Slashdot Mirror
Slashdot Asks: How Long Before Self-Driving Cars Become Mainstream?
Here's the thing, regardless of one's stand on self-driving cars, they are no longer a futuristic idea. Major car companies such as Tesla, BMW, and Mercedes have already released an autonomous vehicle or plan to release one soon. Sergio Marchionne, an Italian-Canadian executive who is currently the CEO of Fiat Chrysler Automobiles, recently said: It isn't pie in the sky. People are talking about 20 years. I think we will have it in five years. ZDNet has published its interview of Jim McBride, technical leader in Ford's autonomous vehicles team, who thinks self-driving cars are five years away from changing the world. At the same time, we must acknowledge the talks about these smart vehicles killing many jobs, and the security vulnerabilities we read every once in a while. What's your take on this?
Just all those other projects that were only 20 years away, since the 1960's.
Correct me if I'm wrong, but we're not quite at the stage where you can just take a car off the road, slap on some sensors and some jazz in the steering wheel/pedals, and the car is ready to self-drive. I would imagine each car model has to be optimized for this, and that will take awhile. So in 5 years, we might have self-driving cars coming off the assembly lines, but it's gonna take a lot more than 5 years before that's a sizable percentage of the cars on the road.
I think we will have it in five years
Sounds like a good candidate for the osborne effect. https://en.wikipedia.org/wiki/...
Didn't see the benefit until I realized how incredibly useful they'll be at the shopping mall. Same at any busy office building where parking is a problem. Pull up out front, get out, and have the car park itself. I'd buy that.
I pulled it because we had already run it a week ago. A few readers submitted it today (actually, plenty of outlets have run the story today), hence the confusion.
If every car on the road is radaring/lidaring everything within 100m/yards, won't the spectrum just become a clogged mess?
When talking about self-driving cars I often ask questions similar to this
"Will my one-month old son learn to drive? Will my 2 y.o.? My 5 y.o.?"
A few years ago, those questions would get a laugh. Over time they've been given much more consideration.
' Self driving cars still require a human to take over whenever it doesn't know what to do...'
And if you believe that humans can do this after several hours or days of nothing happening, you're wrong.
Great minds think alike; fools seldom differ.
There are no self-driving cars today. A self driving car is a car that can follow any standard road, anywhere. They can't. They are plagued by weird things of all sorts. Maybe we're five years away from ones that can handle the city roads. And another five years from ones that can handle effectively all roads. Then adoption can begin.
Video phones have existed since the '70s. I wouldn't say that they became popular until last year. And they still aren't anywhere near the majority.
I don't want to sit in a car, bored to tears, when I could be spending my time driving; driving's fun, and it's certainly more fun than sitting and waiting.
Funny, I have the same combination on my luggage.
Are we actually trying to keep dupes off the front page now? Wow times they are a changin.
Thanks for letting us know.
Minimum threshold fixed. Thanks!
My feeling is that we're "20 years" away from general, broad deployment of self driving cars, mostly because we don't know how they will work when taken out of the hands of the developers and given to the general population.
There's nothing preventing self-driving cars from working just fine in highly controlled situations - local shuttles on private property, maybe in bus lanes. We'll probably see some form of that first, along with self-driving features augmenting car's control systems (adaptive cruise controlling being one that we already have). I hope we see some of these applications in the next 5 years.
But, it's all the things that never occurred during testing that turn out to happen regularly in a broad deployment that will slow down general adoption. Not just in dealing with road hazards, but also in dealing with how everyone else drives.
My personal anecdote with a Google car in Austin highlights this: I was crossing a major street (Burnet) from a side street and was two cars behind the Google car. The Google car crossed with a green light and I entered the intersection with a green light. The Google car quickly slowed down and almost stopped in the middle of the lane immediately after crossing the intersection, leaving me and the car in front of me stranded in the intersection as the light proceeded to change. I doubt the Google car driver ever realized the hazard they created.
Beyond that, there's the liability issues. Self driving cars will kill people, that's a given. We can argue forever about whether or not their programming and decision making/judgement is better than a human's, but the fact remains that accidents will happen that are the direct result of decisions made by the car's software. Until the legal framework for handling this is worked out, even perfectly functioning self driving cars will have a hard time with broad adoption. The legal system moves slowly and it will likely take 5-15 years for these issues to be worked out to everyone's satisfaction.
So, "20 years" is probably right for some definition of "20 years".
-Chris
OT I have a feeling /. got better recently. Articles got little bit more interesting and strangely the inflow of obscenities from disturbed users (or their bots) decreased slightly. I hope the trend holds on.
> Sergio Marchionne, an Italian-Canadian executive who is currently the CEO
I don't see any relevance at all to mentioning Marchionne's nationalities.
Particularly when there's no mention of nationality for the second person mentioned:
> Jim McBride, technical leader in Ford's autonomous vehicles team
Another aspect of the "self driving car" discussion that I don't hear often is how it will cause inequality to increase. Driving into work I past at least 5 cars that are "for-sale by owner" that are under $2000. When, if ever, will the price point for an autonomous vehicle or even a pure electric now be down to the point where those with very limited income can afford one? I remember when the whole "cash for clunkers" promotion happened early in the Obama administration. That destroyed the market for inexpensive vehicles for several years.
While this may not be a huge issue for those in urban settings, where there is at least an attempt at public transportation, it absolutely puts the freedom of movement that the automobile provided for the lower income population in rural settings. The town I live in is a 25 minute drive to the closest grocery store, and at 6 miles away from the nearest pharmacy, and doesn't even "show up" to services like Uber. Without cars, people would not be able to maintain their lives, and would have to move to urban centers, which would then compound the issues that already exist.
This is what I was thinking. It's an all or nothing situation.
If you car is driving itself you are not going to be paying any attention. You may be expected to. You may be at fault for not doing so. But you will not.
Even if you were paying as much attention to the road as you normally would, after a while your response time may start to lapse. At the very least, your normally unconscious decision making will be thrust into the conscious mind which will cause you to choke and make bad decisions.
My eyes reflect the stars and a smile lights up my face.
In order to kill jobs, you have to have a fully autonomous vehicle which is not 5 years away.
Not so. The needs to replace long-haul truckers are a lot different from the needs to replace New York cabbies. A lot of jobs are of the low-hanging variety that can be easily replaced. Others may take more work, but suggesting that it won't happen until it can cover every specific need is like arguing that the printing press won't replace copying by hand because it can't do calligraphy.
As for concerns about sudden and unexpected obstructions, no amount of smarts, whether human or artificial, will allow a car to avoid hitting, say, a guy who suddenly lands in front of them after jumping off a highway overpass to commit suicide. But an autonomous car doesn't need to be able to do that to be better than us, since we can't do it either. For the stuff that is avoidable, the sensors on many of these cars are already good enough that they detect children running into the street and respond appropriately before a human has even noticed the issue. I've heard about a few cases where passengers wondered why their car was suddenly slowing down, only to realize as they moved forward that the car had apparently seen the feet of someone walking out from behind a parked car and taken the appropriate steps to ensure it didn't hit them.
Which isn't to say that the future is now, but it IS close.
Everyone that keeps saying that the autonomous cars are just around the corner all live in big cities. To get to the point they work without a steering wheel (aka manual mode) these companies have to solve for rural driving. Until the cars can reliably drive up a back woods, rocky, single lane mountain road they are worthless.
70+% of Americans live in cities or suburbs. And they produce almost all the GDP. So it's hardly "worthless" if driverless cars have a problem with places that people rarely need to be.
Humans, by and large, are terrible at operating motor vehicles, and can't be removed from the road soon enoug. Unfortunately, I think it's pretty much a given that the Dunning Kruger effect is going to dominate here and the last people to have the steering wheels pried from their hands will be the worst drivers.
Why can't you anticipate that the man who looks like he's about to jump might actually jump?
In my head, he's jumping from the side of the overpass you can't see from below so that he won't see the car coming, so all you'd see is him falling at the last second. Regardless, feel free to insert your own unavoidable obstruction example in place of that one. ;)
I would still want the manual controls left in place. Just in case I need to go off road.
Minimum threshold fixed. Thanks!
Long-haul truckers own their own trucks. Why wouldn't they be allowed to own their own autonomous trucks?
Ditto. People come up with all of these oddball scenarios (A dozen kids suddenly appear in the middle of highway. Hit them or run into a wall?), but fail to recognize the fact that a typical human driver would have only have looked up from his phone after he felt his car bowling over kids like tenpins.
To be successful a self driving car doesn't need to be perfect. It just needs to avoid killing 30,000 people a year as well as avoiding about 5.5 million auto accidents that in turn injured 2.5 million people.
Humans sort of suck at driving, actually, and I've got to think a vehicle with 360-degree sensors that can see and react to conditions in microseconds can do a lot better than us tired, distracted, drunk, road-raging meatbags.
Any sect, cult, or religion will legislate its creed into law if it acquires the political power to do so.
' Self driving cars still require a human to take over whenever it doesn't know what to do...'
And if you believe that humans can do this after several hours or days of nothing happening, you're wrong.
You are certainly correct that self driving cars cannot assume a driver is being attentive. But they could identify a problem is up ahead and give up control if it can warn the driver perhaps at least 30 seconds before he needs to take control. One example could be construction ahead.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
no amount of smarts, whether human or artificial, will allow a car to avoid hitting, say, a guy who suddenly lands in front of them after jumping off a highway overpass to commit suicide.
Actually, the self-driving car is more likely to avoid an accident in a situation like that because it will react from 700 to 1500 milliseconds faster. At normal highway speed, that means about 100 feet (or 30 meters) of extra braking distance.
You've been doing a good job, manisha.
You are welcome on my lawn.
Pulling it is the worst thing you can do, even if it's a dupe, because *comments*.
If you can't detect dupes before putting them up, don't bother.
At the bottom of the
I have the feeling the real litmus test would be the first time they have a really bad bug, like permanent injury or death where you just have to admit the car was dead wrong. You can compare this with for example the medical industry or industrial accidents, a lot of people have died from human error. But despite all the checklists and routines and safety procedures people recognize that there's always the human factor. With medical equipment and industrial robots on the other hand, there's no tolerance for fatal errors. Which is why there's not so much human-robot interaction really, usually they operated in closed quarters and safety cages and if humans are to go in there they're turned off.
But you can't do that with a car AI and sooner or later it'll run into one kind of crazy nobody taught it how to handle. And improvisation is not their best skill, I suppose it'll hit the brakes and come to a complete stop. If that means standing still as that 50 ton out-of-control semi comes barreling down the hill right at you, well it's technically correct driving I suppose. But it's maybe not the best kind of correct for you.
Live today, because you never know what tomorrow brings
Others may take more work, but suggesting that it won't happen until it can cover every specific need is like arguing that the printing press won't replace copying by hand because it can't do calligraphy.
Best analogy yet.
We play the game with the bravery of being out of range
Seriously this. People are always coming up with rare occurrences where a computer might fail when humans fail at driving in huge numbers across the country on a minutely (is that a word?) basis. Dear lord, I drive in Atlanta traffic and if a self driving car can use a blinker, stay in their lane, and navigate a roundabout it is already leagues ahead of meatbag drivers.
Yes it's an anecdote! Were you expecting original research in a Slashdot comment?
Agreed. I've seen a self-driving car signal, move safely towards the right curb when the bike lane changes from solid to dashed, stop at the red light, and then turn.
Most human drivers in the Bay Area can't even signal, let alone do "advanced" stuff like being in the right place or stopping at a red light. I get a great view of their antics from my perilous place in the bike lane.
or if fails to see child running into the road...
A human is much more likely to do this than a self-driving car. The car doesn't get distracted, and a malfunctioning camera should be detected and result in the car taking itself out of service, since it should have redundant sensors. Those disagree too much, problem.
if you're driving in an extreme weather situation
I'll say that this alone is a reason to, more often than not, to simply stay home. Or even at work if you're there already.
That being said, I won't consider a car 'self driving' until you could take out the steering wheel and I'd still be able to get to work from home. From the sounds of it, they have the most critical part down - not hitting anything. They still need to beef up the routing though.
I don't read AC A human right
If you own a steering wheel free self-driver, I'd imagine that getting a manual vehicle would still be easy. Using an app on the latest device, you decide which vehicle you want to rent. Then, if it's capable of self-driving, you simply step out your door at the appropriate time and it's waiting for you and your camping gear.
If it's not capable of self driving, you load up your stuff, and your car takes you to the manual drive lot, that might actually be just outside of the park where you need manual drive. You move your stuff over, your car parks itself in the spot where the manual car was, and you're good to go. Or maybe it drives itself home for the weekend.
Lots of options.
I don't read AC A human right
e.g. say a plastic bag blows off the bridge in such a way that the car thinks it's a person jumping.
Cars use a multitude of different sensors. A plastic bag is easy distinguished from a "solid" body. E.g. it won't show up on radar.
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
Actually, there's a lot more to congestion than just the amount of cars on the road. Adding lanes is actually counterproductive past a point. There's actually an amazing amount of overlap with fluid dynamics.
Roughly speaking, bigman's point about self driving cars reducing congestion could be true - with fewer accidents, not to mention distracted, stupid, lost, or road-raging drivers, you can go from a turbulant flow to a laminar flow, which can result in a much higher effective capacity.
Though I agree with you on the integrated cities. I've proposed 'semi-arcologies' before - take a 100 story building. First 10 floors are commercial sales, next 10 are business/office, the remaining 80 are residential. Put a skybridge to each neighboring building. I'm tempted to say floor 20 so you can have different elevators serving the residential portion and the commercial portions in the same shaft, saving space. Floor 20 might end up being an interesting design that way, actually. I'm thinking that it might end up being the cafe/quick-stop floor. Anyways, I used to advocate for sliding walkways on this floor and in the skywalks between buildings, but with the spread of small lithium-ion vehicles, we might go with them instead. The core point is to allow a person to cover twice the distance in a given amount of time and/or effort. So if they're willing to walk half a mile, under my rules they'd be willing to travel a mile(horizontal difference). The walking is good for health, you're not using vehicles, and people should be able to get an apartment within a few buildings of their work, maybe even in it, which would really cut down on travel time.
I don't read AC A human right
In winter where I live they don't always plow the roads, almost never down to the pavement in order to save on blade costs. The ruts that remain can turn a car sideways if they are not driven through properly and they rarely match up with the lane. I'm not clear on whether an automated car will actually analyze the ice contour of the road and driving according to that instead of according to where the lane is. Not that the lane matters because the markings are under the ice anyway. The car will have to navigate this while anticipating correctly where the other cars will go as they also deal with the ice.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.