Volvo will claim full liability in case of self-driving car crash

[Curufinwe_wins]

I cannot argue with you because you're right. Computers are massively better at us in almost every aspect of piloting any vehicle you can name, doesn't change the fact that without SPECIFIC PROGRAMMING it cannot deal with a situation where a human can.

 

As I said, fully autonomous vehicles will never happen, humans will always be able to override, that said though I fully expect self driving vehicles to almost entirely eliminate road accidents.

 

Don't get me wrong, I'm not against them at all, the problem right now is the legal mess and the moral questions they pose. If these things can be worked out then they will change the world. But eventually it will also happen that somewhere a SDC will kill a pedestrian, and when that happens the shit will hit the fan big time.

The moral questions pop up only because we allow them to. Morality is not absolute and that is arguably the biggest issue with it. (human logic also fails in the same manner because we define our systems differently and hence arrive at different conclusions, even if we had the same process.)

 

I agree there are major road blocks, and I don't see it happening anytime soon, but my point is that absolutely the system would be better off without humans involved whatsoever (I would argue the less human involvement even in emergency situations the better, because increasing human presence increases unpredictability in an unacceptable fashion.)

 

I also believe personally that the biggest issues will occur when most drivers are autonomous with a select minority of actual drivers, because optimizations that are in place for autonomous systems will break down in the presence of human action (whereas prior to majority autonomization, the optimizations take human factors more into account, which may be safer if a human is around, but certainly decrease the overall efficiency of the system.)

[FuzzyYellow]

the other thing is, what happens when the computer in the car gets too much information, and pauses for a second to process everything? do you drive into a light pole? another car? do you go off a cliff? 

 

we need A LOT more real-world testing before these things should be trusted.

[TriviaBeast]

Can't wait for car viruses or Ddos attacks at the mall clogging the parking lot.

Do you think Amazon will sell one for a bit higher price without ads?

Will Google's car drive you to the nearest McDonald's drive thru unless you skip the ad before you get there?

I'm only poking fun but as imperfect as cars and computers are right now it's tough to imagine a smooth transition.

Also, "not it" on testing automated sport bikes.

[pagani123]

self driving cars will be a good thing in about 10 to 15 years. they still need alot more testing and i dont see them commercially viable for at least 5 to 10 years if not longer for now they need to be tested and kept out of the hands of the public. i mean look at quadcopters 10 years ago quadcopters were uncontrollable and a very skilled pilot was needed to actually control the thing. 5 years ago the parrot ar drone came out saying it was really easy to fly but it wasn't and it had tons of bugs i had one for about 2 weeks until it flew away into the woods. now we have quadcopters that actually work as envisioned but they are still expensive and a lot of r and d went into them mostly by the military. driver assisted cars are going to become viable in the next 5 years and they are going to be used in limos and trucks first. we dont even really have laws on these types of cars yet and we expect them to be fitting in with normal cars in 2 or 3 years. 

[Curufinwe_wins]

Here's a good source that shows the difference in long term savings between electric cars and gasoline cars. I don't agree with including the battery replacement in there due to the fact that A. It is covered by insurance and B. The batteries are rated to last 6 years before they start to lose their efficiency, and even then they don't really start to degrade physically until around the 8-9 year mark.

 

Hydrogen or Hydrogen Fuel Cells are not a viable option due to our lack of hydrogen production that is available to us and how long it takes for hydrogen to replenish itself naturally. I've never heard of Hydrocarbon generation from waste heat for cars. Can you explain that to me? I heard about it being used in a few factories to regain lost energy from their emissions, but that's about it, nothing on a small scale level.

Ok time for a power generation side track:

 

Even at peak you are looking at ~85% efficient batteries. Discarding battery cost for a moment. Current reactors systems operating at sub 500C (so mainly steam cycles) are generally on the order of 35% efficient (which makes up the majority of the fleet). Modern Diesel (gas is worse, but still better than large turbine systems) in an internal combustion engine can reach upwards of 48% efficiency (due to its very high operating temperatures). Unfortunately, they don't scale well at all, which is why Diesel isn't really used to produce electricity except at peaking plants. 

 

So an electric vehicle from that perspective (with say a 100% sub 500C generation portfolio) has a real efficiency of 30% (.85*.35). Now there are reactor designs (these include most natural gas plants and the newest highest temperature coal, solar, and nuclear plants) that operate at sufficiently high temperatures to reach 40-50% efficiency, but still that puts electric vehicles behind. 

 

The caveat though is that this doesn't include the differences in distribution costs for electric and direct internal combustion (after all in theory using super conductive transmission lines, you could have nearly 0 transmission/distribution losses while hydrocarbons always have to be transported with significant loss.) The two right now are fairly comparable but that's why I say electric has the POTENTIAL to be both cheaper and more efficient than internal combustion.

 

The idea with co-generation as its called (producing hydrogen or hydrocarbons from waste heat) is that the exhaust temperature of reactors is still far above ambient (think tailpipe), so you could in theory add systems on that even though they are not particularly efficient the opportunity cost of them (since otherwise the heat is "wasted") allows you to generate additional energy in your product stream (in the form of hydrogen or then carbon neutral hydrocarbons, which are carbon neutral because you spent energy to lock the carbon in the first place into the hydrocarbon form so it's a closed loop.)

 

The issue becomes that it's basically always cheaper in the unit cost of whatever you end up producing to skip co-generation and just straight up generate the product as a full stream negating the benefits in the first place (aka the cost of materials for co-generation is massive, and less than linearly scaling with stream volume). Furthermore, in the case of hydrogen, by far the easiest method of generation is steam methane reformation, which as the name implies is no where near carbon neutral, and worse yet takes relatively high temperatures to take place. In fact as a result, well over 90% of all hydrogen product in the world is generated from fossil fuels (as then you can use methane for both the reaction and the heater, simplifying the overall design).

 

Co-generation then makes the most sense in a world where one source of energy is massively (at least two times) cheaper than any other form of energy (to counteract the additional process inefficiencies for conversion). This is of interest to the nuclear industry because unlike solar and wind (which are relatively non-scalable, aka there is a finite amount of energy around to be generated from it even if fairly large) and unlike fossil fuels (in that it is essentially greenhouse gas neutral), nuclear energy is in the long term essentially infinite and clean meaning that in the limit of time, only nuclear can scale to the magnitudes needed (wind and solar only compete semi-long term because the sun is nuclear lol) eventually, and after a certain point, either cogeneration or direct electric transportation (depending on if the per unit battery efficiency loss is less than the per unit transmission and conversion costs) will be a viable and necessary option. Also co-generation could in theory allow nuclear (or anyone else) to operate beyond baseload power (which for renewables would allow them to level/regulate their distribution curves, and for nuclear it would prevent shutdowns when the needed power is below a certain threshold). Again the same issues apply.

 

BTW please don't start any sort of pro/anti nuclear debate from this topic. Everything is stated manner-of-fact not as an indictment or support of nuclear power, and trust me I have heard everything there is to hear on the subject, and it is not a path you want to go down with me (esp because almost no one comes in with a willingness to change if proven wrong, supporters included.)

 

 

Sorry for the super wall, but it was a complicated response.

[Dabombinable]

Now to get Gaben to claim full liability for all of those dying for Halflife 3.

[mattebad]

Ok time for a power generation side track:

 

Even at peak you are looking at ~85% efficient batteries. Discarding battery cost for a moment. Current reactors systems operating at sub 500C (so mainly steam cycles) are generally on the order of 35% efficient (which makes up the majority of the fleet). Modern Diesel (gas is worse, but still better than large turbine systems) in an internal combustion engine can reach upwards of 48% efficiency (due to its very high operating temperatures). Unfortunately, they don't scale well at all, which is why Diesel isn't really used to produce electricity except at peaking plants. 

 

So an electric vehicle from that perspective (with say a 100% sub 500C generation portfolio) has a real efficiency of 30% (.85*.35). Now there are reactor designs (these include most natural gas plants and the newest highest temperature coal, solar, and nuclear plants) that operate at sufficiently high temperatures to reach 40-50% efficiency, but still that puts electric vehicles behind. 

 

The caveat though is that this doesn't include the differences in distribution costs for electric and direct internal combustion (after all in theory using super conductive transmission lines, you could have nearly 0 transmission/distribution losses while hydrocarbons always have to be transported with significant loss.) The two right now are fairly comparable but that's why I say electric has the POTENTIAL to be both cheaper and more efficient than internal combustion.

 

The idea with co-generation as its called (producing hydrogen or hydrocarbons from waste heat) is that the exhaust temperature of reactors is still far above ambient (think tailpipe), so you could in theory add systems on that even though they are not particularly efficient the opportunity cost of them (since otherwise the heat is "wasted") allows you to generate additional energy in your product stream (in the form of hydrogen or then carbon neutral hydrocarbons, which are carbon neutral because you spent energy to lock the carbon in the first place into the hydrocarbon form so it's a closed loop.)

 

The issue becomes that it's basically always cheaper in the unit cost of whatever you end up producing to skip co-generation and just straight up generate the product as a full stream negating the benefits in the first place (aka the cost of materials for co-generation is massive, and less than linearly scaling with stream volume). Furthermore, in the case of hydrogen, by far the easiest method of generation is steam methane reformation, which as the name implies is no where near carbon neutral, and worse yet takes relatively high temperatures to take place. In fact as a result, well over 90% of all hydrogen product in the world is generated from fossil fuels (as then you can use methane for both the reaction and the heater, simplifying the overall design).

 

Co-generation then makes the most sense in a world where one source of energy is massively (at least two times) cheaper than any other form of energy (to counteract the additional process inefficiencies for conversion). This is of interest to the nuclear industry because unlike solar and wind (which are relatively non-scalable, aka there is a finite amount of energy around to be generated from it even if fairly large) and unlike fossil fuels (in that it is essentially greenhouse gas neutral), nuclear energy is in the long term essentially infinite and clean meaning that in the limit of time, only nuclear can scale to the magnitudes needed (wind and solar only compete semi-long term because the sun is nuclear lol) eventually, and after a certain point, either cogeneration or direct electric transportation (depending on if the per unit battery efficiency loss is less than the per unit transmission and conversion costs) will be a viable and necessary option. Also co-generation could in theory allow nuclear (or anyone else) to operate beyond baseload power (which for renewables would allow them to level/regulate their distribution curves, and for nuclear it would prevent shutdowns when the needed power is below a certain threshold). Again the same issues apply.

 

BTW please don't start any sort of pro/anti nuclear debate from this topic. Everything is stated manner-of-fact not as an indictment or support of nuclear power, and trust me I have heard everything there is to hear on the subject, and it is not a path you want to go down with me (esp because almost no one comes in with a willingness to change if proven wrong, supporters included.)

 

 

Sorry for the super wall, but it was a complicated response.

Woah, this is way more than I was expecting. And I definitely understand where you're coming from in regards to the fact that electric cars have the potential to be much cheaper than they currently are. Their efficiency also has to do with the fact that there has been almost no advancements in the battery industry in the past 20-30 years due to the limitations of storage and lithium-ion being the only real way to store a large amount of energy as of now. 

 

Now I understand what co-generation is, it's basically repurposing the lost energy in carbon emissions into usable energy for the energy. It's a great idea for the large scale (factories and what not) but I don't think it will be able to scale down to cars due to the costs and complexities involved in it like you said. I do believe that nuclear is the way to go in the future when it comes to small scale reactors as many other sources of power don't scale well past a certain point. We also have the advent of fusion coming upon us with the building of ITER (International Thermonuclear Experimental Reactor) being complete in 2020 and them being completely on schedule. Now obviously this is completely up in the air if we look at history. However, there has never been something like this before in history, where we can test fusion reactions in a dedicated facility. Previously the United States used the Lawrence Livermore National Labratory in California and the National Ignition Facility, but funding was low for the both of them, and they regularly had to make a lot of compromises. This marks the first time ever that Fusion research is getting real funding, and from multiple countries at that. This is just like how CERN wasn't taken seriously until they discovered the Higgs Boson, something that wasn't even thought to exist, and now they are trying to detect dark matter, another enigma in physics.

 

Don't worry, I won't argue about Nuclear Reactors pros and cons due to the fact that I have tried with people before, but usually they are too disinterested to care, or too ignorant to admit they are wrong

[Curufinwe_wins]

Woah, this is way more than I was expecting. And I definitely understand where you're coming from in regards to the fact that electric cars have the potential to be much cheaper than they currently are. Their efficiency also has to do with the fact that there has been almost no advancements in the battery industry in the past 20-30 years due to the limitations of storage and lithium-ion being the only real way to store a large amount of energy as of now. 

 

 

Power systems engineering (specifically with regards to nuclear material applications) is my career field so I have a lot of knowledge stored up on this stuff.

Batteries are actually much more limited by cost and scalability than anything else. There have been tens if not hundreds of announced battery designs that blow Li-ion out of the water at small scales, but bringing them to a relevant scale for production has been a massive challenge. Furthermore, lead-acid batteries are still by far the cheapest batteries per unit energy stored (car batteries are those), but they are particularly poor in latent energy on a unit mass (and only mediocre in latent energy per unit volume) basis which makes them hard to use in full electric vehicles.

 

Now I understand what co-generation is, it's basically repurposing the lost energy in carbon emissions into usable energy for the energy. It's a great idea for the large scale (factories and what not) but I don't think it will be able to scale down to cars due to the costs and complexities involved in it like you said. I do believe that nuclear is the way to go in the future when it comes to small scale reactors as many other sources of power don't scale well past a certain point. 

In both cases, I was talking about it being an issue the other direction. At some point, we will require so much energy that nuclear is the only way to generate it all in a cost effective manner (not to mention environmental responsibility). Furthermore, the issue has been that co-generation for even the largest reactors currently in use (which are about 1500 MWe Nuclear Plants, 3700-4000 MWth) still do not give off enough waste heat to make co-generation less expensive than pure re-purposing the entire production for hydrogen production. And until that barrier is reached, the price cannot get that high because pure generation plants would prop up first (again this is why fossil fuels generate almost all the hydrogen in the world.) 

 

 We also have the advent of fusion coming upon us with the building of ITER (International Thermonuclear Experimental Reactor) being complete in 2020 and them being completely on schedule. Now obviously this is completely up in the air if we look at history. However, there has never been something like this before in history, where we can test fusion reactions in a dedicated facility. Previously the United States used the Lawrence Livermore National Labratory in California and the National Ignition Facility, but funding was low for the both of them, and they regularly had to make a lot of compromises. This marks the first time ever that Fusion research is getting real funding, and from multiple countries at that. This is just like how CERN wasn't taken seriously until they discovered the Higgs Boson, something that wasn't even thought to exist, and now they are trying to detect dark matter, another enigma in physics.

 

Don't worry, I won't argue about Nuclear Reactors pros and cons due to the fact that I have tried with people before, but usually they are too disinterested to care, or too ignorant to admit they are wrong

I thank you there, but I would be shocked if anything substantial came out of ITER before 2025 and any sort of actual power generation from fusion is in my humble (and semi-expert) opinion extremely unlikely this side of the 21st century.

[iHardware Shelden]

Hmm I can only imagine what an EMP would do to these self driving cars... Other than that cool I guess?

[TriviaBeast]

Hmm I can only imagine what an EMP would do to these self driving cars... Other than that cool I guess?

The same thing it would do to any modern car after obd2. So pretty much incapacitate the overwhelming majority of cars on the road that use computers for any drive train purposes.

[iHardware Shelden]

The same thing it would do to any modern car after obd2. So pretty much incapacitate the overwhelming majority of cars on the road that use computers for any drive train purposes.

Yea we are really not prepared for something like that...

[Oasis]

Dear god, we're probably going to see self-driving car subscriptions to pay for this shit.

[NateGSR117]

But I love driving

[Admiral Naismith]

People are 1000x worse at driving than a machine could be.

 

Why are airplanes safer? 1. Because there are less of them. 2. Because the drivers JOB is driving them.

 

I'd take a machine over a human any day where my safety is concerned.

Airplanes will be interesting. It's inevitable with the way drone technology that someone peoples will try to get self-flying planes in place. But, as you pointed out, cars and planes are different.

 

Anyone who knows anything about flying would be against full on passenger planes being controlled by a computer or controlled remotely, but cargo planes are different..

 

Anyways,

[Curufinwe_wins]

Airplanes will be interesting. It's inevitable with the way drone technology that someone peoples will try to get self-flying planes in place. But, as you pointed out, cars and planes are different.

 

Anyone who knows anything about flying would be against full on passenger planes being controlled by a computer or controlled remotely, but cargo planes are different..

 

Anyways,

I think the more planes continue to fly in the sky the greater pressure will begin to coalesce for more and more automation both from an efficiency standpoint (with more accurate scheduling) and greater overall safety (again with completely predictable behavior traits.) I do however think we are a long way off from the density being great enough (plus a lot of the automation can be seamlessly integrated into human driven systems because the overall skill set is both higher and more uniform.)

[Voldy21]

It's called driving defensively. You leave plenty of room between yourself and the other driver so you have time to react. You also leave for where you need to be early, so you can avoid rush hour traffic.

 

The problem isn't that humans are bad at driving, the problem is that there are humans driving, who shouldn't be allowed to drive because they've no freaking clue what they are doing. If you don't feel safe driving, it probably means you don't have enough experience and skill when it comes to driving. Nothing worse than a nervous driver, they overreact to emergency situations, and cause wrecks.

In other words, there are humans who suck at driving. So the problem is that there are some humans who are pretty bad at driving. Which is the point that you're saying is wrong and yet also showing is right. Some humans suck at driving and even good drivers are fallible to committing certain mistakes that could lead to them getting into a car crash.

 

Here's the thing, you can't take people off the road for not driving well all the time. There are some times that people take their eyes off the road and will crash into someone for some reason. You could take them off the road then but that doesn't change the fact that that could happen to anyone except the most focused drivers. Also, there are assholes who are overconfident in their driving by cutting people off, tailgating, racing and other things that are just dickish things to do. All of these things are problems that we currently face and driverless cars would solve all of that. There would be no tailgating, racing, cutting other people off, taking eyes off the road, or any other human fault that has led to crashing into another person. That's the benefit.

 

Also, I challenge you to present to me a case where driverless cars would actually do worse than humans. I don't mean a vague scenario like, either the driver dies or the pedestrian dies, I mean an actual scenario because I have a hard time imagining a scenario where that could actually exist.

[Trik'Stari]

In other words, there are humans who suck at driving. So the problem is that there are some humans who are pretty bad at driving. Which is the point that you're saying is wrong and yet also showing is right. Some humans suck at driving and even good drivers are fallible to committing certain mistakes that could lead to them getting into a car crash.

 

Here's the thing, you can't take people off the road for not driving well all the time. There are some times that people take their eyes off the road and will crash into someone for some reason. You could take them off the road then but that doesn't change the fact that that could happen to anyone except the most focused drivers. Also, there are assholes who are overconfident in their driving by cutting people off, tailgating, racing and other things that are just dickish things to do. All of these things are problems that we currently face and driverless cars would solve all of that. There would be no tailgating, racing, cutting other people off, taking eyes off the road, or any other human fault that has led to crashing into another person. That's the benefit.

 

Also, I challenge you to present to me a case where driverless cars would actually do worse than humans. I don't mean a vague scenario like, either the driver dies or the pedestrian dies, I mean an actual scenario because I have a hard time imagining a scenario where that could actually exist.

We CAN increase the quality of driver education. The point I am refuting, is the idea that a computer can drive better than a human. People who drive reckless is not a solveable problem, as there will not be a driverless ONLY car, for quite some time (meaning a car without a steering wheel, gas pedal, and brake pedal). Even when such a car exists, there will still be plenty of people who refuse to allow a computer to drive for them.

 

Long story short, the human brain is more complex, and adaptive than ANY computer currently in existence, as such, there will ALWAYS be situations in which a human will have the advantage over said computer.

[QueenDemetria]

That's cool of them.

 

Other thought: I wonder what would happen if you had your car in autonomous mode and your texting, but you get pulled over. I wonder if there is a way of proving you were in autonomous mode?

[Curufinwe_wins]

We CAN increase the quality of driver education. The point I am refuting, is the idea that a computer can drive better than a human. People who drive reckless is not a solveable problem, as there will not be a driverless ONLY car, for quite some time (meaning a car without a steering wheel, gas pedal, and brake pedal). Even when such a car exists, there will still be plenty of people who refuse to allow a computer to drive for them.

 

Long story short, the human brain is more complex, and adaptive than ANY computer currently in existence, as such, there will ALWAYS be situations in which a human will have the advantage over said computer.

And for every one time a human can have an advantage over a computer, the computer will have 100s of advantages over a human.

 

Even if you assume perfect education, the frequency of mistakes even professional drivers make on the road extrapolated out to the entire driver population is a truely unacceptably level of failure.

 

As was posted earlier traffic and damage mitigation after a detrimental effect, are two area's humans simply cannot out-preform computers at all.

[Curufinwe_wins]

That's cool of them.

 

Other thought: I wonder what would happen if you had your car in autonomous mode and your texting, but you get pulled over. I wonder if there is a way of proving you were in autonomous mode?

Be in the passenger seat?

[Trik'Stari]

And for every one time a human can have an advantage over a computer, the computer will have 100s of advantages over a human.

 

Even if you assume perfect education, the frequency of mistakes even professional drivers make on the road extrapolated out to the entire driver population is a truely unacceptably level of failure.

 

As was posted earlier traffic and damage mitigation after a detrimental effect, are two area's humans simply cannot out-preform computers at all.

Considering that Driverless cars aren't officially a thing yet (they're just about to become a "thing" at best) I will go ahead and doubt you there.

 

I'd like to see these driverless cars testing in rush hour traffic, as well as being tested in rural areas. Have they encountered a pack of cyclists on a blind turn yet? There's still too many untested variables to declare driverless cars superior. Hell, if I remember correctly, the Google driverless car had trouble with a single cyclist at an intersection.

 

Like I said, the human brain is more adaptive than the most powerful computer that currently exists.

[QueenDemetria]

Be in the passenger seat?

I guess you could do that, but it looks like Volvos cars are just regular cars with "auto pilot"(unlike Googles car without a steering wheel), so you probably have to be in the driver seat to turn on and do 'some' driving. Thats what it looks like from this video:

[Curufinwe_wins]

Considering that Driverless cars aren't officially a thing yet (they're just about to become a "thing" at best) I will go ahead and doubt you there.

 

I'd like to see these driverless cars testing in rush hour traffic, as well as being tested in rural areas. Have they encountered a pack of cyclists on a blind turn yet? There's still too many untested variables to declare driverless cars superior. Hell, if I remember correctly, the Google driverless car had trouble with a single cyclist at an intersection.

 

Like I said, the human brain is more adaptive than the most powerful computer that currently exists.

A human also has a response time on the order of a second. You know 110 ft at 75 MPH....

 

"had trouble" as in operated in a perfectly safe manner (even if not particularly time efficient).

 

Google's fleet has over 2 million miles logged in densely populated S Cal obviously including quite a bit of rush hour. I do doubt they have done sufficient inclement weather testing thus far (as in ice...) but saying that a human can drive better than a computer over any significant stretch of time whatsoever is complete and utter horseshit that a) makes 0 sense and b) is refuted by the very presence of all the automation that already exists in cars.

 

Whether or not they are ready YET is completely irrelevant to the point you originally were trying to make.

[Curufinwe_wins]

I guess you could do that, but it looks like Volvos cars are just regular cars with "auto pilot"(unlike Googles car without a steering wheel), so you probably have to be in the driver seat to turn on and do 'some' driving. Thats what it looks like from this video:

Yea then you are boned. Although perhaps that's the point, keep people invested in the system until it is mature enough to remove the human mode all together (outside of emergency cases, even though personally I'd think by the time we got there it would be better to NEVER let people drive overall anyways.)

[Trik'Stari]

A human also has a response time on the order of a second. You know 110 ft at 75 MPH....

 

"had trouble" as in operated in a perfectly safe manner (even if not particularly time efficient).

 

Google's fleet has over 2 million miles logged in densely populated S Cal obviously including quite a bit of rush hour. I do doubt they have done sufficient inclement weather testing thus far (as in ice...) but saying that a human can drive better than a computer over any significant stretch of time whatsoever is complete and utter horseshit that a) makes 0 sense and B) is refuted by the very presence of all the automation that already exists in cars.

 

Whether or not they are ready YET is completely irrelevant to the point you originally were trying to make.

Well, we need to define "better" to begin with. I'm talking about safety, not just efficiency and consistency. As far as safety goes, the driverless cars still have a hell of a long way to go before proving themselves.

 

Not to mention the intangible quality of "driving a car is fun". Some of us do enjoy driving you know, even over long distances. Although I can admit, there HAVE been times that I would have preferred to go to sleep and wake up at my destination. That being said, I'm more likely to trust another human being to drive me than I am a car. and I don't really trust other drivers, at all. Even those in my own family.