Slashdot Mirror
Focusing On Tech Alone, You Miss How Autonomous Driving Will Change Society
Hallie Siegel writes The way that consumers interact with and operate cars will transform most functions in commuting, travel, communications, car ownership, and many other as-yet unknown ways. Dieter Zetsche, chairman of Daimler and head of Mercedes-Benz Cars, said at this year's CES in Las Vegas: "Anyone who focuses solely on the technology has not yet grasped how autonomous driving will change our society." Robotics watcher Frank Tobe writes about how imagination is overtaking the ethics debate around autonomous cars."
Bottom line: we probably cannot imagine all the implications and collateral effects driverless cars will cause beginning early in 2020 for top-end and early adopters and progressively more widespread year after year until mid 2030 when these cars will be our major form of transportation.
That's it? That's your substance? Hell, why not try? Here are my own guesses:
These are all, of course, many years off. But it is starting to look more and more inevitable.
My work here is dung.
Why do we *need* to travel at all? Autonomous transportation in many cases is simply very inefficient teleconferencing. At least this is true in business.
I am very small, utmostly microscopic.
Why do we *need* to travel at all?
Because lots of things have to be done in person. I run a manufacturing plant. I can assure you that you cannot run a manufacturing plant from your bedroom at home. It's a little hard to run a restaurant while telecommuting. Good luck operating a retail store while telecommuting. Farming? Mining? Medicine? Freight delivery? Most jobs aren't really compatible with telecommuting if you actually give it a moment's thought.
Autonomous transportation in many cases is simply very inefficient teleconferencing. At least this is true in business.
I assure you that that is quite false in the majority of cases. Autonomous transportation is basically like a very small flexible train system that does not require tracks. It's like riding the bus - someone else is doing the driving but you still have to get there for a reason.
Not a single word in the article about HOW an autonomous car will change our society in a tangible way. You can safely skip TFA, because it actually says nothing about what the title implies. Instead, the author seems to needlessly hand-wring about the "ethics" of these cars.
These cars won't really deal with ethics, per se. Rather, they'll have goals and rules, and these will essentially encapsulate the ethics in an indirect manner. I'm betting these cars will have reasonably simple priorities for safety, like (I just came up with this on the spot, so don't get hung up on the details):
1) Never knowingly drive the car off the road for any reason.
2) Keep the car in the correct lane unless a collision is unavoidable, otherwise allow emergency lane changes.
3) If necessary, allow movement across the entire roadway, but only if it is otherwise unoccupied and can't cause a collision.
4) If all else fails, slow down or stop and tell the human to make sense of the situation
The trolly-switch dilemma that people keep bringing up is so ridiculously contrived that I just don't see it having a bearing on the reality of day to day driving and safety of the vehicle for a couple of reasons.
First, autonomous cars are much less likely to be surprised by someone cutting in front of them or other obstacles. They don't have blind spots, and their reaction times are many thousands of times faster than a human. As such, the choice of "hit A or B" is much less likely to come up in the first place, because the car would have been following a safe distance behind and would have hit the brakes at the first sign of trouble. So in the vast majority of cases, the car starts braking before the human occupant even realizes there's a problem. No accident at all, or a survivable collision at 10 or 20 mph instead of 70.
Second, in the rare situation an accident is inevitable, the priorities will be straightforward: protect the occupants of the car first within the constraints of keeping the car on the road, and if possible, in it's own lane. That simply means avoiding collisions if possible. If that's not possible, the car will simply attempt to brake as much as possible before the collision to protect the occupants. There will likely be no "swerve to miss the human and hit the bus instead". The car will brake as hard as physically possible, but if it can't safely swerve, it really has no choice but to continue forward in the safest path for its occupants.
I think people are making more of this than is actually necessary by constructing ridiculously overly-complicated and completely hypothetical scenarios and saying "how would an autonomous car deal with this?" Humans are almost never put into a situation where they have to make such a complicated choice in a split second. I'm not sure why we expect our machines to properly make choices that *we* could never make it in real time either. They're going to be better than humans in almost all situations that really matter, such as concentration, navigation, and reaction time in emergencies.
Irony: Agile development has too much intertia to be abandoned now.
Carpooling is a pain because you don't have the car with you during the day. If something unexpected happens and it isn't you driving that day, you are in trouble. With driverless cabs this problem disappears -- you will most likely have to accept a delay when you request an unscheduled cab and possibly a higher price, but you are not stuck.
You forgot the most annoying part about car pooling, you must be on schedule like a clockwork. At work I have to be there certain "core hours" of the day, while mornings and evenings I have a bit of flexibility as long as I get my total hours done. Can't find your keys in the morning? Need to leave an hour early? Work an hour late? Should have stopped to buy milk on the way home? Heck, even those who take the bus can mostly catch one leaving half an hour later. You get the door-to-door service, but it's the least flexible solution. If any of you are the least bit sloppy and unorganized, chances are big they'll either be annoyed with you or you'll be annoyed with them. It's not all of my friends I'd carpool with, to put it that way.
Live today, because you never know what tomorrow brings
The problem with public transport in the burbs is that they only provide timely service during rush hour.
If your shift is different, or if you have to go in early, or leave unexpectedly late, you could be waiting an hour or more for the bus, at a bus stop without a shelter, after an up to 30 minute walk to get to that stop, lugging a briefcase or a laptop bag, in inclement weather.
The reason this happens is because it costs too much to pay the bus drivers to keep the system running all day.
And on top of all that, it has been shown that there is a non-zero chance that you could be sitting next to an ax-crazy murderer.
Public transport really, really sucks in the burbs - it's why we all have cars, even if we don't like to drive them.
There will certainly be some people that will need to have a dedicated vehicle based on the cargo they are carrying, but for the most of us, why have a car at all? Thing about the space savings if you didn't need to park all of those cars downtown and during the day they could drive themselves somewhere else and drive someone with a different schedule, sort of like a driverless Uber, where everyone just shares the cost of the fleet of cars based on usage.
Given it took about two decades for anti-lock braking systems to become widely available on cars (and that is a relatively simple technology compared to autonomous driving systems), most of us will be dead and buried before there is a significant percentage of self-driving cars on the road. First, we'll see "super-cruise control", where the driver can engage a partial system on a highway but the system will revert back to the driver if certain tolerances can't be met (e.g., weather, traffic levels). You'll still need a licensed driver behind the wheel. This will be an expense option on high-end cars for 5-10 years, then will trickle down to more mid-priced vehicles. In the meantime, the auto makers, in an attempt to cope with the shift in liability from driver to manufacturer, will introduce incremental changes to the system to increase those tolerances, but will still insist that the driver take over "just in case" a deer suddenly jumps out on to the highway. All the while the manufacturers will be collecting data on how the system copes with real-world driving to try to determine at what point a truly autonomous car is possible, and their risk in getting sued for a faulty system is acceptable. Given how risk-adverse most corporations are, especially auto makers, I don't see an autonomous car in my future (and I'm 52). Personally, I think you'll get more insight on the future of autonomous cars from the insurance companies than the auto makers.
"Could be worse...could be raining." Igor
The #1 job for men in the United States is.. driving a truck.
It pays well.
Those two things make it ripe for disruption as there is a clear economic incentive; autonomous trucks don't need to stop. It's not clear even if you'd ever have to turn them off, save for regular maintenance. That is a huge economic motivator.
Trucks also follow well defined routes that are easier for the autonomous systems to deal with right now.
The Teamsters will of course freak out; but change, it is a comin'.
..don't panic
This does not solve the problem of pollution when millions of individual cars are manufactured and operated. Nor the impact on environment when habitable land is consumed by sprawling suburbs rather than compact cities. With sensible urban planning, buses and subways can solve the same problem much better.
Self-driving cars can make incremental improvements to safety and pollution levels, but are just delaying the changes achievable with older technology in wide use in many places in the world.
Will we be legislated off the roads as hazards to society? Anyway didn't Google just say that they are nowhere near being able to handle winter conditions. Yet we keep hearing that by 2020 these will be on the roads. tick tick....
I can see things that aren't really cars as well. an autonomous motorcycle with no room for a rider, but with a very large tool chest, so a plumber (for instance) doesn't need his own truck. The drone meets him at his work destination, unlocks and lets him access the tools of his trade.
A similar motorcycle acts as a delivery van. covered with drawers, each of which can lock or unlock independently. It goes to a destination, sends a message to the people inside the building and waits ten minutes. after the person inside authenticates with their cell phone (maybe by taking a picture of the drone) the drone unlocks the one drawer, and waits for the person to remove or add a package.
Make emergency vehicle drones and put them in strategic locations all over the city. call 911 and one of these drones pops out of the police box and could be there long before a human response, Then it could provide SOME assistance while waiting for the rest of the emergency team.
The party drone opens into a full tailgate bar. Flash mob raves.
It is impossible to synch traffic lights in both directions. I would have thought that was obvious. Computers can only do so much.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
What I meant by "super-cruise control" would be that it would handle acceleration AND steering, as long as the car was travelling at a fairly consistent speed with little traffic around it.
There are cars with that now, too. The only thing preventing them from driving the car with your hands off of the wheel is regulation.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"