Google Gets Driver(less) License

[SMOGZINN]

My biggest question relates to stuff on the road. That includes oil, or ice/snow.

I am sure the car understands the current road conditions based on the traction of the road just underneath the car, but what about oil, or black ice 50 feet in front of it.

Sure the car can drive in a straight line and measure distances between itself and the car in front of it, but what about the northern neighbors who deal with roads that are temporarily slick or where there has been something spilled on the road?

I am sure that people will still be held 100% accountable for the decisions made by their cars. Which means you will still have to be paying attention in any situation where the car wouldn't notice a problem but a human would.

This might be great in southwestern states, but Boston/New York? I don't think this will fly.

An autocar with a decent sensor package will be much better at detecting changes in road condition then a human would be. Not only does it have all the sensor data from it's camera's, radar, thermometer, hygrometer, accelerometer, and traction, and probably a dozen other instruments I haven't even thought of, but it has all that data from the cars in front of it as well.

And when things do go wrong, and it does not detect the road anomaly in time it will be much, much better at performing the proper action to correct for the sudden change, while most humans do exactly the wrong thing.

 

[SMOGZINN]

I don't see it happening within 25 years at a minimum. We still have train engineers, and the rail system is much more easily controlled than our roadways.
But it is interesting to think about. RV's that drive non-stop to your destination. Tractor-trailers that don't stop for food/sleep. Freight companies are probably drooling at the potential savings in labor costs.

I think (hope) you are wrong about this. I think the time is right for this to happen. People are getting used to automated equipment doing things for them, and I think that a little effort we could sell the public on the idea of autocars.

 

[mnewsham]

I don't see it happening within 25 years at a minimum. We still have train engineers, and the rail system is much more easily controlled than our roadways.
But it is interesting to think about. RV's that drive non-stop to your destination. Tractor-trailers that don't stop for food/sleep. Freight companies are probably drooling at the potential savings in labor costs.

I would honestly say we are already part way there, Higher end BMWs, Mercedes-Benz, etc. Have adaptive cruise control, set a speed it will follow traffic and even come to a complete stop if needed, all without driver intervention, I say 10-15 years before we see the first fully driver-less cars for the public, 20-25 years for widespread adoption.

 

[HumblePie]

My biggest question relates to stuff on the road. That includes oil, or ice/snow.

I am sure the car understands the current road conditions based on the traction of the road just underneath the car, but what about oil, or black ice 50 feet in front of it.

Sure the car can drive in a straight line and measure distances between itself and the car in front of it, but what about the northern neighbors who deal with roads that are temporarily slick or where there has been something spilled on the road?

I am sure that people will still be held 100% accountable for the decisions made by their cars. Which means you will still have to be paying attention in any situation where the car wouldn't notice a problem but a human would.

This might be great in southwestern states, but Boston/New York? I don't think this will fly.

When you learn to comprehend that all decisions made, even human decisions, are based upon If Then Else style statements then you can comprehend how a computer can do everything a human can. The difference is a computer can only do what it is programmed to do and nothing more. That means the more complex the scenario, the more code that needs to be written to account for those variables.

Your scenario is no different that a human for the first time driving on those conditions as well. I know plenty of people from where I live that would have zero idea what black ice is or how to drive on it. They would most certainly get into an accident the first time if no one tells them before hand about what those conditions are and how to handle them. Which is no different than telling a computer except you use code to do the same thing.

 

[Jeff7]

When you learn to comprehend that all decisions made, even human decisions, are based upon If Then Else style statements then you can comprehend how a computer can do everything a human can. The difference is a computer can only do what it is programmed to do and nothing more. That means the more complex the scenario, the more code that needs to be written to account for those variables.

Your scenario is no different that a human for the first time driving on those conditions as well. I know plenty of people from where I live that would have zero idea what black ice is or how to drive on it. They would most certainly get into an accident the first time if no one tells them before hand about what those conditions are and how to handle them. Which is no different than telling a computer except you use code to do the same thing.

A human can't do more than he's programmed to do either - sort of. 🙂
1) We can generate and retain new information on a regular basis, incorporating existing memories to make new ones. Computers can do this.
2) The difference between us and a computer is that we don't (yet) have a debugger that we can plug in and take a complete system snapshot of someone's brain, or step through the code one line at a time. And we simply lack access to the source code. 😉

Kind of like free will: If you don't know what someone's going to do, then it's free will, right? But if you knew the underlying programming and its current status, and you know what stimuli are going to be generated, you should be able to determine the result. Stimulus, response.


Another thought, that would kind of suck: "Oh, you bought a car that can drive itself? Your workday now begins as soon as you leave the house, and ends when you get home. On a related note, god damn I'm glad that you're salaried."

 

[swanysto]

It would be interesting to see how well the cameras perform in bad storms and stuff. Not just seeing ice or anything, but having to see through a Chicago snow/wind storm. Might have to use manual mode then.

 

[Nintendesert]

I am just curious as to how Google is going to use this technology to spy on us and monetize our personal data. Perhaps the self driving car will be spying on us and sending us ads while we sit there doing nothing.

 

[Phoenix86]

If it speeds, who gets the ticket? Points?

 

[Phoenix86]

Also, if it pulls itself up to a valet, does it tip? :hmm:

 

[SMOGZINN]

It would be interesting to see how well the cameras perform in bad storms and stuff. Not just seeing ice or anything, but having to see through a Chicago snow/wind storm. Might have to use manual mode then.

The car would certainly see better then my old eyes! It's camera's can have multiple filters that let it see in a much larger spectrum then I can, ranging up into the UV spectrum and down in the IR. It is not limited to my limited binocular view, it can have triocular, hexocular, or even docecaocular vision, and analyze the minute differences between the views. As well as radar and Lidar that will have no problem cutting though the thickest fog or heaviest rain.
And to top all of it off, it can combine all that with both the sensors on other cars, as well as stationary sensors already in place on the road. Properly done a autocar would appear to be nearly prescient in it's ability to detect and react to changing road conditions. A humans reaction time is glacially slow compared to a computers.

I am just curious as to how Google is going to use this technology to spy on us and monetize our personal data. Perhaps the self driving car will be spying on us and sending us ads while we sit there doing nothing.

They have already hinted at wanting unmanned cars for it's 'steetview' vans. I would imagine that fleets of unmanned cars with extremely advanced sensor packages driving around collecting real time data would be quite useful.

 

[SMOGZINN]

If it speeds, who gets the ticket? Points?

I think that speed limits are something that we could do away with once most cars are automated. The system would determine what the proper speed is based on a number of factors like road conditions and gas economy. If something like a speed limit was in effect and the car broke it, I would think the car owner would get a ticket for failure to maintain the vehicle since they would certainly be programmed not to speed.

Also, if it pulls itself up to a valet, does it tip? :hmm:

I don't think valet would be needed for autocars. They would just park themselves, possibly in large parking lots on the edge of town that are packed bumper to bumper in solid bricks. When you want your car you call it, and a game of shuffle would automatically occur to get your car out, and then it would meet you at the door, or if you knew when you wanted it you could have it programmed to be waiting for you.

 

[SKORPI0]

Watch 'Brave new world with Stephen Hawking - Machines' (Science channel), it shows how the Google car works.... amazing that they've done about 175k miles testing it. :awe:

video - http://www.youtube.com/watch?v=EtdnW_AVycQ

 

[BlahBlahYouToo]

i trust driving next to this thing more than 95% of the driving population.