Oops...self-driving cars perhaps not quite ready for prime time yet.

[Harry P.]

If a car manufacturer offers a truly autonomous car that needs no driver input... does the owner of the car still have to buy auto insurance? I mean, he is no longer a "driver," just a passenger in his car.

I wonder what the auto insurance companies think about that?

[Ace-Garageguy]

If a car manufacturer offers a truly autonomous car that needs no driver input... does the owner of the car still have to buy auto insurance? I mean, he is no longer a "driver," just a passenger in his car.

I wonder what the auto insurance companies think about that?

This is the kind of question that makes me wish I'd picked a different career-path.

Being a layer in idiot times like these must be an absolute guarantee of richly remunerative employment forever.

[Ace-Garageguy]

                

Edited July 2, 2016 by Ace-Garageguy

[Tom Geiger]

 Yup. One of my requisites for female companionship for the past several years has been "must be able to drive a three-on-the-tree". Needless to say, I'm a lonely guy. 

Both my wife and my girlfriend prior to her learned to drive on three-on-the-tree cars!   Our first family car (wife was the primary driver) was a Nissan Stanza with a five speed. 

I also taught my sister, my wife's sister and her brother.  One Saturday morning I get woken up by a phone call from my wife's friend and her husband.  They had bought a brand new Mercury Topaz at a local car show. They went to pick it up at the dealership and didn't realize they had bought a car with a clutch. So I had to go to the dealer and drive it home for them. Then I spent the day teaching them to drive it.  And the husband got all mad that his wife was getting it and he wasn't!  That was a fun day.

 

[Joe Handley]

If a car manufacturer offers a truly autonomous car that needs no driver input... does the owner of the car still have to buy auto insurance? I mean, he is no longer a "driver," just a passenger in his car.

I wonder what the auto insurance companies think about that?

I'll bet that if everything gets straightened out tech wise, they'll prever autonomous over human driven, since humans will make more mistakes and will probably either lobby the government to make only self driving cars legal and/or crank the insurance through the roof on the stuff that isn't.  That said, I can see an advantage to having some mix of self driving and enthusiast driven cars sharing the roads.  Put those who would rather have a rolling living room, drunks, and those that consider driving a wasteful chore in the self drivers that will stay in the outer lanes then allow those of us who actually prefer driving to have the open lanes without tue sudden surprise of somebody randomly changing lanes in front of you, deciding that they want to run 10 below the limit on a day like we're having today, ect.

Good question. The bozo who drove into my old truck a couple weeks past is now refusing to pay for the damage, through his attorney, on the grounds that "just because Mr. Zippy was given a police citation for 'improper lane change' doesn't automatically imply responsibility for causing the collision, much less financial responsibility for repairing any damage".

OK.   

Uhg, I hate people like that, although at least he stuck around after the impact, mine usually up and take off.

[Ace-Garageguy]

 ...And the husband got all mad that his wife was getting it and he wasn't! 

Gives a whole new meaning to "male performance anxiety", doesn't it? 

[Ace-Garageguy]

I'll bet that if everything gets straightened out tech wise, they'll prever autonomous over human driven, since humans will make more mistakes and will probably either lobby the government to make only self driving cars legal and/or crank the insurance through the roof on the stuff that isn't.  That said, I can see an advantage to having some mix of self driving and enthusiast driven cars sharing the roads.  Put those who would rather have a rolling living room, drunks, and those that consider driving a wasteful chore in the self drivers that will stay in the outer lanes then allow those of us who actually prefer driving to have the open lanes without tue sudden surprise of somebody randomly changing lanes in front of you, deciding that they want to run 10 below the limit on a day like we're having today, ect.

The state of "autonomous" vehicle technology is another one of those things where the popular perception is far from the reality.

A substantial number of engineers and scientists actually working on the integration of strong AI, machine vision and all the various and REAL technical obstacles understand that the perfection of the tech may be still 20 years out.

I personally don't want to trust my life to the beta version.

Machines do a LOT of things far far better than humans, but driving is going to take some really smart machines. No doubt the day will come, but multiple redundancy in case of failure (and fail-safe systems), as well as hardening of the systems against malicious hacking are far from fully developed now...or even really understood in the context of what it's going to take.

 

Edited July 2, 2016 by Ace-Garageguy

[GT4494]

I don't think we will ever see totally autonomous cars.  Every city in the country would go broke if there were no traffic violations.  A computer operated car by definition would obey the rules that are programed in it.  If the program code was set to violate the law who would be at fault.  couldn't be the " rider".  Therefore the car would "Never"  violate a speed limit, red light, stop sign etc..  

[GT4494]

and as to the Manual shifting.

Learned to drive on a pickup with three on the tree.  now have a car with a 5 speed and a car with a 3 speed non synchronized.  Its fun double clutching too. 

[Ace-Garageguy]

Consider this: in a future world where all or even most of the cars are "autonomous", there will still be failures. Even the MOST robust system experiences occasional failures, and ANY failure of a vehicle transporting living human beings will HAVE to be considered "statistically significant" from a moral and legal perspective.

Assume the car you're riding in has a massive system failure of some sort and the redundancies built in fail as well, possibly due to inept maintenance. If the car is in a tightly packed traffic stream, like YOU encounter every time you go on an interstate, the vehicle can't just shut down and coast safely to a stop. Nope, it's going to have to be able to communicate with the vehicles around it and 'ask' for permission to invade their space while it tries to pilot itself over to the side of the road...EXACTLY what YOU do if you blow a tire or experience some other kind of malfunction. All the OTHER vehicles will be responsible for reacting safely to the failure of the first one too.

Getting a machine to be able to multi-task in a situation like this and correctly weigh all the possible variables to take life-saving action is a pretty good trick. Most humans can barely do it, and not always.

Think about how YOU'D react at 70 MPH if you blew a tire in the center lane of 5, or if the car next to you or just ahead of you did.

Get it?

Edited July 2, 2016 by Ace-Garageguy

[Harry P.]

I agree that there would be a massive number of issues to be solved if we ever went to a fully autonomous system of transportation. It seems inconceivable that we could ever develop and implement such a system.

But then again, back in the day people believed that if a human being traveled at any speed greater than that of a galloping horse, the body could not stand the stress and you would die.

And of course, once upon a time the idea of a human being actually setting foot on the moon was laughable.

So you never know what might develop...

[Ace-Garageguy]

I agree that there would be a massive number of issues to be solved if we ever went to a fully autonomous system of transportation. It seems inconceivable that we could ever develop and implement such a system.

But then again, back in the day people believed that if a human being traveled at any speed greater than that of a galloping horse, the body could not stand the stress and you would die.

And of course, once upon a time the idea of a human being actually setting foot on the moon was laughable.

So you never know what might develop...

I never say "never", but it won't be next week or next year or probably even next decade.

Speaking of "never", it's beginning to look more and more likely that faster-than-light space travel as envisioned by Star Trek's "warp" drive (and others) is going to be possible after all, and that the upper limit of light-speed for moving between two points (implied by Einstein's work) may in fact be neatly side-stepped.

"So you never know what might develop... "  Indeed. 

[Spex84]

And also, there are "autonomous" bum wipers, Ace...don't they use water jets in Japan? Skipped the whole robot arm thing altogether! lol

 

 

[Ace-Garageguy]

And also, there are "autonomous" bum wipers, Ace...don't they use water jets in Japan? Skipped the whole robot arm thing altogether! lol

You mean bidets? They've been in use in Europe since the 1700s. Not really "autonomous", and you still need to dry the water...far as I know.   

[Spex84]

*Checks Wikipedia* Huh, looks like despite the fancy features, those receptacles technically ARE bidets. Just...a little more advanced than they were in the 1700s.

[Ace-Garageguy]

Boy...that's quite a lot of gadgetry and complication just to clean your backside. Looks like it even has a blower. And an LCD screen readout too.

Man. That's some high-tech carp. 

Ain't technology grand?  (Does it come with an emergency roll in case of power failure?)

[Tom Geiger]

Consider this: in a future world where all or even most of the cars are "autonomous", there will still be failures. Even the MOST robust system experiences occasional failures, and ANY failure of a vehicle transporting living human beings will HAVE to be considered "statistically significant" from a moral and legal perspective.

Assume the car you're riding in has a massive system failure of some sort and the redundancies built in fail as well, possibly due to inept maintenance. If the car is in a tightly packed traffic stream, like YOU encounter every time you go on an interstate, the vehicle can't just shut down and coast safely to a stop. Nope, it's going to have to be able to communicate with the vehicles around it and 'ask' for permission to invade their space while it tries to pilot itself over to the side of the road...EXACTLY what YOU do if you blow a tire or experience some other kind of malfunction. All the OTHER vehicles will be responsible for reacting safely to the failure of the first one too.

The past few years I've been working in Reliability Engineering for pharmaceutical manufacturing.  It's really an interesting field of study, all very logical based on predictive maintenance, proper preventative maintenance and  active monitors on the equipment that alarms on any out of tolerance activities.  We are piloting a system called "Machine Learning" where all the characteristics of a single machine or a production line are constantly monitored.  Through experience, the system will report the very first out of tolerance message because it's programmed to recognize bad behavior.  This is all recorded in a central system that is monitored by a live operator every shift.  Once an issue is discovered, the operator has the ability to shut down immediately (in the case that the deviation will result in bad product that will be rejected) or if the machine will make it through the run and can be repaired on scheduled downtime.  All of this data is kept and is the basis of the system recognizing bad behavior.  When there is a production run that fails quality control, this data is pulled up and studied to determine the cause of the failure.  Then a root cause will be determined... was it machine failure or human error?   If indeed an operator does the run setup wrong, the system immediately determines that and refuses to start.   These type of systems are real and all around you.  For instance the Flavia coffee machine in my office will refuse to run if your cup isn't properly positioned. 

So I believe the cars of the future will have these onboard systems.  The main brain computer will probably have a redundant unit in case the main one suddenly fails.  It will be monitoring all systems and will report vibration and rotational issues such as a bad wheel bearing. Blow outs are caused by out of tolerance tires. The system will predict that blow out rather than have to react to it. Also, today's Teslas have a motor on two or four wheels, so if one does fail you can get home on the other(s).

If indeed that blow out does occur, the car will immediately put it's flashers on and send out a mayday signal.  The other smart cars around it will react accordingly.  And in the case of human controlled cars, I've witnessed blowouts and gave the guy plenty of room!  But in the case of an unpredicted blow out, the vehicle will then isolate the recorded behavior of that tire prior to the event, and it will send that report back to the factory.  By the next morning every like vehicle on the road will be aware to look for that event pattern to prevent a blow out.

Cars of the future will not tolerate inept maintenance. They will warn you there is an issue, and tell you to take the car for maintenance.  There will be a cut off point where it simply will refuse to move. And in the case of self driving cars, it will make it's own appointment for service and go to the dealer at the correct time!

 

[Joe Handley]

I believe that Teslas with the Autopilot systems will already have many of those features Tom, they've even been using info learned by the individual Teslas to teach each other how to do things better, this issue could cause Tesla to swap the optical sensors for more or a radar type system that the newer cars have.  Still won't fix the whole "Watch the freakin' road, ya dummy!!!!!" Problem though.

[GT4494]

To me the question is  "What happens in the transitional phase"? 

New autonomous cars are developed and put on the road.  People buy them and use them.  Something happens and they have to "Borrow" a car for daily use.  The borrowed car has no self drive and the driver gets a phone call and forgets that minor fact..  A very likely  scenario  in the new generation.

Could make for interesting times. 

But I'm still waiting on my "Jetsonmobile" folding car.    

[Joe Handley]

I think Tesla provides another Tesla as loaner, so that shouldn't be an issue, and if they are like me and have one "modern" car and one "dinosaur", it shouldn't take much to readapt to the older car.......unless it's a column shift auto, that always catches me out in my Parent's Town and Country..........too many years of driving console shift cars and Jeeps.

[Matt Bacon]

I'm sure there were Ike Asimov stories where people were prosecuted for driving cars on "manual"...

Surely the only way this works (for at least a few decades) is if the human "driver" is always responsible for the behaviour of "their" car? It's gonna be a while before cars are smarter than horses, given that the current state of AI is about at ant level, if that...

The biggest factor is going to be communication, not intelligence. As Tom says, it's the ability of one vehicle to tell all the others around it if there's a problem developing. And the ability of a "cloud" computer somewhere to learn from the behaviour of all cars in the same place over months and years.

To take a specific example: cruise control. In the 50s/60s, you had a cruise control that held a specific speed. By the 80s you had one that had a "default" speed that it returned to even when you speeded up or slowed down with the pedals. In the 90s my Mondeo accelerated when you pressed the gas pedal, and slowed down when you braked, and held whatever speed you ended up at. Nowadays, they use the front parking sensors (and maybe radar) to lock onto the speed of the car in front and track it. Tomorrow they'll get a feed of "I'm thinking this...." from the cars in front, and those behind, and adapt their driving style accordingly. It's still a smart robot, though...

...when they start to think "I think this guy's a bit timid, I'll overtake him, but I'm not going to be tempted into a drag race with this idiot over here..." THAT'LL be AI...

bestets,

M.

[Longbox55]

I don't think we will ever see totally autonomous cars.  Every city in the country would go broke if there were no traffic violations.  A computer operated car by definition would obey the rules that are programed in it.  If the program code was set to violate the law who would be at fault.  couldn't be the " rider".  Therefore the car would "Never"  violate a speed limit, red light, stop sign etc..  

After doing a little research on this accident, it seems that the system in this can in fact be set to violate the speed limits. According at least one witness, the Tesla was going at least 25 mph over the posted speed limit for that road (60 MPH), which was the speed she was going  (she admitted to speeding herself) when he passed hew about a mile or so before the accident.

[Ace-Garageguy]

The past few years I've been working in Reliability Engineering for pharmaceutical manufacturing.  It's really an interesting field of study, all very logical based on predictive maintenance, proper preventative maintenance and  active monitors on the equipment that alarms on any out of tolerance activities.  We are piloting a system called "Machine Learning" ...etc.

All well and good, Tom, and it all sounds lovely.

One little fact is that I'm more cognizant of the mechanical aspects of machine failure modes, particularly surface vehicles (and ALL of their components), and their "predictability" than probably anyone on this board, and I'm also much more well versed in the realities of Artificial Intelligence and "machine learning" than you could possibly imagine. No brag, just stating fact.

It's going to be rather harder to accomplish this on a large scale than the VAST majority of people seem to realize.

I'm fully aware of all of the possibilities for failure and mis-communication between various "intelligent" machines in our coming future, and I'm also fully aware of the massive amounts of code and prediction of machine behavior by humans that will be necessary to turn everything INTO code before the fact.

The appalling lack of early-on industry-wide standardization in the information-management sector should be a warning that if the need for these systems to talk to each other...which implies they'll need to have EXACTLY the same coded definitions across the board from day one... isn't addressed IMMEDIATELY, we'll end up having the equivalent of Microsoft and Mac (and who knows how many other 'proprietary' self-drive systems) that do essentially the same things, but speak entirely different languages and don't even really think the same way. That will be an intolerable situation, and as little standardization as there exists even in the automotive industry today mechanically (which is a HUGE waste of effort and money that seems to be totally ignored), where is the indicator that things will be any more rational in a self-drive-car world?

When I stand back and look at the lack of a coherent energy policy worldwide, which is a perfect indication of humans' inability to work well together on large-scale issues of global importance, even though the root technologies to end our dependence on traditional fossil fuels have existed for far longer than I have, I'm not encouraged that humans will get this right. At least, not for quite a while, if ever.

There's also every indication that the self-drive world will be every bit as vulnerable to political posturing and idiot rule-making by inept government intervention as most things are today. You only have to look at the state of the "climate change" controversy to see how people REALLY work together. There is NO DOUBT THE WORLD IS GETTING WARMER. Any fool with access to the internet can find raw data (as in glaciers that were there when I was a child now simply GONE) to PROVE it. But rather than accepting the reality of the situation and saying "OK, it's happening; what do we need to DO about it", there's still an ongoing petty war about WHY it's happening and who should be to blame...or not...instead of a concerted global effort to DEAL with the reality.

AI is rather a lot farther along than most people realize too, because of advances in the understanding and modeling of the hierarchical information processing in the human brain and its mathematical simulation that have occurred over the past decade. AI development hit a wall when people were trying to code for every eventuality, a series of "if > then" scenarios that soon became entirely unwieldy. When it became apparent that modeling a thinking machine on our emerging perception of how the human brain assembles concepts and pictures from tiny bits of data was the way to go, and then allowing it to learn based on its experience, everything changed. As of 2012, there were three essentially stand-alone systems that, if integrated, could easily have passed the classic "Turing Test" that is considered to be one proof of true artificial intelligence. I would not be surprised to find that a thinking, self-aware artificial mind exists as I write this. We can only hope that they'll proliferate, and do a better and more efficient and cooperative job of managing our future than we, as a species, have done up until now.

I personally don't want to live in an interim self-drive world where, besides being continually on the lookout for sleepy doofy unskilled texting human morons, I'll have to be keeping in the back of my mind the very real possibility of self-drive failures of the vehicles near me. The transition period from the way it is to the way people envision it is going to be a killer. 

 

 

Edited July 3, 2016 by Ace-Garageguy

[Joe Handley]

After doing a little research on this accident, it seems that the system in this can in fact be set to violate the speed limits. According at least one witness, the Tesla was going at least 25 mph over the posted speed limit for that road (60 MPH), which was the speed she was going  (she admitted to speeding herself) when he passed hew about a mile or so before the accident.

That was one big question I've yet to see an answer to, how fast was that car moving at the time of impact!  That could determine if the truck driver knowingly pulled out infront of the Tesla, how the car (which blew the safety bypass on the NHTSA roof crush test equipment) managed to lose the roof, then continue down the road through 2 fences, and then hit a power pole all 900' emerging from the wrong side of the trailer!

[Ace-Garageguy]

 

Cars of the future will not tolerate inept maintenance. They will warn you there is an issue, and tell you to take the car for maintenance.  There will be a cut off point where it simply will refuse to move. And in the case of self driving cars, it will make it's own appointment for service and go to the dealer at the correct time!

If you had spent a large part of your life working with greasy fingers on machines, you'd know that this is pretty much the way things already work, to an extent. It's called OBD II (Onboard Diagnostics) and constantly monitors many of the vehicles mission-critical systems and reports to the operator, through the "check engine" or "service vehicle soon" warning (and will report specifics through the OBD interface port to a suitable reader).

The problem is that in real life, inept maintenance often doesn't show up until a system failure is the result. For example, you can have correct fuel pressure (which the sensor reads and relays to the computer) even though a fitting is not torqued correctly. There is no "prediction" (other than statistically) of WHEN that incorrectly torqued fitting will loosen and spray fuel all over hot exhaust manifolds and incinerate the car and its occupants.

Failures of vehicle systems due to unknowable poor workmanship (and downright stupid design and material selection) happen constantly every day. There can not possibly be enough sensors ever placed in a vehicle to monitor EVERY DAMM THING. It just doesn't work that way. And electronic systems are NOTORIOUS for failing intermittently, or suddenly, and the onboard diagnostics have no clue as to what the problem is. Spend 20 years under hoods, and you'll know this.

Another LARGE part of the problem we see with vehicles of every description from almost every era and every manufacturer is that they were designed by engineers who'd never spent any time with tools, under cars, with barked knuckles. Some "engineered" stuff I see defies belief in its inept stupidity, as do many of the "repairs". It will make you ashamed to be human, and I sometimes cringe when I'm identified as an "engineer".

The problem gets worse and worse with every generation of vehicles. Stupid, poorly-thought-out over-complication combined with insane cost cutting, with clean-hands dweebs doing all the design and management without ever having input from highly skilled AND critical mechanics. Of course, you can find a lot more clean little mediocre middle-managers and computer-console "engineers" than you can find competent and articulate mechanics.

There's no reason to expect things to change, especially as long as most management types refuse to even acknowledge these problems exist. They build their retard carp, they get their paychecks, they play all day on social media, they neither know or care they're making junk. The consumer sees shiny new things with all kinds of bells and whistles and never thinks beyond that.

                                                                                                    

 

 

Edited July 4, 2016 by Ace-Garageguy