Hopefully they won’t suddenly stop once a month while you’re using them to “download upgrades”.
I started working at an Avionics factory in 1966. At that time, the engineers said they had the capacity to produce airplanes that could take off, fly, and land without pilot intervention. The FAA would not permit it for many of the reasons stated here.There was a joke going around in the factory at that time.
The first automatic flight to Paris was announced, and the plane was full.
Back in 66…the physical size of the computer(s) (and custom air-conditioning) to handle that would take up the whole airplane. They were barely into SS computers by then. The IBM 360 was just introduced a couple years earlier. And the computer in you cellphone is 1000 times more powerful that the 360 (which was top of the line Big Mainframe back in 66).
Wow, this is pretty amazing, and settles it for me: http://www.google.com/about/careers/lifeatgoogle/self-driving-car-test-steve-mahan.html
Well, they already are on the road in Calif., so that’s a bet I’d take. So far they’ve been operated with a driver ready to take over in case of a malfunction. The new one is the first to do without the emergency driver. I’m sure this will come about sooner than most of us expect. The logic of driving isn’t very tricky. Driving games have been around now for decades and do a pretty good job and computers tackle much harder tasks all the time. The problem will be more with the lack of uniformity in our roads than anything else. A freeway or any road with striping will be pretty easy, but a car that can only be driven on paved roads won’t be very useful. I’ve been on some challenging unpaved mountain roads I wouldn’t trust a driverless car to manage. Most had to be driven right down the middle to avoid ending up halfway down the mountain.
The problem will be more with the lack of uniformity in our roads than anything else.
No…the most difficult problem is for the computer to interpret what we visually see. That has always been the biggest challenge. The second biggest problem is mechanical failure. Not in the drive system, but with one of the many sensors. Looks like Google car may have a handle on that. But it is a challenge.
The logic of driving isn't very tricky. Driving games have been around now for decades and do a pretty good job
Well, yes and no. Computer driving simulators don’t need to make “Go/no go” decisions, or “continue/no continue” decisions when the weather starts to get hairy. A game is a contrived setup with limited choices: Pac-Man never has to weigh his options, and decide if his interests would be better served by staying out of the maze, breaking down and going to “Pellet-Eater Anonymous” meetings.
The included video shows a new, (presumably) meticuously-maintained Prius driving in great weather on well-defined roadways. Granted, this is the most likely scenario, but weather alone would create enough “non-trivial” go/no go decisions that just pretending the issue doesn’t exist would be unacceptable.
So…not to be argumentative, but to find out if anyone knows…how does the Googlemobile handle such decision-making? Does it leave this task to the human? (In which case, is it really meet the definition of “autuonomous”?) Or does it just assume all is well?
The other big issue I see is reacting to other humans when right-of-way is in question. Many city roads I travel are 1.5 cars wide (when cars are parked on both sides), and have no center line striping. Motorists generally determine ROW in real time, using a variety of means of non-verbal communication. I wonder how a Googlemobile would a) process that info and b) communicate intentions back to another (presumably human) driver?
In line with the above paragraph, I dealt with a phone line crew today, who had one of the two lanes “coned” off, and a flagger directing opposing traffic on the remaining lane. Now, flaggers (generally) have a stop/go sign, but a police officer responding to an INOP traffic light might only have two hands and a whistle. How would Googlemobile handle this? (Heck, could it differentiate between a uniformed police officer directing traffic, and a random nut in a blue shirt playing in traffic? 
(But a more direct issue is that I’d never trust my life to an electronic device, installed on a >10-y.o. vehicle, with a chassis-ground system, subject to “typical” maintenance. I’ve seen too many goofy things happen to current in a “questionable ground” scenario!)
@meanjoe75fan, If your “go/no go” decision is based on the weather conditions, just don’t go until you have a big enough opening that you can go even if the road is wet. Seriously, if your margin of safety is that narrow, you’re taking unnecessary risks. All you have to do to solve this problem is program the car to behave as though the road is always wet. You might have a little extra margin of safety when the roads are dry, but what will that hurt?
There. Problem solved.
“But a more direct issue is that I’d never trust my life to an electronic device, installed on a >10-y.o. vehicle, with a chassis-ground system, subject to ‘typical’ maintenance.”
Who, exactly, decides how well your car gets maintained? Don’t you see to that personally, even if your hire someone to do the work?
If you can afford to trade in your car more often than once a decade, you’re set. If you’re like me, and you like to keep cars longer than that, you already trust electronic devices that old for your safety.
You act as though buying an automated car would mean the loss of control of the car’s maintenance. You would still own it, you’d still be responsible for maintaining it, and you’d still be held liable if neglect led to criminally negligent damage.
Actually, MikeinNH, in 1966 they used analog flight computers, not digital. It’s hard to describe how they worked, but they did. And, they weren’t that big. Amazing stuff.
And, I think the engineers were talking theoretically, like people today talking about men on Mars. I was never clear on that.
Wait a minute. I just recalled something. Isn’t there something in scripture about how self-driving cars will be a sign of the apocalypse?
Thankfully, I’m a card-carrying member of the Church of the SubGenius, the only church that offers a triple money back guarantee if I’m left behind during the rapture.
Actually, MikeinNH, in 1966 they used analog flight computers, not digital. It's hard to describe how they worked, but they did. And, they weren't that big. Amazing stuff.
I was comparing what is currently being done in computer flight control and what is proposed , using the technology of 1966. The system in place now is digital. To replicate that with technology of 1966 would be impossible.
Work on neuromorphic chips is proving to be promising on developing new methods of machine learning. This allows for the ability to learn and drive to occur without giant 1000-processor supercomputers, instead using specialized versions of smart-phone processors.
http://www.technologyreview.com/featuredstory/526506/neuromorphic-chips/
"Those self-driving cars Google is experimenting with might not need your help at all, and more adept Roombas wouldn’t get stuck under your couch. "
I think it’s time for me to bid this thread goodbye. Once posters stoop to the point of mocking religious beliefs, it’s time for the sober-minded to clear out.
It’s ironic that my basic objection was simply one of semantics: driving consists of more tasks than wiggling the steering wheel and stomping on the gas; specifically, it consists of a series of judgment calls that computers execute poorly (or abdicate to the humans entirely…it really hasn’t been established how the Googlemobile does this)…ergo, it cannot be “autonomous” if wetware is still the decider…the car’s a glorified autopilot.
Realisticaly, Whitey, people are risk-averse about $150k investments, as are multi-billion dollar firms. Up to this point, automation of the driving experience has been incremental–throttle-by-wire, ABS, and stability control–and there is no good reason to bet against this trend continuing. Remember that the first firm to go “full Monty” is effectively putting a liability “kick me” sign on their back! Google does not manufacture cars, and that is 99.9999% likely not to change. At “best,” adoption of faux-autonomous cars would be a boutique good, sales limited to a small collection of well-heeled fanboys–a bemusing asterisk to motoring history.
(Who knows, maybe several of us will still be on this site in a decade to comapre notes!)
The only important thing is that you’re disparaging religious belief. While my private beliefs (though nobody’s business) are fairly casual at this point in my life, plenty of decent folks seem to get a lot out of faith. I think your latest post is needlessly harsh to those folks–for no good reason, really–and I urge you to take this to heart and consider revision.
I wasn’t mocking anyone’s religious beliefs and you know it. I was demonstrating that your willingness to assume the worst seemed dogmatic, like a doomsayer standing on the pulpit preaching about the apocalypse. I don’t even know what your religious beliefs are. They’re irrelevant to this discussion, and mocking your religious beliefs without knowing what they are would be impossible.
… and I am a card-carrying SubGenius minister. I can officiate at weddings and everything.
Hi, from my reading Whitey didn’t intend to mock religious belief, but no matter what he meant to do, it wasn’t received well by meanjoe. The thread is still alive, and I think it might be worth a step back to let things sit for a minute before it escalates further.
Cdaquila, I tip my hat to you.
Focusing back on the subject of self-driving cars, one thing I’ve noticed is that most people when they consider the possibility think of a vehicle that contains all of its “brain function” autonomously within itself. And while I absolutely agree that agree that onboard radar and optical sensing systems will be a critical part of future self-driving technologies, it will be combined with “streamed” knowledge. Mercedes is already using satellites, GPS, and terrafirma data technologies to “read” the roads ahead of the S-class to the point of even adjusting the suspension for dips in the road.
I personally believe that truly self-driving vehicles will be reality in the very, very near future. And it’ll use a combination of onboard intelligence and streamed intelligence. The technology is already here. I think the Google car is just the beginning.
IMHO it’s analogous to the internet. The scifi movie that predicted “Hal”, “Space Odyssey 2000”, envisioned one massive computer. Nobody, not even the scifi writers, could have envisioned millions (billions?) of computers all feeding off of one another. I think this self-driving technology may be viable when all the different attributes of the various technologies finally come together.
that onboard radar and optical sensing systems will be a critical part of future self-driving technologies, it will be combined with "streamed" knowledge.
Streamed information from satellites is important, but if you put a google car in traffic today…MOST of it’s driving will have to be autonomous. Now when all the cars are google cars…when all the cars on the road are transmitting information about themselves to every other car on the road…that information becomes very useful and makes the google car much more realistic. The autonomous driving of the google car becomes less and less.
I agree, Mike. But your statement just might be obsolete in another 10 years. The Google car and the Mercedes S-class are only the beginning.
I just learned driverless vehicles are being used in Abu-Dhabi right now. http://www.tripadvisor.com/ShowUserReviews-g294013-d2409120-r137977859-Masdar_City-Abu_Dhabi_Emirate_of_Abu_Dhabi.html