New Cars With So-called Safety Features. / General Chatting / Forums

Forums

4hv.org :: Forums :: General Chatting

« Previous topic | Next topic »

New Cars With So-called Safety Features.

1 2 3 next

Move Thread

LAN_403

Patrick

Mon Apr 21 2014, 05:40PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

In the US, I see more and more commercials advertising auto executing features such as braking and lane correcting without the human drivers' say so.

My question is: has the legality of all this been sorted out?
What if the drivers become dependent on these features ( because many people are stupid) and then someone gets killed, the driver says "hey the car stopped itself 9 other times just fine, but this time it didnt, and it's not my fault so and so is dead."

of course the human user is always supposed to be in control and making all decisions and therefore responsible for everything the machine does and doesn't do, and for CNC operators and fighter pilots that's fine. But common folks don't have any training or education in machine operating philosophy and decision making or control. So are there unforeseen, legal issues that arise from this well intentioned add on?

And for instance, what if the safety feature kills more people then it saves. Would these features and the autonomy behind them be capable of rejecting a human drivers decision to damage the vehicle by hitting a K rail, saving a child on the road. But refuse the human command, and instead run the child down killing the child?

(After all GM can't even get my ignition or power steering to work.)

HV Enthusiast

Mon Apr 21 2014, 05:50PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Honestly, with all the idiots on the road today, the amount of times these things may work incorrectly probably GREATLY outweights the number of people that would forget to brake altogether because they distracted.

I don't think its a solution that is suppose to work 100% of the time. Its just supposed to reduce x% of accidents overall.

I'm all for automatic lane correcting and braking. I've been driving long enough to see many people get rear ended or side swiped because of distracted, or tired drivers.

However, that said, as a Libertarian, i firmly believe it should be up to the driver to decide if those features should be activated or not, and not just mandated by the government.

Patrick

Mon Apr 21 2014, 06:06PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

EVR, fair point. but does the user have the ability to turn these features off? And are there minimum legal standards and a formal certification process?

And my two test cases, (1) the incompetent dependent driver, and (2) the competent overruled driver. I think is the legal test as to whether these corporations have outsmarted themselves.

There's a guy who is consulted by TV shows, mythBusters, military, science and history channels, but I can't remember his name... I'd like to email him as a humble college student who builds and operates autonomous systems for flying robots.

EVR remember saving 99 people by killing 1 child to gain 6 cents on a stock ticker is a master case for a cowardly ambulance chasing lawyer.

Steve Conner

Mon Apr 21 2014, 06:32PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Google's self-driving cars are allowed on the highway in California provided they have a human in the driving seat to "supervise" them. I think that is the extreme test case and all other driving aids are just subsets of it.

Patrick

Mon Apr 21 2014, 06:37PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

It will be a long time before many Google cars are out there, the cars I mention are in full rate production..

Steve Conner

Mon Apr 21 2014, 06:43PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I agree. My point is, that if I were a lawyer I would argue that the existing cars are just Google cars with less features.

If the Google car isn't allowed on the road without a human driver, it follows that the human must be responsible for the car's actions, otherwise what would be the point of having him there?

Therefore I think the same argument holds for the existing cars, if your collision avoidance computer goes berserk and crashes you into another car, you are responsible for the crash and would have to go to court to plead otherwise.

Dr. Slack

Mon Apr 21 2014, 08:02PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

I think the UK had the right approach to this at the turn of the last century, have the autonomous car optically follow the red flag carried by a pedestrian in front.

It's very difficult to see how guidance automation that's not all or nothing would work. A car with no features will be driver controlled, and as safe as the driver. A car with 100% features already exists, it's called a train, or a bus. You get on it, get the kindle out, and ignore the scenery for the extent of the ride. The driving system is fully qualified, responsible for accidents (they happen on trains and buses, and the law copes with them, 'the train now arriving on platforms 3, 4 5 and 6 is coming in sideways' - it happened at Potters Bar, several dead). A car that behaved like that, we could live with

I sometimes get surprised by my cruise control, and that is really, really simple (you've heard the story about the RV'er who set the cruise and went in the back to make a sandwich). Now if I had some damned software driving instructor nudging my wheel to centre the lane, or easing in on the brake 99 days out of 100, boy am I going to miss it on the day the situation is too complex for the sensors to make sense of.

I learned to fly sailplanes some decades ago, and there was a very clear protocol to go through on dual control teaching aircraft, a 'you have control' 'I have control' exchange so that we knew the hell who was driving. My ears pricked up when helping out at launch point on a very windy and low-level turbulent day to overhear the instructor saying to a very competent but physically small female pupil, 'if you feel me coming in on the controls on takeoff, it's not because I don't trust your ability, I'm concerned about your strength'. I thought sheesh, I am happy that I'm not in that plane. Statement of interest, I have roughly N=100 takeoffs, and N-1 landings.

I suspect that we will not be able to cope with systems grabbing control from us from random time to time until the 'fully autonomous PA' has been developed, and is in the loop. Many people say 'you need a human in the loop', for all sorts of automatic systems from authorizing debit cards to machine-gunning terrorists. However, I'm a bit more optimistic, you need intelligence in the loop. This means that the system will have to be more rational than, for instance, my wife when driving. That is either very easy (have you driven with my wife?), or very very hard indeed (my wife is human, and given the amount of shouting I do when testing any form of AI, it nowhere remotely close to human performance in understanding communication side-channels, invective, rhetoric, abuse, satire, yes, I know I don't know exactly where I am, but we must intersect a major road within the next few miles!)

Carbon_Rod

Mon Apr 21 2014, 10:23PM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

I've had some understanding of the systems since the beginning of the Darpa Grand challenge.
They may gamble with their own lives to go fully autonomous, but there are literally countless edge cases that can occur in field robotics. Indeed, Google's cars have almost killed people on a number of occasions. â€œAlmostâ€ is the part everyone likes to overlook... A tractor-trailer truck blocked the guidance systems edge detector projection, and fooled the LIDAR into thinking it was a canyon wall. Therefore, the vehicle started to head for â€œOpen roadâ€ when the truck changed lanes triggering the obstacle avoidance routine.... trouble was... it was a real canyon wall the truck was obscuring, and GPS does not work properly in such a situation. The driver literally had to wrestle the wheel back, as there was not enough time to hit the kill switch.

Insurance? Yes, fully autonomous platforms have a registered driver for insurance purposes, but they may not be directly liable... So typically people recommend the maximum Third Party Liability coverage available.

Yes, the law will still consider you accountable if you're behind the wheel and someone gets hit.

stop4stuff

Tue Apr 22 2014, 07:00PM

Registered Member #64 Joined: Thu Feb 09 2006, 06:25AM
Location: Southampton, UK
Posts: 68

I'm all for 'safety features' that make driving safer;

ABS is awesome in slippery conditions, however so is letting off the brake pedal a touch and steering the vehicle (in a safe manner) to avoid trouble - but then that is only useful if the really unexpected happens, like a kid running out between parked cars... every other situation when ABS cuts in can be attributed to poor driving skills.

TCS, again only really needed due to poor driving skills.

Just about every 'safety system' applied to vehicles can be attributed to poor driving skills.

I don't want to come across as antagonistic, however the way I see a motor vehicle is as a killing machine when in the wrong hands, and I'm all for safe driving and safety on the roads.

So yep, whatever the situation, the onus is on the operator of the machine not to harm anyone.

As already pointed out with the Google car & canyon example, a 'system' cannot replace a real driver's intuition - by that, I mean, a system cannot judge what another vehicle is going to to by the attitude or placement of the other vehicle, an alert human in control of a vehicle can and always will have the extra edge due to intuition.

Sulaiman

Wed Apr 23 2014, 08:09AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

I expect that there will come a time when additional insurance will be required
if a motor vehicle owner wants to operate the vehicle manually
overriding the much safer automatic systems.

During the transitional period there will be casualties caused by fully automatic systems
and the press will condemn fatalities and injuries 'caused' by the autonomous systems
even though statistically erratic humans perform less reliably.

The main 'problem' will be that even though autonomous vehicles will be
designed, built and maintained by the manufacturer,
manufacturers will not accept legal liability.

1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob