Slashdot Mirror
Autonomous Cars and the Centralization of Driving
New submitter arctother writes: Taxicab Subjects has posted a response to a Morgan Stanley analyst's recent take on how driverless cars will shape society in the future. From the article: [R]eally, 'autonomy' is still not the right word for it. Just as the old-fashioned 'automobile' was never truly 'auto-mobile,' but relied, not only on human drivers, but an entire concrete infrastructure built into cities and smeared across the countryside, so the interconnected 'autonomous vehicles' of the future will be even more dependent on the interconnected systems of which they are part. To see this as 'autonomy' is to miss the deeper reality, which will be control. Which is why the important movement reflected in the chart's up-down continuum is not away from 'Human Drivers' to 'Autonomous' cars, but from a relatively decentralized system (which relies on large numbers of people knowing how to drive) to an increasingly centralized system (relying on the knowledge of a small number of people)."
Just as the old-fashioned 'automobile' was never truly 'auto-mobile,' but relied, not only on human drivers, but an entire concrete infrastructure built into cities and smeared across the countryside
The original "horseless carriages" started out by following the paths their horse-drawn peers used. No special infrastructure just for them.
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
I will never own a self driving car.
1. Too expensive
2. No fun
3. Dangerous
4. No one controls when and where I go
When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
Autonomous cars become mainstream and law enforcement gets a kill switch that locks the person into their car and drives them to the nearest police-approved pull over spot or station?
So basically farmers will be the only ones who know how to drive, and the only ones who know how to use a gun, and they'll have all the power. It'll be like 19th century France all over again.
Sounds reasonable.........~
"First they came for the slanderers and i said nothing."
I don't think anyone has addressed the real question about self driving auto's. Does human's want to give up their control of driving their own vehicle and trust it to basically computers and GPS. Remember the times people follow GPS and end up in a ravine or at a dead end or worse. Now we know human's are dumb about recognizing the obvious sometimes. But do you really think a computer driving a car will be any better if the information it receives is wrong? The feeling I get is that the geeks who embrace self driving cars are infatuated with the very ideal and not addressing the negatives. We have yet to come close to creating systems that cannot be hacked and imagine hundreds of thousands of cars being hacked and taken over. I think we should take baby steps in evaluating how self driving vehicles can be implemented if they even can. Just because we can do something on a very small scale as Google has done. Does not mean it will work on a much grander scale.
Because that's what the teeming masses want. They are fleeing the world of distributed control of computing towards walled gardens, because of the benefits they feel they receive. The transition is not complete yet, but it's strongly along the way, when you compare it with the computing world of say 1990.
And as they are willing to give up control over their computing to a centralized authority, they will be just as willing to give up control over their transportation to a centralized authority. It is inevitable because it is what the majority of the market will prefer. Alternatives will become more expensive, then outright illegal because think of the children.
have you learned nothing from distributed systems? large systems do not need to be centralized to function at skill -- in fact, they work better if they are decentralized and distributed. you may have some control plane for managing emergency events, but even that does not strictly require centralization, and it certainly doesn't carry most of the "data-plane" (ie, automotive) load.
Fixed that or you.
People who come up with this crap usually live in urban areas and have never driven on anything but city streets and urban highways. I somehow don't see the autonomous car getting me up an old mining road in the Colorado Rockies that doesn't show up on any road map. I also don't me see trusting said car to pick it's way around, over and between the various obstacles like wash outs and large lose rocks that take some very careful driving to get over or around. Especially when there's a 1,000 foot drop on one side and a cliff face on the other (Places like the Alpine Loop between Silverton and Lake City or the road to Argentine Pass) to name just two places I've driven.
Cheers,
Dave
Fixed that or you.
People who come up with this crap usually live in urban areas and have never driven on anything but city streets and urban highways. I somehow don't see the autonomous car getting me up an old mining road in the Colorado Rockies that doesn't show up on any road map. I also don't see me trusting said car to pick it's way around, over and between the various obstacles like wash outs and large lose rocks that take some very careful driving to get over or around. Especially when there's a 1,000 foot drop on one side and a cliff face on the other. Routes like the Alpine Loop between Silverton and Lake City or the "road" to Argentine Pass to name just two places I've driven.
Cheers,
Dave
They that can give up essential liberty to obtain a little temporary safety deserve neither safety nor liberty.
Ben
Why do you think that a vehicle that can see in 360 degrees around it in the visible spectrum, infrared spectrum, and LIDAR -- including underneath itself -- and knows exactly where it is within a few millimeters would be worse at navigating between obstacles than you are?
If anything, static obstacles are the easy part. Predicting what crazy human drivers are going to do is hard.
Karma: Terrifying (mostly affected by atrocities you've committed)
I will never own a self driving car.
1. Too expensive
If you consider your time worthless, then yes a self-driving car might be seen as a car + expensive (?, to be determined) self-driving capability.
But suppose you're using that car to drive between workplaces, you can do real work on the way (a la work from home), and your hourly rate is considered $10 worth (or whatever arbitrary number). That's a choice between:
I'm willing to bet that as the technology matures, the latter option will be preferred by many people, real fast.
Note that the added cost of your 'robot driver' can be spread out over the car's lifetime, or at least over the time you spend in the car, driving around. Given the amount of time some people spend in their car every f***ing day, that buys you a lot of robot / electronics / software / whatever. So don't be surprised if that cost dives under your hourly rate - that happens all the time with other jobs that people used to do. Heck, maybe they'll build a robot that'll drive your car the old fashioned way, then doubles to help you with household chores @ the end of the day.
Of course, paid work is just one of useful / valueable / worthwhile things you can do in a car, while getting from A to B.
Fixed that or you.
Fixed that for you. 3rd time's a charm :-)
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
Why do you think that I prefer to be driven than to drive myself? It's an awfully big assumption don't you think?
In the city and on commercial transport corridors there is some scope for this, but a lot of other use cases, people actually enjoy the process of operating a machine. How does your robot car solve that problem?
What I'm curious about is how will different algorithms from different manufacturers all react to each other? Is there a standard set of rules that Google/Tesla/etc are all working together on? i.e. What happens when one mfgr's car does something mfgr B's car doesn't recognize or "agree" to?
Further, if the rules states "minimum distance is 3 car lengths" then how does it know when the brakes have worn out and 3 car lengths is no longer a safe distance? Is the onus on the car or the owner at that point? Will the car just refuse to turn on if it senses that work needs to be done?
What happens when a tire blows and the car goes into a spin? (I've seen this a number of times). Has Google's autonomous car blown a tire at highway speeds?
I didn't assume that, and that has nothing to do with your original assertion that a robot car would be somehow worse than navigating rural roads than you are.
Karma: Terrifying (mostly affected by atrocities you've committed)
Goodbye speed limits, hello 100+ mph!
Simple: it wouldn't be any fun or a challenge. Still no dings in the skid plate.
Cheers,
Dave
They that can give up essential liberty to obtain a little temporary safety deserve neither safety nor liberty.
Ben
As a side note, I just realized you're not even the person I was replying to. So... I'm not sure what your statement had to do with anything I said.
Karma: Terrifying (mostly affected by atrocities you've committed)
Why do you think that a vehicle that can see in 360 degrees around it in the visible spectrum, infrared spectrum, and LIDAR -- including underneath itself -- and knows exactly where it is within a few millimeters would be worse at navigating between obstacles than you are?
If anything, static obstacles are the easy part. Predicting what crazy human drivers are going to do is hard.
I don't think that the autonomous vehicle would be willing to take the changes offroad that a practiced offroad enthusiast is willing to take and has a degree of experience with the ramifications thereof. Heavy offroading requires understanding how the vehicle will react when used other than for its original on-road intent. It means knowing how it'll work in extremely low traction, when wheel(s) are lifted off of the ground, when the ground conditions are constantly changing, and how speed versus braking will affect all of these things. Autonomous vehicles will probably be limited to hard surface paved roads and to low speed driving in parking lots and flat unimproved ground.
Do not look into laser with remaining eye.
Oh, they probably haven't driven on city streets either. With America's crumbling infrastructure. city streets are pothole-ridden messes, with traffic-calming, school buses, bike lanes, and other interesting twists; and unpredictable traffic including cyclists darting between motorized vehicles. Not only is it unpredictable, but it changes day-to-day - with construction, schools being in or out of session, and any of the other obstacles that the city likes to throw in drivers' way.
The people who come up with this crap are those who will benefit from it financially if/when idiot-lawmakers allocate money to it. The real solution is funding planned infrastructure renewal and intelligent traffic planning and control. Butt that is hard, unsexy, actual work that requires real engineers - not political appointees and tenured "seniority" employees.
Predicting what crazy human drivers are going to do is hard.
Really nothing is 'static' to a moving vehicle.
“He’s not deformed, he’s just drunk!”
You mean like in the DARPA grand challenge?
Yeah, Ophir Pass != Autonomous Driving, not even if they install pucker sensors in the seats. I don't even want to think about it trying to react while I am pulling a trailer over Monarch Pass with ever changing weather conditions.
And will I be able to buy that vehicle for $40,000, put almost 200,000 miles on it, carry five occupants, and still drive on the highway at 75 miles per hour?
Do not look into laser with remaining eye.
Why do you think that a vehicle that can see in 360 degrees around it in the visible spectrum, infrared spectrum, and LIDAR -- including underneath itself -- and knows exactly where it is within a few millimeters would be worse at navigating between obstacles than you are?
Because a computer programs while they can gather lots of data cannot interpret what they are seeing.
Who cares if autonomous cars can't take you up an old mining road in the Colorado Rockies. The number of trips along those roads is small enough that the EXISTING set of vehicles will satisfy all demand for many decades EVEN if no more are built.
On the other hand, for the other 99.999% of required commutes autonomous vehicles will do fine.
Because one of the first perks that well off people get is to be driven around in cars by other people.
A lot of people like driving, on some roads, for pleasure, some of the time.
That does not describe the vast majority of required driving in most conditions. I.e. to and from work or the mall.
Again, small enough demand that driving clubs will accommodate it. Just like some people own and ride horses, other people will own and drive cars.
The vast majority of people won't own horses or drive their own cars.
There will never be autonomous cars on anything but highways and maybe city streets. In more rural areas, they will only be autonomous when you enter a highway infrastructure. There's no way any tech company will convince people to give up their ability to drive their own car at least some of the time.
The solution is pretty simple: you'll have to purchase a 'non-autonomous driver's license' to use that particular road. Of course, you won't be allowed to use that non-autonomous car on anything but that road, and it will be mandatory to equip that car with the full suite of government-monitored cameras, GPS and remote override. To keep you safe, citizen.
What about traffic shaping?
Will the rich get home quicker, while the rest of us queue?
If that road doesn't show up on any map you have no business traveling there, citizen. And why are you risking your precious life driving in such dangerous places? You need to be forcefully relocated somewhere safer for your own good.
No reason that autonomous vehicles can't handle most unpaved roads eventually -- after decades of development and a lot of "incidents" -- some amusing, some tragic. And a LOT of lawsuits incidentally. Unpaved rural roads that are well maintained are fairly common in rural areas of the Eastern US. They really aren't much different from urban and suburban surface streets except for more washboarding, more washouts, more livestock in the road, no curbs, and perhaps fewer potholes. Poorly maintained unpaved roads are possibly going to lead to an issue of the car telling the occupants, "You want to continue down this purported 'road' feel free, but you're driving it, not me."
That said, I think fans of autonomous vehicles vastly underestimate the difficulty of navigating anything other than expressways or the variety of unusual and hazardous situations that need to be dealt with maybe once a year or once a decade even on expressways. 99.99% reliable and capable is great. But if the other .01% puts one in the hospital or morgue many folks are going to be a tough sell. Keep in the back of your mind that the automotive industry has yet to master even the comparatively simple problem of designing intelligent braking systems that work worth a damn on ice and snow even after decades of trying.
You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey
While this is one of my favorite discussion topics, TFA is extremely badly written. If someone is not writing clearly and concisely, they are probably not thinking clearly either. Why is this below average blog post getting treated like a serious rebuttal to a report from a major financial instution?
You made a big deal of the technical accuracy of a robot car, but didn't address the biggest reason people actually own cars, the personal freedom and enjoyment of operating them. This is especially important in rural areas where there is freedom to roam, and is precisely the opposite of the features a robot car offers.
You were making sense until the last line. There will be a small demand, but price and practicality will put it in the same ballpark as the horse or Limo. The vast majority of people won't own horses, limos, or their own cars - robot or otherwise.
All they have to do is look where they're going and they're level with the average carbon unit.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Presumably these detect when the occupants are frightened?
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
didn't address the biggest reason people actually own cars, the personal freedom and enjoyment of operating them.
This isn't the biggest reason people actually own cars. People own cars because they need to get to work.
-- Let us endeavor so to live that when we pass even the undertaker shall be sorry. -- M. Twain
"but from a relatively decentralized system (which relies on large numbers of people knowing how to drive) to an increasingly centralized system (relying on the knowledge of a small number of people)"
Yeah a small number of people who know how to drive.
How is this even supposed to be a problem? We buy our cars from a small number of companies, but I don't see anyone complaining about that.
Autonomous vehicles will have to scrupulously obey the law. But society and business depend on the mass NOT obeying the law. We get to work on time because we cheat. How long would it take you to do anything if you could not depend on the timing from route selection and speed? You would not be able to project the future. And huge chunks of your life would disappear into these prison-pod vehicles full of people desperately checking their watches. Also, how about the fact that actually driving the vehicle takes up the time. Do you really want to sit staring out the window for the entire trip? Are we there yet? Are we there yet?
E Proelio Veritas.
look at any crowded metro highway system and you might agree - very few care for the activity of such a mind-numbing and anxiety ridden activity like driving every day of their lives. i drive a 23 mile commute through terrible traffic on a daily basis - it crushes my soul. After the luxury market, you'll see the major metro market commuters buy in a big way. Dont fear the reaper drivable car guys, if you're still buying drivable cars, somebody will make them. and spare me the big picture control society this and that garbage - just a bunch of philosophical bs. people will buy cars that make their lives more enjoyable - and being able to live more safely and without having to mind numbingly control a machine that could kill them for an hour or two a day is definitely an improvement.
Really nothing is 'static' to a moving vehicle.
"Static" in autonomous vehicle terms means that an object is not moving with respect to the world around it: trees, rocks, barriers, and so on. That is as opposed to "dynamic" obstacles that move with respect to the world, like cars and pedestrians.
So you can try to quibble over semantics if you want, but the fact is that the former class of objects are far easier to plan around than the latter.
Karma: Terrifying (mostly affected by atrocities you've committed)
That's a nice theory, but take into consideration that the army has autonomous vehicles right now that drive offroad constantly. Yes, it's complex, but that kind of terrain really is much easier to deal with than human drivers who behave erratically at high speeds.
Karma: Terrifying (mostly affected by atrocities you've committed)
Well, for one data point - I used to use the local bus system (RTD in Denver), until I finally got the money for a car, and then I enperienced the JOY, the pure unadulerated JOY in being to go anywhere I want, without having to wait for a centralized infrastructure to get me where it wants to go. I've tried using the bus system since then (including almost a whole month of light rail before I gave up and went back to driving), but in the end the autonomy I experience is HUGE. And having to ask "central control to please send my car along this complex path between Denver and Colorado Springs, I want to go see what's there.....no, no destination, just want to explore.....oh, never mind!" is not something towards which I look forward.
. Define sqrt(x) as something really evil like (x / rand()), and bury it deep. Watch your coworkers go nuts.
Ill ride in one of these cars when the AI can beat the KM.
Sure, they don't cost $40,000. But neither does the human-driven vehicle, when you add in the 2k+ man-hours * you're going to need for the control system.
*more if you actually take it off-road...
Can you be Even More Awesome?!
There are a thousand universes in that one word "see" that you throw out so casually.
Having a lot of sensors and seeing are not the same thing.
Ultimately no one will be allowed to drive their own car. This will be done for safety reasons of course, but it will be our autonomy and freedom that is further eroded. People's movements will be further restricted and that old mining road will be off limits.
Probably not long after you can get the on-road version.
There are very real commercial applications for OR autonomous driving, and keep in mind that this was 10 years ago and those were self-funded (or by whatever sponsors they could round up) university teams doing one-off vehicles. If you look at what they achieved for what they spent, and extrapolate it to mass production it's very reasonable to expect off-road autonomy to be available on the same time scale as on-road.
And as zippthorne notes, you have to put a ton of time in to get any good at driving off road. With the autonomous version you pay some money and get tens of thousands of hours of experience built into the system at delivery.
Driving a high-speed vehicle is too complex for the average human being.
I think in a relatively near future - 25 years or so - manually driven cars will be regarded much in the same way that we regard horses today. They'll still be around, albeit in smaller numbers. There'll be lots of people who love them and keep them as a hobby. There will be special trails were you can go ride them. There will be enthusiast meetings and the like. But no one will use them as transportation to go back and forth to work every day.
And will I be able to buy that vehicle for $40,000, put almost 200,000 miles on it, carry five occupants, and still drive on the highway at 75 miles per hour?
Please don't move the goalposts - he answered your original concern fairly succinctly. If he's anything like most nerds I know he won't appreciate having to puff and run to keep up! :-)
..Mullah or Pope, Preacher or Poet, who was it wrote: "Give any one species too much rope and they'll fuck it up"?
This is the exact difference between the Weak AI and Strong AI approaches. Weak AI needs complex specialized infrastructure and dependency on centralised systems. Strong AI type machines can directly replace human drivers in individual cars or trucks with no modification to the road system. Strong AI machines could even include human body robots to add additional functions like loading or unloading. Of course such robots will probably cost at least $100,000 each (or more), and even the most basic Strong AI system for a car will cost at least $20,000 to $40,000. Of course Strong AI is at least still 10 years away and certification for driving probably 20 years away. :D )
(I first made those estimates (or similar) about 20 years ago and they are still true today.
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
Why do you think you wouldn't be able to go where you want, when you want Wroth am autonomous car? That's the primary reason to use one instead of mass transit.
Karma: Terrifying (mostly affected by atrocities you've committed)
Fixed that or you.
People who come up with this crap usually live in urban areas and have never driven on anything but city streets and urban highways. I somehow don't see the autonomous car getting me up an old mining road in the Colorado Rockies that doesn't show up on any road map. I also don't see me trusting said car to pick it's way around, over and between the various obstacles like wash outs and large lose rocks that take some very careful driving to get over or around. Especially when there's a 1,000 foot drop on one side and a cliff face on the other. Routes like the Alpine Loop between Silverton and Lake City or the "road" to Argentine Pass to name just two places I've driven.
Cheers,
Dave
Whoa, hold on, that's getting way to advanced for autonomous cars. They're going to suck in city traffic too.
What the proponents of autonomous cars often ignore is the fact that they'll all be using the same navigation data, so that means they're all going to pick the same route without manual human intervention. Anyone who drives in a city who has half a brain knows that sometimes a longer route gets you where you want to go faster because it avoids congestion.
Calling someone a "hater" only means you can not rationally rebut their argument.
Sounds like you've got zero experience using these technologies in the real world.
The first problem you have is that these technologies aren't as good as you think. Rain and snow tends to have a very negative effect on the LIDAR, IR and Visible parts of the electromagnetic spectrum (which are actually quite close to each other). There's some very good reasons Google is testing their cars in sunny, clement California.
The second problem you have is that these systems have a lag when decisions need to be made in real time. I've dealt with LIDAR terrain mapping, It takes hours for an analyst to get a good capture into a usable state. A capture that has a lot of cloud (most of the imagery I deal with is aerial) will take a lot longer. Sure you dont need that level of precision, but you're still going to end up with the computer being presented with incomplete data.
The third problem you have is that computers aren't predictive, humans are. The computer reacts to what has just happened. A human looks at the risks and evaluates them before they happen. Computers are reactive.
Something that always gets pointed out in favour of autonomous cars is the fact that they're predictable... Which is actually a bad thing when you're putting them into inherently unpredictable situations. So if a computer encounters a problem, it's designed to stop, that is predictable. When you're driving up a mountain road or even in moderate traffic, suddenly freezing is a very bad thing. Humans may not be as predictable but at least they have a chance of being able to make the right decision in an complex and unpredictable emergency, we can guarantee current software cant precisely because it's so predictable.
Please note, this isn't an argument against autonomous cars. It's an argument against the misconceptions people have about autonomous vehicles. We haven't even come close to replacing human controlled vehicles in areas that can be controlled to a pathological level like mine sites (and I've been hearing about the autonomous mine site since 1994) because the technology isn't as reliable as a human even after decades of development... And here's a news flash, it will be decades before you see fully autonomous vehicles on the road. We've come a long way, but there is still a long way to go before they're usable.
Your next car wont be autonomous, your next, next, next car wont be autonomous (and before you say that Mercedes has said... remember how many concept cars the industry puts out that never make it into production).
Calling someone a "hater" only means you can not rationally rebut their argument.
Why do you think that I prefer to be driven than to drive myself?
I am not the one you replied to but I don't believe you will want to be driven. I just want the idiots who can't be arsed to drive with their eyes on the road instead of on their phones to be driven. I also assume you don't fall in that category because those people don't seem to enjoy driving. They probably just want to get from A to B. Exactly what a self driving car would provide.
If I had a car I would also want it to be self-driving because I find driving to be boring. I do, however, realize it is a 1 ton weapon with significant kinetic energy.
Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
Traffic data is included in modern navigation systems. Last modern car I was in just asked me something in the lines of "Traffic congestion ahead. Reroute?".
Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
Well, no. They don't seem to. They're talking about autonomous vehicles And there is at least one far enough along for photo shoots. http://rt.com/usa/driverless-a... But it's often a long way from capability demonstration to proven capability. Not to mention that there may be some significant differences between the appropriate method for an autonomous APC to deal with a couple of cows in the road and the same situation in a Fiat Panda.
You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey
Also, what are they going to do, outlaw motorcycles and scooters? Never happen either.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
The tech companies won't have to, the insurance companies will do the job for them. Human error causes most accidents, therefore they can tripple your insurance if you drive the car. Never mind that accidents will diminish and insurance payouts will plummet with automatic cars, they will still need to drive up your insurance rates.
Yeah, I'm not buying that argument. Everyone makes it sound like it will be this wave type event. But I'm pretty confident that autonomous cars will take decades to attain any sort of market saturation to influence insurance rates in that fashion. All you have to do is look at the history of transportation to see this. But's that's not what I'm talking about. People will not give up their ability to drive willingly. Not in the city and certainly not in the rural areas. Even with supposed higher insurance rates, people will still want to control their own cars outside of the highway commute scenario. IMO, I see this as a Silicon Valley bubble. Inside the bubble there is an animosity towards commuting and driving in general. It keeps them from seeing that a majority of the country actually enjoy driving. I live here and I see it every day. There is huge desire here to eliminate drivers because....well drivers here SUUUUUUCK.
How did people get to work before cars?
Because they don't exist? And when they do you can be sure as hell you won't be able to go anywhere you like. As a real world example I was on holiday this Easter and discovered (you won't get that on a preplanned automated journey) some sand dunes which I could take my 4WD on. We spent about an hour hooning around having fun in the sandy wilderness, in fact it was the highlight of the drive. I can't see you having that experience in your robot car.