As they’ve revealed on their blog today, Google has developed a technology for cars to drive themselves. And they haven’t done it on a computer, or in some controlled lab, they’ve been out on California roads testing this out. “Our automated cars, manned by trained operators, just drove from our Mountain View campus to our Santa Monica office and on to Hollywood Boulevard. They’ve driven down Lombard Street, crossed the Golden Gate bridge, navigated the Pacific Coast Highway, and even made it all the way around Lake Tahoe. All in all, our self-driving cars have logged over 140,000 miles. We think this is a first in robotics research,” Google engineer Sebastian Thrun (the brainchild of the project who also heads the Stanford AI lab and co-invented Street View as well) writes.
Further, The New York Times, which has a bit more, says a total of seven cars have driven 1,000 miles without any human intervention (the 140,000 mile number includes occasional human control, apparently). These cars are a modified version of the Toyota Prius — and there is one Audi TT, as well.
So how does this work? The automated cars use video cameras, radar sensors, and a laser range finder to locate everything around them (these are mounted on the roof). And, of course, they use Google’s own maps. But the key?
This is all made possible by Google’s data centers, which can process the enormous amounts of information gathered by our cars when mapping their terrain.
Google says it gathered the best engineers from the DARPA Challenges (an autonomous vehicle race that the government puts on) to work on this project. They also note that these cars never drive around unmanned in the interest of safety. A driver is always on hand to take over in case something goes wrong, and an engineer is always on hand in the car to monitor the software. Google also says they’ve notified local police about the project.
No. For the simple fact it is a prototype. If accidents occur it should be due to human error not mechanical error. And definately not because some scientists decided to use motorists as Guinea pigs.
Jucon said:I wonder when this technology will be released to the public?
Jucon said:But I'd NEVER let it drive itself in harsh weather.
Jucon said:But if it ever got into an accident with someone I expect Google would face huge public criticism.
Well, everything is a prototype before it is a finished product. DARPA has been testing these under controlled conditions for years; at some point, you NEED to test it on actual roads to make sure that it's plausible. Obviously at this point, they aren't good enough to drive unsupervised, which is why they have someone behind the wheel (just like driving instructors typically can control cars when student drivers do something dumb.)
I envision a future where humans are banned from driving because it's too dangerous, since they aren't as capable as computers...and that would be a great thing.
Computers can sometimes mess up,get hacked or something else to cause it to improperly work.
jamesrage said:A good reason to be against self-driving vehicles. I want the ability to be able to drive where ever the hell I want and I want my kids and grandkids to have that same ability.
Look at it this way James, with a self-driving car, you can get as drunk as you want, and let the car drive you home
Stop signs and speed limits are easy enough, RFID transmitters or pre-programmed maps with the pertinent areas mapped out ahead of time will easily suffice.How does the car read stop signs, speed limits, construction zones ect? Today's optical recognition software can't reliably read the hundreds of different signs in different lightning conditions, fog, weird angles ect. How does the car react to someone on a bike giving a hand signal or people jaywalking in an urban area?
Sure, that's why there's a human sitting there who is able to override the system until the bugs get worked out. But eventually that won't be necessary. Let's not forget that humans can also sometimes mess up. Pretty often, in fact.
It does not justify stripping people of their ability to drive.There are millions of people a day who do not die from accidents.Why? For most people, driving is the most dangerous activity that we do. If not for the massive advantages that being able to drive provides us, there is no WAY that we'd be allowed to do something so dangerous that it kills 40,000 people a year.
If robotic cars were able to provide us with those same massive advantages without the danger, why would you be opposed to that? Wanting the freedom to do something that seriously endangers others (while providing no benefit that couldn't be achieved in some other way) is selfish, and is equivalent to wanting the freedom to throw bricks off the overpass just because it looks fun.
Stop signs and speed limits are easy enough, RFID transmitters or pre-programmed maps with the pertinent areas mapped out ahead of time will easily suffice.
Visual sensors could detect the prescience of something in the roadway that was not supposed to be there and make the car react accordingly or we could go the opposite route, embedding sensor networks into our roads to make them able to detect such things and transmit that data to vehicles on the road.
You'd have to build up infrastructure around it, but that shouldnt be too much of a problem. Our transportation infrastructure is constantly being replaced and repaired and it wouldnt be terribly difficult to slip in this new layer. With things like stop signs and signals, RFID transmitters could be added long after the sign was in place by simply tacking it onto the back of the sign.I was wondering how the prototype handles signs, as it doesn't have any infrastructure built to support it.
I agree it's a difficult proposition, but I havent seen indications that it's impossible yet.That might work in some areas, but there are 2 major problems. The first in that city driving in urban areas is beyond the ability on sensors to handle. Safe driving techniques require societal knowledge of how other cars will operate and reading the body language of pedestrians. The second is that maintaining millions of senors is both expensive and inevitable failures will be a problem. Staying limited to highways is much more practical if you want a controlled environment.
How does the car read stop signs, speed limits, construction zones ect? Today's optical recognition software can't reliably read the hundreds of different signs in different lightning conditions, fog, weird angles ect.
rathi said:How does the car react to someone on a bike giving a hand signal or people jaywalking in an urban area?
rathi said:I think that his technology should be used primarily on freeways and highways. Freeways have far less variables than street driving, making it much easier for a computer to handle. I can envision a system where you drive your car onto the on-ramp, punch in your exit, and let the car drive until you reach the on ramp. Even better would be networking all the cars so they work in concert, letting everyone travel at optimal speed and minimizing traffic.
rathi said:Street driving is probably better done by humans who can better handle its chaotic nature.
There is this service where you call someone to pick you up and they drive you where ever you want for money,its called a taxi.
Better a human messing up than a machine. You can sue a human.
jamesrage said:It does not justify stripping people of their ability to drive.There are millions of people a day who do not die from accidents.
jamesrage said:Not even comparable.Tossing bricks off the overpass is a deliberate act to hurt or murder someone. You driving your car to work is just you driving your car to work.
Actually, it can. Object-recognition technology is quite good. It can literally "read" text as long as it isn't in some weird font...and aside from that, it can recognize the shape and design of common road signs anyway.
These are the hardest part, and the reason that it still won't be commercially available for a few years. But the basic principles have already been established: Recognize any potentially dangerous objects (like a human, a car, or a large animal) and avoid running into them. Even here the computers do reasonably well right now...MOST of the time, they'll notice kids running out into traffic and hit the brakes. But obviously, most of the time isn't good enough. They'll get there in a few more years. The technology has already progressed immensely in just the last three years, since the DARPA Urban Challenge.
There is this service where you call someone to pick you up and they drive you where ever you want for money,its called a taxi.
How does the car read stop signs, speed limits, construction zones ect? Today's optical recognition software can't reliably read the hundreds of different signs in different lightning conditions, fog, weird angles ect. How does the car react to someone on a bike giving a hand signal or people jaywalking in an urban area?
I think that his technology should be used primarily on freeways and highways. Freeways have far less variables than street driving, making it much easier for a computer to handle. I can envision a system where you drive your car onto the on-ramp, punch in your exit, and let the car drive until you reach the on ramp. Even better would be networking all the cars so they work in concert, letting everyone travel at optimal speed and minimizing traffic. Street driving is probably better done by humans who can better handle its chaotic nature.
No it can't. Too many uncontrolled variables prevent such a system from working reliably enough to drive on. Graffiti on the sign, odd shadows, weird angles faded paint, reflected sunlight, fog, obscuring branches all are serious obstacles. Even worse, the system has to figure out the difference between writing that is traffic information and other signs along the road. To top it off, this all has to be done in real time. Maybe in the future it could be done, but not with the technology that exists right now.
rathi said:The problem with writing software is that it can only handle variables that have been anticipated and planned for. In an urban environment, that isn't likely ever to be possible. What happens when the car in front of you has engine failure at a traffic light and the only way around is to do something that is normally a bad idea like reversing or driving into the opposite lane? With collision avoidance software you can probably avoid accidents even in strange circumstances, but you still need a human who knows how to drive to navigate your way out.
Kandahar, while I do agree that the technology is exciting, I have to side somewhat with rathi on this.
To my knowledge, there has not yet been a proof-of-concept run that demonstrates we have software capable of successfully navigating city streets without incident or with a minimum of incidents. If we can get that far, then we can look at this idea a little bit more fully. Until then, I dont see that the benefits would outweigh the costs.
Another issue is that the car doesn't actually know precisely where you want to go. It can't figure out which parking space you want, nor can it tell which house the party you are attending is. Manual control is always going to be required unless we develop AI capable of truly understanding what we want.
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?