any safer. You still have to account for cyclists, pedestrians, animals, and the billions of cars driven by people etc etc.
What drive less car can do and the system in the whole can do is to be able to know where and when a car is driving in the vicinity where normal human cannot perceive. So the situation you describe will not happen at all if those systems are in place. Actually, you just describe a scenario where it is preferable to have a system that monitors human error that could prevent accidents like the one you describe to happen.
Also I don't know whether you have been following tech recently, and if you have seen the intro to drive less car, you would probably seen or heard about the infamous laser and radars mounted around the car as shown in the article in the link: http://www.slashgear.com/back-to-basics-how-googles-driverle...
It is virtually impossible for the car to crash into anything because the car can literally "see" everything around it. Also the algorithms is smart enough to anticipate or plan an alternative path to prevent collision with the objects. You sounded like you have never heard or been following tech especially machine learning, so I forgive you, but you should do some research first before posting comments like this...
It's not my plan. It's the inevitable path of innovation and technology. What I described should be obvious.
Eventually, all cars will be driver-less, and the "old style" of car will be rare to see on the road. Like pre-1974 cars are extremely rare to see now. Everyone will want the newer, safer, cooler, energy efficient cars that drive you home safely, even when you're high.
"You're high. It's not going to happen."
Yes, I'm high. But it will happen. The bicycle situation you describe is one where the computer could still take action to minimize injury to the cyclist, and reduce the risk of causing a major accident with other vehicles. Most drivers would panic and "swerve into the next lane" as you describe, which is not the best action to take. It's about controlled evasive action, not "holy crap... slam on brakes...over-steer... enter skid... over-compensate over-steering... enter another skid... crash."
If the cyclist collides with your car because of cyclist error, then what you have firstly is an extremely unlikely situation, as cyclists are usually focused on what they are doing. They're not adjusting the air-con, or thinking about sex while riding through traffic - they are concentrating.
"You still have to account for cyclists, pedestrians, animals...".
Come on bolder88, get real. Driver training teaches us to avoid swerving into the path of oncoming vehicles to avoid impact with wildlife. The computer would not allow you to do so, and that's how it should be. However, the computer WOULD allow you to swerve into the other lane if there was no oncoming traffic. Otherwise, the safest option is to run over the animal.
With pedestrians, it's certainly a challenging scenario for the driver-less car, but I still think that the automatic hazard avoidance systems can save the day better than manual intervention. That's because it takes great skill to control a car in an emergency situation. Most drivers don't know how to control a skid, because most drivers don't get any practice with such driving.
Perhaps then, to meet you half way, some drivers can apply for a special license that gives them full control. For the rest of us, the computer is the safer option for controlling the vehicle during an evasive maneuver.
Anyway, your right to drive a car probably won't be taken away. Your ability to afford the insurance necessary to operate on public roads is probably a different matter.
a) We let our freedom be gradually eroded over a long period of time, so that people don't notice it, and so that any one generation can't really remember how great things were.
b) There's a wake up call event that shows people we should not allow ourselves to go down this road at all.
During the Vietnam War, more Americans were killed in auto accidents on the DC beltway alone than in the actual war. Auto accidents are one of the leading causes of death, and driver error is the leading cause of auto accidents. No, I think you just want to watch people die.
Auto deaths are almost half what they were in the 60s in the US, despite more cars and more miles driven. And that figure is coming down by a pretty good percentage each year.
http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in...
To put in in perspective, every year in the US:
* People die from heart attacks: 600,000
* People die from auto accidents: 32,000
Getting human-operated cars off the road doesn't diminish freedom, it enhances it.
Insurance companies will probably love to be collecting rent on every vehicle while having to cover 90% less accidents.
I wasn't aware they wanted it now. People buy it because they are legally obliged to.
People don't like the cost, but they like the benefits when they need it.
This is going to be much bigger for the rest of the world than in the roads-a-plenty USA.
Pesky governments building infrastructure. In a libercrazian future, the free market would make such trivialities free!
There are many countries around the world where the government has no money to take care of the infrastructure and you need private ventures to step in.
Well, when you have 15 trillions of unpaid debt behind your door, I can hardly call that functional. If it were any private business they would have gone bankrupt 50 years ago. If you cut the money supply, the government stops in the hour.
I would love to argue more, but my Beijing to Detroit flight is boarding.
if 1/2 of your fare went to the driver normally, now it can go to having more cars available.
All I'm saying is the economics might work
Honestly, the biggest issue (assuming the technology is flawless) is going to be keeping them clean and maintained because of the very high usage rates.
Accidents happen to other people... until they don't.
This only follows through if you don't introduce new technology and innovations to your software (and thus new bugs), which isn't going to happen.
If you look at air travel, the number of fatal accidents occurring each decade since the 1950s has stayed around the same, even though we've significantly improved automated systems and reduced the potential for pilot error.
visavanv, it looks like you got hellbanned between 10~14 days ago, possibly because of: https://news.ycombinator.com/item?id=6548892
This is just it. Most people aren't rational, and they don't think statistics apply when they are causally involved. How many people have you met who think that "don't drink and drive" doesn't apply to them because they "can hold their liquor" or "have had enough practice driving drunk"?
The match up isn't even when you have "random thing happens and you die" on one side and "dangerous situation happens where your personal ability plays a small role" on the other. The only way I can see to spin it is to convince people that they don't want to be on the road with other human drivers.
Insurance companies can take care of that too. Just charge a huge premium to drivers who want to drive by themselves. You'll see even the most confident macho driver submit to the AI overlords.
That said, it does seem inevitable; just a matter of time (perhaps a lot of it).
Finally, I suspect it will be necessary (more for PR than technical reasons) to require a capable driver ready to take over in case the autopilot were to fail, much like with commercial airliners, at least during the initial adoption period. And if so, it also seems inevitable that the majority of crashes ostensibly caused by these vehicles will in fact be due to drivers taking control and making poor decisions at inopportune times. You know when people complain about GPS because it led them astray, but it turns out the real problem was they didn't trust it in the first place, so ignored its directions?
In fact, while I don't know this for a fact, I would highly suspect that the number of airline accidents that are the result of a flaw in the automated systems is dwarfed by the number of accidents caused by humans overriding the automated systems.
(If you want a feel for just how automated flying has become, I particularly like this video of an A380 landing at SFO: http://www.youtube.com/watch?v=k3HKN-FWNq0 . Just look at how much the pilots aren't doing.)
Cars can't stop instantly and the energy has to go somewhere.
I am also led to wonder, who will be implicated if an accident happens? What if the auto-driving car has a bug? Or if it was infected with a virus or hacked? What if the driver deliberately modified a self-driving car to cause injury to a target?
The same as what would happen today if someone mess with the brakes?
Consider: Car #1 suddenly realizes (for whatever reason) it is going to impact car #2 or car #3 at approximately equal energy for either crash, pings both, #3 isn't "smart", no reply. #2 pongs back, car #1 heads for car #2, reports this to #2, car #2 deploys air bags proactively using a lowered-energy inflation mode since there is some advanced warning of the impact and sets its own transmission mode and braking (or lack thereof) to be optimal for safest possible (to inside occupants and other detected cars in the impact area) energy transfer.
This is not a new problem. Automation has already rendered a very large part of the world's population economically unnecessary.
Anyway, truck driving is exactly the kind of job (e.g., http://io9.com/5933246/why-truck-driving-is-one-of-the-most-... ) that we should make sure that no homo sapiens should be forced to do in the future.
Also call me when they get to tens of millions of miles, eg what's driven every day in the US.
http://en.wikipedia.org/wiki/Google_driverless_car#Incidents
I'd rather have them say, we can encountered 100,000 different conditions (and combinations) and we know how to react to them.
I'm more concerned of what these cars will do to highway capacity and therefore traffic congestion due to the excessive lag/follow distance their algorithms impose. While they should be following at < 1sec lag due their super-human reflexes, what you actually see (http://www.youtube.com/watch?v=TRIOE1IZrq4) is sub-human follow times and it's apparently the only car on the road that's frequently pumping on the breaks (which consequently makes following cars increase their lag time as well).
I've also read that they won't exceed the speed limit, which is significantly more dangerous than driving with the flow of traffic. I hope it knows enough to move over when being overtaken from behind. If you're driving 65mph on I-280 between San Fran and San Jose with no traffic, you are the slowest car on the road by between 20-40mph. The road has long straightaways, often 1+ miles of visibility, well paved, and I'm often overtaken while going 90mph.
I expect that the experience of having to share the road with these devices, as they start to become more common, will transition from, "Gee that's cool" to something more like "Geez, get off the road Goorama!"
I don't see any way they can exceed the speed limit right now - the legal liability alone would be huge. The best solution to the problems you're posing would be to adopt an German autobahn-like system.
It seems simple to give the user control of a 'speed adjustment factor', and hence transfer liability to the user. Should be no different from adaptive cruise control letting me pick a speed. Cruise control doesn't stop working at 100mph.
Personal liability for operation of your automobile, regardless of the level of digital augmentation, must be very hard to relinquish.
I mean, they are willing to take over steering the car, but oh no, they can't break the speed limit while doing it?
More realistically, they don't break the speed limit because their error tolerances fall too much when they do.
But that was some time ago - were there other accidents, they've probably driven much more miles now?
And regarding speed-limits - I think that it's a great thing; speed limits need to be set to an honest number where I cannot get ticketed, ever, for simply driving at the appropriate speed, but apparantly the only thing that can force the bureaucrats to update those limits will be an army of robots that actually try to obey the law.
Any sufficiently advanced self-driving AI will try to avoid being rear-ended like that. Even better, when a collision is unavoidable, an AI has a much better chance at making the 'right' decision as to how to crash.
All these things are near impossible to an average human driver, but there's no reason why an AI would have more trouble looking behind than forward.
* If there are enough self-driving cars, it will also (hopefully) teach those drivers who feel entitled to go 90mph at 3am that tailgating is no longer a working strategy.
"The central zones of $CITY are self-driving only". You could still use your normal car to get to parking lot near the city and switch to public transport but if you want to get INSIDE the zone then you need to be self-driving.
(On sidenote, you could have self-driving electric taxis that ONLY serve inside the zone as well as zone-only buses)
Then gradually expand the zone.
Will the laws change to have 'autonomous only' lanes? Will they replace HOV lanes? Will autonomous vehicles be allowed to make maneuvers that human driven cars cannot in order to improve traffic flow, such as driving through red lights if determined safe?
I'm really, really hoping the manual drive cars abandoned for autonomous cars are maximally recycled and not just left to rot in myriad junk yards.
Sadly technologies like these have not even point percentage acceptance. Trains and planes are capable of auto driven since ages but we don't actually see them put to use.