Article on driverless cars

[TominMO]

Thought this was good background reading on issues with driverless cars.

http://www.foxnews.com/leisure/2014/12/21/are-driverless-cars-safe-california-regulators-weigh-issues/?intcmp=HPBucket

[TorqueEffect]

I don't even need to read it to know driverless cars are not going to happen anytime soon, because there is so many variables in weather, and road conditions that there is no way a computer can get that perfect, at least not yet, also parking lots, unpaved roads, or if you want to go to a festival where you have to park out in a field. Sure there can be manual overrides but that kinda defeats the purpose of a self driving car doesn't it? A computer can't see the truck or car way out ahead swerving around and choose to avoid it like you can. Can they perform emergency lane changes? Nope, only thing they can do is slam on the brakes.

[AlaskaFestivaGuy]

"This is the first fully-automated plane, flown by a computer" ... watch it crash:

https://www.youtube.com/watch?v=bzD4tIvPHwE

This is about the first French Ariane V rocket crashing due to software errors:

http://en.wikipedia.org/wiki/Ariane_5

This is a about a Canadian software-controlled X-ray machine overdosing by factors in excess of 1,000, killing patients:

http://en.wikipedia.org/wiki/Therac-25

[drddan]

I read an article in Autoweek about computer controlled cars. Essentially the car IS a computer, and computers get hacked all the time. (re: N. Korea!)

Basically, the article stated, if someone were to program a virus into a cars computer, and say have all the cars under control of the computer with the virus, they could make all the cars make a left turn at the same time, it could shut down entire cities! There were a few more examples given, but you get the drift!

[TominMO]

Man, what a bunch of Negative Nancys!

Just kidding. I too think driverless cars are a very bad idea, for several reasons, some already mentioned.
1. You would need an artificial-intelligence level of computing to make this happen safely, and an array of sensors. This will also raise the price of driverless cars dramatically.
2. The computers can still crash or be subject to external control--whether by hackers, or police who want to shut your car down. While this may be a good idea to end dangerous police chases, this technology will be used on all of us by the new police state. Right now the "authorities" can jam our cell phones at will, taking away communication. Do we want to let them take away transportation too?!
3. How will the computers be programmed for potentially-fatal situations? Will it "sacrifice" the occupants of your car to protect the occupants of another car, if there is no other choice? These cars will know how many people are in each car, via seat sensors, and be able to communicate with each other. Who decides on the programmed-in "ethics" of the car?
4. Why would we want to give up free choice/personal responsibility in yet another area of our lives anyway?

[AlaskaFestivaGuy]

TominMO,

Regarding cars communicating with one another, the technology has been in place for aircraft for 20+ years:

http://en.wikipedia.org/wiki/Traffic_collision_avoidance_system

The cost would have to come down significantly (certainly to a fraction of the value of the vehicle), and the technology adapted to the 2D world (telling one car to "climb" and another to "descend" won't work!) as well as deal with cars only seconds from collision and the possibility/likelihood of scenarios involving more than two cars (I don't think TCAS can work "three-way" resolutions). For now, simply having everybody stomping on the brakes is probably about as good as it's going to get, then deal with those vehicles getting rear-ended, i.e. chain reactions.

Your comment about whether such a system would "sacrifice" a Festy over a loaded school bus is interesting, even disturbing.

At the end the day, the whole subject should keep the legal profession gainfully employed.

[sketchman]

There was a BMW (IIRC) on TopGear that could not only drive itself but race itself. BUT it had to "learn" the track first, so I doubt any real world app was even close.

Though we do have autonomous cruise control, automatic parking, launch control, stability control. In the more high-end vehicles all this stuff works well. Not all the time in the same car, but the tech is there. I could see at least a full on auto-pilot for highway use real soon. I mean if the car can sense when you're about to crash into someone's rear end and stop you, or slow down while in cruise if you get to close, why can't it sense when it's safe to pass or how close is safe to get to the median? And of course anybody with a GPS knows the car could already easily know most of the speed limits, so what's stopping it, technologically speaking? I'd say not much.

[AlaskaFestivaGuy]

I think all Google is claiming is that their technology will work on hard-surfaced roads with good-condition painted lines AND where Google has previously mapped the road into its big database in the sky. That's a far cry from driving unknown dark country roads without any painted lines. With the computer's "finger on the map" (comparing what it "sees" with the mapping data), any gaps in the painted lines, say at intersections, can just be "coasted" over. Much like a big jet's navigation system uses INS (International Navigation System) always know fairly accurately where it is [independently of GPS, which is a relatively new development], even in the absence of ground-based navaids, such as over oceans. As long as you have the painted line, you can keep "re-centering" yourself, equivalent to updating the INS with "known" positions you fly over vs. simply "dead reckoning."

If the software ever isn't sure of its location, or even which side of the road it's on, all it has to do is stop. That's equivalent to an airplane's autopilot doing a "disconnect" whenever it isn't sure of its ability to keep flying properly, at which time it plays a little "ring tone":

https://www.youtube.com/watch?v=8XxEFFX586k

or

http://www.hark.com/clips/qywlflllfq-a320-autopilot-disconnect

to tell the crew to take over.

[AlaskaFestivaGuy]

From today's San Jose Mercury News (Silicon Valley's major daily newspaper):

http://blog.nwautos.com/2014/12/googles_driverless_car_looks_goofy_and_friendly.html?prmid=4751

It's not exactly clear what "milestone" has been achieved here -- "fully functional?" Yet, it sounds like there still has to be a human driver ready to take over.

[TominMO]

From today's San Jose Mercury News (Silicon Valley's major daily newspaper):

http://blog.nwautos.com/2014/12/googles_driverless_car_looks_goofy_and_friendly.html?prmid=4751

It's not exactly clear what "milestone" has been achieved here -- "fully functional?" Yet, it sounds like there still has to be a human driver ready to take over.

PR nonense.

[Movin]

Sounds like the perfect driver is the computer. The only accidents were caused by errant human drivers.

[TominMO]

UK plan to control private citizens' cars. The lame excuse they give is that it's for fighting global warming and for traffic congestion. Of course it's about control.

http://www.infowars.com/uk-government-announces-plan-to-remotely-control-vehicles/

DEA tracks millions of license plates as cars move around:

http://www.theguardian.com/world/2015/jan/27/millions-of-cars-tracked-across-us-in-massive-real-time-spying-program

[TominMO]

Sounds like the perfect driver is the computer. The only accidents were caused by errant human drivers.

Unless te software is buggy and crashes the program--and the car. Or when hackers take the car over via wi-fi or GPS signals from or to the car.

[TominMO]

Automakers can't protect against hacking of new cars with wireless technology:

http://www.foxnews.com/leisure/2015/02/09/report-automakers-fail-to-fully-protect-against-hacking/?intcmp=features

[drddan]

Moar:

https://autos.yahoo.com/news/consume...--finance.html