Drinking The Haterade On the Driverless Car
This is something I’ve wanted to write about and challenge ever since I read this recent Bonus Banana article in Forbes about Google’s driverless car: http://www.forbes.com/sites/chunkamui/2013/01/22/f...
Forbes is one of my favorite publications. To be blunt, the slobbering love affair that this writer has with the driverless car is painful to read. His “analysis” is faulty right from the beginning, as he simply accepts Google’s projections on cost and benefit with no challenge. I realize anyone can have an argument about assumptions and you’d expect them to be on the generous side.
But really? A 90% reduction in traffic accidents and fatalities? Ninety percent?
It’s not that I don’t like the idea of a driverless car, or that it can’t be done. It’s a noble and worthwhile pursuit. But as the analysis takes shape based on Google’s truly incredible projections, Forbes is forgetting one of the most important questions—”How?”
How exactly will having a driverless car result in a 90% reduction in accidents and deaths? I get that removing the human element from driving is supposed to cut it down, but how will that work in the interim period between all-human drivers and all-driverless cars? Won’t there be more accidents, not less, as we shift between systems? Automated systems aren’t foolproof by any means, and somebody still has to monitor it and repair it.
“Come on Flesh, it’s Google—it won’t need that much monitoring,” I can hear the fans interject. “It’ll just work.” Really? Can you explain how?
Another extrapolation from the series here is that we won’t have to purchase our own cars anymore. “The car will just come to you,” the article promises. How?
Another claim is that the driverless car will dramatically reduce congestion, eliminating the need for traffic lights. How? Will the car be smaller? Go faster? Be driven by satellite? What about parking? Pedestrians crossing?
And also since the Google car will be able to see in the dark, there won’t be any more need for streetlights. How, and why? Streetlights and their early counterparts, gaslights, weren’t developed with cars in mind—they were installed to reduce crime and enable people to stay out later, so that the workday didn’t have to end with sunset. Just because the car doesn’t need outdoor lighting doesn’t mean we won’t, unless we plan on spending our entire waking lives in them.
Again, I have zero problem with the driverless car—more power to it, and good luck to Google. What I take issue with is a serious business writer melting over this like a schoolgirl at a One Direction concert. He ought to know better. He ought to ask how.
Comments
Well, since 90% of traffic
Well, since 90% of traffic accidents and/or fatalities are driver error related, it stands to reason that when we transform entirely to automated vehicles we'll cut down on about 90% of accidents.
I see your frustration with the author, but much of what he says will probably come to pass. I can't "prove" it, but since Google and others have mentioned it, this is the kind of stuff they've mentioned--self parking, the ability to drive safely at very high speeds, communication between cars that will allow the computers to run the math when coming to intersections, etc. There's actually a pretty cool animation of this online.
How difficult would it be for
How difficult would it be for a hacker to drive your car into the Hudson from his laptop?
"A man generally has two reasons for doing anything. One that sounds good, and the real one." - J.P. Morgan
I read about this. The car
I read about this. The car has traveled thousands of miles across the country and it only had one minor accident. The accident happened when the driver inside the car decided to take over control from the computer and he ended up rear ending another car. Perfect example of driver error.
We have partially autonomous
We have partially autonomous vehicles today. They aren't on an open computer network. We're 20 or 30 years away from them running on an open network. At least at the beginning these vehicles will run on a closed network and a GPS.
But we'd probably be in as much danger of a hacker as we'd be in danger of someone cutting your breaks.
TeddyTheBear: I read about
I read about this. The car has traveled thousands of miles across the country and it only had one minor accident. The accident happened when the driver inside the car decided to take over control from the computer and he ended up rear ending another car. Perfect example of driver error.
For what it's worth, hundreds of thousands of miles. ;)
Congestion would be reduced
Congestion would be reduced because cars would be able to communicate with each other and accelerate in unison. I hit slow-downs in traffic all the time that are not caused by accidents or closures, but by thousands of drivers punching the gas and slamming the brakes just to move up a few feet. All of those inefficiencies build on each other and create a jam.
a lot of congestion is from
a lot of congestion is from idiots hitting the breaks instead of just taking their foot off the gas, drives me mental
I can explain how it will
I can explain how it will reduce accidents and fatalities, actually, because I've been through this once before.
Most of you guys are probably too young to remember, but mid-air collisions used to be a BIG problem with aircraft, and it was almost always (and I'm being generous there - it was always) pilot error. Somebody zigged when they should have zagged.
I was consulting for a small regional airline in 1995 when the FAA made it mandatory that all FAR Part 121 carriers retrofit the TCAS system into their aircraft. TCAS stands for Traffic Collision Avoidance System and it basically takes the pilot out of the picture. When the aircraft senses a nearby aircraft the TCAS kicks in and the two aircrafts' TCAS systems decide among themselves who is going to do what. Mid-air collisions are now a thing of the past.
The Google cars that are on the street presently obviously have some sort of TCAS installed, only two decades more advanced. That's how they keep from running into shit. It doesn't communicate with other cars yet (because there aren't enough driverless cars on the road), so my guess is that it's high-speed RADAR based at this point.
Now, to the point that there will be increased collisions in the transition to everyone going driverless: I can see that happening, but I can't imagine it will be worse than current accident stats. On top of that, once everyone is driverless accidents will pretty much be a thing of the past and your drive time will actually become a lot more predictable and productive (yay! more work!).
Good point, Eddie. We have
Good point, Eddie. We have "autonomous" aircraft and commercial aviation in the United States is nearly accident free. I'm not sure there's been a single death since September 11, 2001 in the U.S. on a commercial passenger airliner.
DCDepository: Good point,
Good point, Eddie. We have "autonomous" aircraft and commercial aviation in the United States is nearly accident free. I'm not sure there's been a single death since September 11, 2001 in the U.S. on a commercial passenger airliner.
American 587 crashed a month after 9/11 and was one of the deadliest crashes in U.S. history. But, of course, that was due to... wait for it... pilot error.
NorthSider: DCDepository: G
Good point, Eddie. We have "autonomous" aircraft and commercial aviation in the United States is nearly accident free. I'm not sure there's been a single death since September 11, 2001 in the U.S. on a commercial passenger airliner.
American 587 crashed a month after 9/11 and was one of the deadliest crashes in U.S. history. But, of course, that was due to... wait for it... pilot error.
Pilot error--good point.
Well, close to the same amount of time. It's been a solid decade since a commercial passenger crash in the U.S.
In The Flesh: How exactly
How exactly will having a driverless car result in a 90% reduction in accidents and deaths? I get that removing the human element from driving is supposed to cut it down, but how will that work in the interim period between all-human drivers and all-driverless cars? Won’t there be more accidents, not less, as we shift between systems? Automated systems aren’t foolproof by any means, and somebody still has to monitor it and repair it.
Automated systems may not be "foolproof", but they are a hell of a lot better than humans driving the car. On the contrary, I find it hard to believe that, in the long run, there wouldn't be a 90% reduction in auto crashes.
“Come on Flesh, it’s Google—it won’t need that much monitoring,” I can hear the fans interject. “It’ll just work.” Really? Can you explain how?
There is great explanation that has been provided by Google on this matter. There's even an entertaining video where a news reporter is asked to walk in front of the car while it's driving to prove that it would swerve and stop. The reporter was scared to death and barely stepped off the sidewalk and, sure enough, car stopped well short of her. I think human beings are just inherently skeptical of the proposition that a machine could do something better than they.
Another extrapolation from the series here is that we won’t have to purchase our own cars anymore. “The car will just come to you,” the article promises. How?
Imagine a fully autonomous taxi fleet that could be queried on demand via an app. Instead of driving a car to work, you simply have enough cars in an area to meet peak demand and people pay to use a more efficient system of transportation. The average car spends 95% of its lifespan parked - that's incredibly inefficient. Imagine increasing that rate by even a few percent and it has tremendous effects on the cost of transportation.
Another claim is that the driverless car will dramatically reduce congestion, eliminating the need for traffic lights. How? Will the car be smaller? Go faster? Be driven by satellite? What about parking? Pedestrians crossing?
The vast, vast, vast majority of traffic is caused by inefficient driving patters. If you're really interested in this subject, I suggest you read a book called Traffic by Tom Vanderbilt, who goes into great depth about how these phenomena occur. But to keep a long story short, most traffic occurs not because of too little road capacity, but because of the accordion effects assocated with unnecessary braking.
When you're driving in closely connected traffic on the highway, human drivers need to give a significant amount of clearance between their front bumper and the rear of the car in front of them. When the car in front of them taps on the brakes, they do as well, and this filters back through all of the cars driving by that point. These create ripples in the system which ultimately lead to traffic. Moreover, human drivers are inexplicably inefficient at managing traffic flows: one highway might be completely packed with traffic while alternate routes are moving smoothly.
With autonomous cars, traffic can travel near bumper-to-bumper, can navigate around congestion, eliminate unnecessary braking, and significantly reduce one of the prevailing causes of traffic: car accidents.
And also since the Google car will be able to see in the dark, there won’t be any more need for streetlights. How, and why? Streetlights and their early counterparts, gaslights, weren’t developed with cars in mind—they were installed to reduce crime and enable people to stay out later, so that the workday didn’t have to end with sunset. Just because the car doesn’t need outdoor lighting doesn’t mean we won’t, unless we plan on spending our entire waking lives in them.
That we would eliminate street lighting admittedly sounds farfetched - not sure I agree with that.
DCDepository: NorthSider:
Good point, Eddie. We have "autonomous" aircraft and commercial aviation in the United States is nearly accident free. I'm not sure there's been a single death since September 11, 2001 in the U.S. on a commercial passenger airliner.
American 587 crashed a month after 9/11 and was one of the deadliest crashes in U.S. history. But, of course, that was due to... wait for it... pilot error.
Pilot error--good point.
Well, close to the same amount of time. It's been a solid decade since a commercial passenger crash in the U.S.
There have definitely been others in the last decade (Comair 5191 in 2006 and Colgan 3407 in 2009 come to mind - both due to pilot error), but your point is definitely true: crashes have been significantly reduced by technology, and the ones that remain are usually a result of
NorthSider: There have
There have definitely been others in the last decade (Comair 5191 in 2006 and Colgan 3407 in 2009 come to mind - both due to pilot error), but your point is definitely true: crashes have been significantly reduced by technology, and the ones that remain are usually a result of
Also Chalk's 101 in 2005 due to metal fatigue. Sorry, I have a strange obsession with plane crash documentaries. And there was a Southwest plane that slid off the runway at Midway due to pilot error.
If you guys have driven in
If you guys have driven in any recent luxury cars like an s class or a8, you'd know they already have much of these technologies in place.
You can merge on the highway, set the cruise, and never touch the break even if you have to come to a complete stop. The car detects another car in front of it and will never collide.
"A man generally has two reasons for doing anything. One that sounds good, and the real one." - J.P. Morgan
BTbanker: If you guys have
A lot of people do certain things to add days to their life. I do things to add life to my days.
DCDepository: But we'd
Minor glitch -> system
SureThing: DCDepository: Bu
Also, just to assuage any
NorthSider: SureThing: DC
73 good sir!
NorthSider: You're missing
SureThing: Oh... and with the
Not worried about Google
I tell you, we are here on Earth to fart around, and don't let anybody tell you different.
Kurt Vonnegut
Not in the US, but there was
mhart1119: Not in the US, but
NorthSider: mhart1119: Not
A lot of people do certain things to add days to their life. I do things to add life to my days.
I hope there's going to be
mhart1119: Not in the US, but
blackrainn: mhart1119: Not
the person behind the wheel
RedWood: the person behind
How do you figure a computer
mhart1119: How do you figure
SureThing: DCDepository: Bu
Speaking from the point of
mhart1119: How do you figure