I've been saying similar to the article for years. AI is limited in what it can consider; we don't have the power to make an unlimited consideration AI, so we must limit what it considers. That means separate roads for them, or roads for them that we can drive on, too, if we have transponders.

 

I've also said for years that the hype couldn't be supported by the engineering. I'm well aware of the software validation and verification that we have to do in the aerospace industry, and these autonomous car efforts aren't even close to the same level of rigor. This has been all about jumping on the bandwagon to hype stock prices. And how about Elon's BS about full self-driving Teslas, except that they aren't...

https://www.caranddriver.com/news/a3...-self-driving/

 

Billy Durant, Preston Tucker and Glenn Davis are smiling...

 

The more things change...

 

I found this a bit surprising. Without the backup human driver, the AI vehicles would have racked up 160% more collisions. Kinda says the software isn't ready to be left on its own.

https://venturebeat.com/2020/10/30/w...over-20-month/

Quote:

Originally Posted by

Waymo’s vehicles were involved in 18 accidents with a pedestrian, cyclist, driver, or other object and experienced 29 disengagements — times human drivers were forced to take control — that likely would have otherwise resulted in an accident.

Waymo claims all of the collisions were the fault of the other drivers, but this is not entirely accurate:

https://www.phoenixnewtimes.com/news...oblem-11541896

Quote:

Originally Posted by

A driverless Waymo vehicle caused a crash in October by stopping unexpectedly in the middle of the road, displaying a technical malfunction the Google-related company claims is rare.

 

I've held off saying this, but my spidey-sense has convinced me that the concept of AI capable of completely self-driving a vehicle is mental masturbation for technogeek types.

I also suspect the fact that a lot of technogeeks are inherently poor drivers may be fueling this quest.

 

I've said all along that there will never be a fully automated personal self driving vehicle unless it's on rails or a closed track. The software and optics systems simply cannot recognize all of the issues that pop up while driving.

 

Another issue on the people side is the death of expertise. Most anyone who frequents this site is a wrencher that understands the mechanical and electrical side of the world. People who have no technical interest or aptitude hold forth on the internet, base their judgment on a google search, then pretend they know as much as anyone and more than you, while the information they just skimmed may have been hype for corporate profit maneuvers like said above. You can't argue with them because they deny the existence of the facts of your argument. I've never seen ignorance wielded like a weapon before.

 

Yeah you have. Clueless people have been crapping on engineers for decades.

 

My question for the people who embrace the idea of self driving cars is where will the liability begin?

Sooner or later, the technology will fail. Either a mechanical issue, software glitch, it’s only a matter of time before there is a collision or fatality. Who will be responsible? The owner of the car? The car manufacturer? The company that developed or licensed the software? All the above?

 

Well....you're not wrong.

 

And not exactly right, but , we'll leave it at that.

 

Just imagine how wonderful it would be on your 20 minute commute to work; thumbing away on your smartphone and texting away with strangers on facebook.

 

And that's different from how people drive today?

 

Great reply !!!!!!

 

only difference is people will then be completely oblivious to what’s going on around them. Now they have to split their concentration on the road AND their phone.

 

I have found that the people that want self driving cars are typically the bad drivers in urban areas. Not only do they want them, they want them on demand, like calling up a taxi at their beck and call, but for as long as they want it. They also do not understand that the country is 95% not like their city-center and this does not need to be yet another thing pushed upon us by them.

This is right up there with the BS about airplanes being safer than cars. Statistically, overall, maybe so, but, when you average too many things together, your data is manure, just like if you lumped all the ingredients in your dinner together; you'd get manure. Flying in an airplane might be safer than a horrible, distracted driver in a car with bad brakes, shot steering, rust, and an ailing engine on a Chicago interstate during the start of rush hour, but not where and how and in what I like to drive.

 

Everything works on paper (or CAD, these days).

 

I don't understand your point. Deaths per passenger mile on commercial airlines in the United States between 2000 and 2010 was about 0.2 deaths per 10 billion passenger-miles. For automobiles over the same period, the rate was 150 deaths per 10 billion vehicle-miles. Unfortunately passenger miles vs vehicle miles aren't exactly the same units, but assume there were six people in every one of those cars (which is grossly optimistic - most probably had one or two) and you're still two orders of magnitude lower death rate in commercial aircraft. Yeah, this is per passenger mile. Aircraft travel a lot faster than automobiles. If you use deaths per hour of exposure, the ratio will be different. The fact remains that you're far less likely to die in a commercial aircraft than in a car.

 

My point is that airplanes may be safer than some doofus driving, but they may not be safer than MY driving.

 

Yeah, keep telling yourself that.

 

The risk associated with flying commercial aircraft can vary a lot with foreign carriers. Some of them I avoid completely. Even Air France, a Delta partner, who you think would know better, still allows cockpit tours in flight by randomly selecting pretty girls from coach. (I have seen it in person.) I had an African experience in which it was obvious that an air charter pilot in Tanzania was not instrument rated yet was flying on autopilot with altitude hold engaged in T storm associated instrument conditions---which is a big no, no. ( I could see the cockpit instruments from my seat and was just proceeding to relieve the pilot from his seat when we popped out in VFR conditions.) In South America I flew in a Boeing 737 with live chickens in the passenger compartment. I saw a stream of oil trailing from one engine and reported it to the captain. After a stop in Nowheresville, he took off anyway because the engine was still making some power. If I recall correctly, this was a Chilean national carrier.

 

See?!?! Get off my lawn, Joe!

 

Surprised they didnt replace you with some more chickens who didnt know the difference

 

Perhaps a change in some posters thinking is in order. Ask yourself what could AI driven cars become. How about dedicated AI lanes for all the bad drivers to get from A to B along with stricter "manual" driving license requirements, so those of us that think we're better drivers will have a license to prove it and open roads to pass everyone by.

Whether you want it or not some form of advanced AI driven cars will take over the roads. I wouldn't expect much real change for 30-50 years. It will take about 25 years to turn over the fleet of cars meaning more advanced AI cars on the roads than our current "manual" cars.

Don't fight the future, mold it to our will.

 

Had a conversation a while back with an acquaintance about this. Apparently some manufacturers have separated driving programming challenges into levels.

Level 5 (if I remember it right) was programming a car that is going to crash to make a decision between hitting a nun in a cross walk or a pregnant lady. Humans might not have the reaction time to make a choice, but if you posit that a machine can react that quickly, which should you program it to choose? That’s a hard problem.

I spent most of my career automating risk decisions which used to be based on race, religion and all sorts of not-really-risk related opinions. I’m not against the machines, but I haven’t seen a demonstration I’d trust my wife, or children, or myself with yet. Then again I haven’t looked.

One way to encourage adoption if we wanted to (as a society) would to be to force DUI convicted and otherwise more dangerous drivers to ride in slow moving automated cars as a safety measure. Since it’s slower, perhaps it’s safer in the early years of driving automation. For people who are safe drivers, If you demonstrate skill & courtesy on the road, you get to self-drive. This way you create an incentive for people to become skillful courteous drivers.

Heavy handed, yeah. Anti-American, probably. But I can think of a lot of drivers I’d rather have riding in a slow moving automated car than cutting off my hard-to-replace ancient cars.

I suspect this will come about a few years after they repeal the 2nd amendment.

 

You hit a good point that was mentioned in one of the articles I linked above. Waymo specifically said they are operating on surface streets and not on highways for that exact reason.

 

While I have never been accused of being clueless. A group of Ford engineers figured out how to shoe horn a 6.0L diesel in a 2004-2007 Econoline van. It is an engineering marvel to be sure, but working on almost any part of it is akin to peeling an onion to it’s core one layer at a time. And yes while working on them I have crapped on the engineers responsible for this atrocity. If you have never worked on one you have no room to comment to the contrary.

 

Maintenance on cars is, increasingly frequently, just designed to be possible, not easy. Jamming more stuff into the same sized bay may mean either remove the engine or go through hell to get there. I'm not sure I can deny once or twice dismissing a user's input during a design phase because of phrases similar to yours, however.