Oh HELL no... no, no, no (auto-driving cars)

[SirBrass]

Originally Posted by Rayme

The funny thing with self driving cars is that they will be able to drive without occupant.

Doubly more "hell no!" than before.

[strat61caster]

I love 'em, was seeing about 1 a day when I was commuting with traffic, now that I take surface streets I only see about 1-2 a month, they are always noticeable because they are literally the only cars on the road allowing for proper following distance, merging procedures and in general going with the flow of traffic while obeying all traffic laws.

Here's an old video about how they programmed the decision making into it, might assuage some fears, there's other parts to the vid but this is the interesting section:

[ame="https://www.youtube.com/watch?v=YXylqtEQ0tk"]How Google's Self-Driving Car Works - YouTube[/ame]


Edit: just realized this will never reach consumer hands, Google or another company will build a taxi fleet and put cab drivers out of business while creating an upper class public transit system of sorts. Solves the insurance issues of private ownership of a vehicle equipped with this tech.

Edit 2: So I don't get attacked, I'm never giving up driving myself, a decent robot taxi service once in awhile would be a nice convenience.

[SirBrass]

I still wouldn't trust it, even if we had auto industry standards for auto-drive that were the equivalent or better than DO-178B, and other associated DO standards in aviation.

Same reason I wouldn't trust an airliner to totally fly itself. The more I see in the industry the less I trust computers and the more human override control I want. Problem here is that operators are so poor and drivers so few, and no real way to correct that.

Though I suppose I could appreciate it more if these allowed "auto drive taxis" to be used by those who never like to drive... as long as it frees up the roads on which I travel and lets me be more spirited b/c of less congested roads. THAT would be an improvement, but only if this kind of software and tech has definite engineering-informed standards for reliability and safety. Otherwise, keep that accident-waiting-to-happen the hell away from me.

[Model Citizen]

[strat61caster]

Originally Posted by SirBrass

Though I suppose I could appreciate it more if these allowed "auto drive taxis" to be used by those who never like to drive... as long as it frees up the roads on which I travel and lets me be more spirited b/c of less congested roads. THAT would be an improvement, but only if this kind of software and tech has definite engineering-informed standards for reliability and safety. Otherwise, keep that accident-waiting-to-happen the hell away from me.

Hah the reduced congestion alone would be worth it, we all know that jams are so often caused by dumb drivers, rubbernecking, tailgating, improper merging etc. all instantly eliminated with computers. Following proper traffic protocols will reduce congestion, reduced demand for gasoline due to an operator that actually understands fuel saving principles will lower gas prices for the enthusiasts.

They don't have web developers working on this, they've got a boatload of engineers, physicists, mathematicians, biologists, and even the soft sciences from all disciplines working on this. After nearly five years and hundreds of thousands of miles later (I'm surprised they haven't claimed a million yet) there has never been a Google Driverless car involved in an accident that a human pilot could have prevented.

http://en.wikipedia.org/wiki/Google_..._car#Incidents

Who would review the standards for reliability and safety? It's a small group who understand all the consequences well enough to determine if something is sufficient or lacking.

[strat61caster]

Oooh I recognize about half these intersections:

Edit: oops I'm a dummy, this video is referenced in the article that stemmed the conversation, still good for those who didn't follow the tunnel.

[ame="https://www.youtube.com/watch?v=dk3oc1Hr62g"]Google Self-Driving Car on City Streets - YouTube[/ame]

[chas3wba0]

Originally Posted by strat61caster

Hah the reduced congestion alone would be worth it, we all know that jams are so often caused by dumb drivers, rubbernecking, tailgating, improper merging etc. all instantly eliminated with computers. Following proper traffic protocols will reduce congestion, reduced demand for gasoline due to an operator that actually understands fuel saving principles will lower gas prices for the enthusiasts.

They don't have web developers working on this, they've got a boatload of engineers, physicists, mathematicians, biologists, and even the soft sciences from all disciplines working on this. After nearly five years and hundreds of thousands of miles later (I'm surprised they haven't claimed a million yet) there has never been a Google Driverless car involved in an accident that a human pilot could have prevented.

http://en.wikipedia.org/wiki/Google_..._car#Incidents

Who would review the standards for reliability and safety? It's a small group who understand all the consequences well enough to determine if something is sufficient or lacking.

No accident YET... of course the Google Car looks like a Godsend. Most people won't wake up to the risks and realities until a Google Car malfunctions and runs over a bunch of students crossing the street or something. Then, it will suddenly be cool to reject this technology.

Just playing devil's advocate here

[strat61caster]

Originally Posted by chas3wba0

No accident YET... of course the Google Car looks like a Godsend. Most people won't wake up to the risks and realities until a Google Car malfunctions and runs over a bunch of students crossing the street or something. Then, it will suddenly be cool to reject this technology.

Just playing devil's advocate here

Like I said, 5 years and hundreds of thousands of miles and counting.



I was incredibly skeptical until I was driving with them, like I said they're noticeable mostly because of how well they behave, "Wow that SUV has proper following dist... oh there's the spinning camera, it's a Google car"

[reni]

It's probably inevitable to become widespread within a couple of decades. This thread sounds like old people complaining about the internet circa 1997.

[chas3wba0]

Originally Posted by strat61caster

Like I said, 5 years and hundreds of thousands of miles and counting.



I was incredibly skeptical until I was driving with them, like I said they're noticeably mostly because of how well they behave, "Wow that SUV has proper following dist... oh there's the spinning camera, it's a Google car"

True, their capacity for self-restraint and discipline will always surpass those of us mere humans :p

However, I still see this as just a fun little project for the kids at Google who are just innovating for the sake of innovating. If they want a fully automated transportation system, it would take a lot less effort (but probably a lot more material) to put everything on rails, no?

[Rampage]

Originally Posted by chas3wba0

If they want a fully automated transportation system, it would take a lot less effort (but probably a lot more material) to put everything on rails, no?

Americans will never buy into mass transit like trains or buses. The government has been trying to push that for year but it is too inconvenient and too structured for us "born free" types. At least, we live in that illusion (or is it delusion?).

[strat61caster]

Originally Posted by chas3wba0

If they want a fully automated transportation system, it would take a lot less effort (but probably a lot more material) to put everything on rails, no?

Bleh responding to this has scrambled my brain.

Too many people thinking in short term benefits, you're right mass transit would eliminate a lot of desire for self driving cars. All these cool programs in the 50's-70's were going to make a brighter tomorrow but ended up getting slashed when people with power realized they weren't going to see many benefits in their lifetime, from experimental motors to college education to NASA to nuclear energy.

I would absolutely push for public transit over self driving cars but self driving cars are the path of least resistance right now, nobody wants to fund public transit out here, BART and MUNI and VTA basically run a skeleton service that only applies to a rather small percentage of the population.

Why would I pay for a train ticket when it costs less and takes less time out of my day to drive to SF or Sacramento (and yes, I considered the complete $/mile on the car I was driving at the time, insurance and maintenance). My commute is 15 minutes by car, 30 minutes by bike or 50 minutes by bus. Which would you choose?

[JohnJuan]

Originally Posted by SirBrass

Actually, it's ability to react to immediate conditions is something I'm alright with. Problem is that there's not really any way to tell if it selected the correct countermeasure to the problem.

Example: a generic Google DriveMe (speculative name invented by me right now) unit which can be used in FF, MR, RR, FR, and AWD vehicles, is driving along at speed, and suddenly traction is lost. Depending on the slide, traffic around, etc., it can select an action... except it can't be specifically tuned for THAT CAR, which a human would instinctively know how to do if he or she was a good driver (note that I don't use "driver" and "operator" interchangably, and "operator" is not a nice term by me in this context). DriveMe does the normal numbskull thing and voila back end steps out further.

OR, it's in a curve and an emergency maneuver is needed and the car is a MR car or a car setup for minor oversteer like ours (a proper car). Throttle lift, brake on... hello snap oversteer.

No way for the computer to know. It only knows based on the aggregate input from all kinds of cars driving in normal traffic and a selection of maneuvers for stopping in a straight line. True this is what is mostly used in emergency freeway maneuvers is it in any sense the only kind of emergency maneuver a driver should be prepared for.

This is why they dont sell it as a module. The better way to go about it would be to only sell it pre-installed on the car it was programmed for.

Originally Posted by SirBrass

And then there's the Asimov speculations. What if the car decided that letting you crash into the wall and die is safest b/c then only YOU die, not others. And what if that code got inserted during an update without your knowledge (Go watch the Doctor Who multi-part episode about the Sontarans and the ATMOS control system in vehicles.... ), oops.

Problems such as this should be heavily mitigated by the vehicle's spatial awareness and the "buffer zone" that it puts between itself and the cars around it. From what others who have seen Google's cars in action, it leaves propper following distance between itself and the car in front.

[SirBrass]

Originally Posted by chas3wba0

No accident YET... of course the Google Car looks like a Godsend. Most people won't wake up to the risks and realities until a Google Car malfunctions and runs over a bunch of students crossing the street or something. Then, it will suddenly be cool to reject this technology.

Just playing devil's advocate here

Exactly. No accident yet b/c of few vehicles that they're carefully developing with. Give one to Joe schmo to own & let him treat it as he does his car normally & we're going to start having issues right quick.

Just like when software goes from a rigorous public beta with studious qa to on sale, there's always bugs that crop up. Except here it'll be bugs induced by hardware issues they didn't see crop up.

Also, I'd like to see SAE & IEEE be fully involved in the protocols for these devices.

On phone I am. Therefore, if mistakes I make, Frank Oz blame you should.