Autonomous Cars

[Just_a_fan]

Like I said ,.. IF you abuse the privilege that they grant you.

Nah, it's no different to freedom. You have a right to freedom. Unless you mess up and then society removes that right for a period of time. You have a right to drive provided you meet certain conditions.

Anyway, it's all moot. The bastards will do us over one way or another...

[Phil]

Autonomous cars?

I don't see it happening. Not for a very long time or unless limited to very few places where traffic is predictable. The problem I see are two fold:

1.) technology/unpredictability
Imagine an autonomous car. The only way I can see the technology ever becoming good enough (fit for purpose) is if the circumstances in which these cars drive are predictable enough. If you don't have that, e.g. you have normal 'manual' cars on the road, it's too unpredictable for any machine relying on sensors to react to situations better than a human being would. And mind you, I am not only talking about other cars on the road here. I am talking about the drive through villages with potential children, pedestrians walking into traffic. Yes, a "machine" might react to the threat more quickly and perhaps in a more safe manner if the sensors are good enough to pick up on - but what if not? What if the person or animal running onto the road is not distinguishable from the surroundings? What if the sensors are dirty/compromised? Are there safe guards for these kinds of things?

Getting to traffic, a drunk driving a manual car into your car - the autonomous vehicle will not fare better than a human being. Chances are, either way, you will end up with a huge accident. Or will the autonomous vehicle swerve into a pedestrian instead? Sure, these things happen in every day life with humans. We're not perfect, we make mistakes. And when we do.... (which brings me to the next point)

2.) liability
...who is liable? If you commit a driving offence, it's your responsibility and you get charged and if necessary a court deems what, who and why is responsible. If you are driving an autonomous vehicle, who is liable? The company who built the car? Who built the faulty software? The faulty sensor that triggered a manoeuvre into on-coming traffic? The company who built the fail-safe that didn't work? Or the driver, who was sitting in the drivers seat, but wasn't following the road because he was sitting in an autonomous vehicle and didn't think it's his responsibility or judgment that was required? Even if the software wasn't faulty, there will be accidents. Who will be responsible and liable for those?

The only, ONLY way, I ever see autonomous vehicle working, is when you only have autonomous vehicles on the road. On roads which little to no unpredictability. All cars connected to some overall AI/program that monitors all cars and has full control over all the participants on that connected road they share. No sensors required to 'read' what the distance to the car in front is, since the overall program will see, know and be able to predict everything it controls.

But wait, we already have that in most countries. It's called public transportation, e.g. a train, that runs on tracks and has a very predictable path it follows. The railway company controls the trains and (sometimes) you only need a few 'pilots' for the trains itself. In theory, one would think that even that could be remote controlled too some time in the future.

Sure, some countries, it's probably less feasible to built train tracks into every village. Living in Switzerland, perhaps we are fortunate enough to already have that. What's the point of having an autonomous vehicle that's most likely prone to failing or too many circumstances that could create a problem?

Cheers,
A skeptic.

[Just_a_fan]

AV will only happen if legislated for by government. Government will have to say "from date X, only AVs will be allowed on the roads". Date X will have to be suitably far ahead to allow the manufacturers and car buyers to comply. Anything else is doomed to failure for the reasons of conflict between AV and human drivers. It might start with motorways (freeways/autobahns/autoroute etc) where traffic is all going in the same direction. You can only use the motorway if in an AV with the car in charge, for example. If those routes are the ones that move most freely and people see the benefit of AV use, the market will move to AVs quite quickly. Either way, it will require government action.

Of course, given time, it might be possible to assume, for legal purposes, that the AV is always right and that any crash involving AV and human drivers is the human's fault. Can't see that being a vote winner, however.

[AJI]

...The only, ONLY way, I ever see autonomous vehicle working, is when you only have autonomous vehicles on the road. On roads which little to no unpredictability. All cars connected to some overall AI/program that monitors all cars and has full control over all the participants on that connected road they share. No sensors required to 'read' what the distance to the car in front is, since the overall program will see, know and be able to predict everything it controls.
...

Announcement: Sydney, Australia ~2036

Due to ongoing safety concerns it has been determined that no human being shall be allowed, or able, to control any vehicle in any public space. Effective, January 1, 2040.

[Phil]

Even assuming restricted zones with autonomous cars. How well does the technology have to work in order to be 'safe' (e.g. safe enough)?

Another example:
A bird flies into the path of the car. Is, or will the technology going to be ever good enough that the "threat assessment" is good enough to know that in such a case, it would be more dangerous to rapidly slow down the car (or take avoiding measures) than to simply continue and see what happens and risk a collision with that small object?

What if it's a big bird? What if it's a rock on the road. A wooden plank. A puddle of water. More than a puddle?

These cars will be probably be expected to be driving the same speeds as normal usage now days, so anything up to lets say 30m/s. Connected to 4 tires. You have temperature fluctuations to deal with, varying degrees of sunlight (possibly irrelevant to sensors, however) and varying degrees of surfaces from slippery (ice) to grippy. The sensors need to be good enough to accurately control the car while considering all these factors. There's really no margin for error here.

I'm thinking, it's just going to be too much, too expensive, to ambitious to build anything like this controlled environment that it will need to ensure a safe drive on a mass scale. Wouldn't it just be easier to invest into public transportation (trains) and encourage people to use that instead of cars?

Reality is; with an every increasing population (and one that is able to afford and want cars), we are left with less and less room and as a result, accidents are happening more frequently. I'm doubtful the answer lies in a future with autonomous vehicles. Public transport, yes. Cars that still offer the benefit of door-to-door transport while fully autonomated? That's where I have my doubts.

It's a lovely idea and experiment (and especially from a technological perspective) to try to achieve and IMO I wouldn't doubt technology progressing to the point where it'd be close enough to fulfill some sort of statistic that it may become "safer", but I think it will still fall short. Perhaps because in the end, who would want an automated vehicle anyway beside a few technology geeks? At that point, it won't be more convenient anyway.

I think people are hung up on this 'idea' because it represents a huge technical challenge and perhaps could be seen as the next 'big thing', but do people really want this and use it too? Those that really see a benefit in it, how is using an automated car in the future than simply using public transportation? In countries that have a good train-network, it will most likely get you there safer and quicker (and cheaper).

[AJI]

...how is using an automated car in the future than simply using public transportation? In countries that have a good train-network, it will most likely get you there safer and quicker (and cheaper).

Because it's personal transport and human beings are fundamentally selfish… Us westerners want our own private door to door shuttle, that's why we currently ignore the public transport systems on offer in favour of driving ourselves.

The other thing that constantly gets glossed over on this thread, even though it has been raised several times, is; what about those who can't or don't want to drive? Children, the elderly, the disabled. How about those who are simply not confident in their own driving skills, whether that be permanently or temporarily? How about those who simply don't like driving?

It's all very easy for us, a bunch of petrol heads who do track days and love our cars and bikes and motor sports, to argue that we don't like it, but we're the minority…

[Phil]

....use the train, e.g. public transportation?

Fair enough, I don't know where you are from, and perhaps I do live in a rather small country (unlike possibly most others) that also happens to have one of the most dense public transportation networks in the world, but looking past the infrastructure in my own country, I'd say the public transportation network across most countries in Europe is fairly good, meaning you can pretty much get from your village to most other villages around you. I have no idea why children (in your example) should want to travel in a car anyway - or the elderly mind you.

Besides, driving cars is awfully inefficient from a cost/energy to person perspective, unless you have a bus parked in your garage and take multiple people with you everytime you use it.

As much as I hate to say it and yes, saying this as a petrolhead, the way population is increasing, driving cars is not going to be very feasible in the future. I'd argue by the time technology has made it possible for automated cars, we will be long past the point where it will actually be needed or wanted.


As for why *I* prefer using my car: Well, because I can. Because driving equals freedom. I don't have to satisfy someone else's time table, but my own. I want to go somewhere? Easy. I get in the car and drive. I don't have to stop in locations i don't want nor do I have to wait for the public transport to arrive. That's the only reason. And, last but not least - I actually like driving. If I had an automated car, the best thing about driving (for me) would be gone. I could just as well use public transport then.

But my point is; in this future where automated cars may become feasible, the world will likely be a very different place. Cost of driving will have probably increased exponentially as a result of far too many cars on the road and neglecting that, you also have parking spaces to consider. Driving to the city (most cities I would think) is already a challenge. I'd argue in most places, actually using the bus or the train is a much more quicker and cheaper way to get from A to B.

Now saying that, I am fully aware we also have countries like the US and Australia that are just too large and population (those that live outside) too wide spread to dream of a public transportation only future. But Europe is way more dense population and traffic wise.

[AJI]

....use the train, e.g. public transportation?

Fair enough, I don't know where you are from,

Australia

I have no idea why children (in your example) should want to travel in a car anyway - or the elderly mind you.

Because sending children somewhere door to door in a driverless car has a built-in safety mechanism. I wouldn't have ever sent my child (who is now an adult) in an Uber or a taxi because of the driver, but I would have considered it if the car was autonomous (the assumption being that the autonomous system is robust).
As far as the elderly go, why should they, at the end of their lives and generally physically less-able, be expected to walk to the train station when they have chauffeured themselves around in their own personal transport for their entire adult lives? In 2040 this will be me...

Besides, driving cars is awfully inefficient from a cost/energy to person perspective...

It doesn't stop us now, so why should it stop us in the future?

As much as I hate to say it and yes, saying this as a petrolhead, the way population is increasing, driving cars is not going to be very feasible in the future. I'd argue by the time technology has made it possible for automated cars, we will be long past the point where it will actually be needed or wanted.

That's a big call. Bigger, I dare say, than my 2040 autonomous call… That would require a social shift that I don't think the western world will ever accept. If you give people the option of their own personal vehicle that they are not allowed to drive, or the train, which option do you think they'll take?

As for why *I* prefer using my car: Well, because I can. Because driving equals freedom. I don't have to satisfy someone else's time table, but my own. I want to go somewhere? Easy. I get in the car and drive. I don't have to stop in locations i don't want nor do I have to wait for the public transport to arrive. That's the only reason. And, last but not least - I actually like driving. If I had an automated car, the best thing about driving (for me) would be gone. I could just as well use public transport then.

Okay, so to be clear I've been playing the devils advocate role here, but from my own PoV I'm a petrol head boy racer. I love driving and riding and I love the freedom and convenience, exactly the same as you, but I have to refer you to the two options above and ask you to answer honestly.

my point is; in this future where automated cars may become feasible, the world will likely be a very different place. Cost of driving will have probably increased exponentially as a result of far too many cars on the road and neglecting that, you also have parking spaces to consider. Driving to the city (most cities I would think) is already a challenge. I'd argue in most places, actually using the bus or the train is a much more quicker and cheaper way to get from A to B.

And I'd argue that AV's can solve the problems you've listed. How can perfectly sequenced traffic flow make traffic worse? How can a car that doesn't have to park near you (or ever for that matter) create a parking problem? Individual car ownership may actually fall if there was a car sharing program?

Now saying that, I am fully aware we also have countries like the US and Australia that are just too large and population (those that live outside) too wide spread to dream of a public transportation only future. But Europe is way more dense population and traffic wise.

And this is my own selfish take on it. I currently live in rural country about 400k's north east of Sydney. The only way I can quickly and efficiently (and cheaply) get to Sydney (which I have to do about 20 times per year) is to drive myself. If my car drove me I'd be totally fine with that, even if it was at the cost of never being allowed to drive on a public road again. There'll always be the track to satisfy my need for speed...

[Andres125sx]

Like I said ,.. IF you abuse the privilege that they grant you.

Permissions are not privileges. To get a permission you need to receive some formation and pay for it, unlike privileges.

[Andres125sx]

The funny thing about this debate is people take a demagogic approach constantly, their position is "if AV are not perfect in every way, at every situation, and 100% safe, they can´t be used".

Then no human should be driving, period. Even if they´re not 100% safe and perfect, they´ll be several orders of magnitude safer than 90% of humans.

Take all those questions about AV and do it to an average driver..... What will an average driver do when a bird goes directly to his windshield? Many will panic, turn the wheel sharply tryinig to dodge it, and will even go over a pedestrian if there´s any. Should we ban human drivers then?

What if a small kid jumps into the road and the driver is distracted because he only has two eyes and they´re looking at the radio or dashboard? Do we ban human drivers then?

Perfect or not, they will be much much safer than the average driver because of three basic reasons:

- They never get distracted, wich is the reason for 99% of accidents (because the driver was drunk, lightng a cigarette or whatever, but the ultimate reason is almost always a distraction)
- They can use dozens of sensors 360º around the car
- Their reaction times are miliseconds instead of several tenths (humans)

Obviously they´re not perfect, but they´re much closer than we humans

[hollus]

Another point on how good humans are right now: humans often drive while a bit drunk, or a bit on drugs, or overly tired.

[Phil]

Because sending children somewhere door to door in a driverless car has a built-in safety mechanism. I wouldn't have ever sent my child (who is now an adult) in an Uber or a taxi because of the driver, but I would have considered it if the car was autonomous (the assumption being that the autonomous system is robust).
As far as the elderly go, why should they, at the end of their lives and generally physically less-able, be expected to walk to the train station when they have chauffeured themselves around in their own personal transport for their entire adult lives? In 2040 this will be me...

I still don't understand where a parent would want to send its child without adult supervision any place, other than perhaps to school? And if that is the place, how has society dealt with that in the past? When I was living in Australia (Canberra) I took the bus to travel there (~40 minutes using two separate buses). Children at a younger age, I'd assume there's a elementary school in your village.

Elderly, same point really. Most elderly can still safely drive cars beyond 70. Eventually, there will come a time though when it is no longer safe for them to. Depending on where you live and how chaotic traffic is, this may happen sooner or later. When this does happen, I would assume the public transportation system is an adequate way to get from one place to another. Beyond that, there will inevitably come a time when they can no longer care for themselves and will live in a place that cares for them and their needs. I don't quite see the requirements for AVs here for them, sorry.

Australia isn't the norm. I appreciate the fact that it's a very large country with a very small population and that the public transport network is perhaps not great, so many people (like in the US) that live outside the metropolitan areas rely on cars from getting them to places. If you take a look at Europe though, I think you will see that it is possible to have a very good network that works, is efficient, safe, fast and quite convenient, even when living outside the cities.

On the topic of AVs; you bring up a good point. Are we talking about AVs that are owned privately or are we talking about a system of 'public transportation' where AVs replace the buses/trains? If it's the latter, I suppose get that an AV system is more convenient because it's effectively door to door transport with no interruptions, but thinking into a future where such a model could exist, I would think the practicality of it would diminish given by the sheer number of people wanting to use it and it would still result in traffic jams, delays and congestion. So again - we're back to the point that a public transportation network is probably better.

It doesn't stop us now, so why should it stop us in the future?

Because it's becoming a bigger more crucial factor. Prices are increasing. More people are on the road. The whole world (most of it) are playing the 'environment card'. We want more efficient use of resources. With population increasing, this is inevitable. The dream of AVs down the road in my eyes doesn't really add up into this future of people getting into their AV that picks them up and drives them to their location. Even if these cars are fully electric, electric is still energy and energy is a (finite) resource.

That's a big call. Bigger, I dare say, than my 2040 autonomous call… That would require a social shift that I don't think the western world will ever accept. If you give people the option of their own personal vehicle that they are not allowed to drive, or the train, which option do you think they'll take?

If you put the question like this and today, sure, everyone will be shouting AV! It's an exciting dream for sure. An ambitious target. But to be honest, I really have my doubts that people really want this if they sit down and think about it. The only benefit to AV is the door to door transport. You can have that with Uber or taxi, the former quite affordable too. Why does it need to be autonomous (and yes, you already played the children and elderly argument)?

As to me, how I'd answer your question? Probably public transport system. But to be fair, the public transportation system here [in Switzerland] is fantastic. No AV will solve the congestion problems we have on the roads here as a result of increased people on the road. The train does. It's faster, cheaper and safer. And you get to read the news paper in a way more comfortable position than in a small car. Ever tried reading a newspaper as a passenger in a car vs on the train? Introduce corners and I know which I find more comfortable. Even the bus is way more comfortable than a car in that regard (lower lateral forces).

And I'd argue that AV's can solve the problems you've listed. How can perfectly sequenced traffic flow make traffic worse? How can a car that doesn't have to park near you (or ever for that matter) create a parking problem? Individual car ownership may actually fall if there was a car sharing program?

Perfectly sequenced traffic yes, but how much traffic? Assuming at least as many AV cars on the road like we have today (it's gonna be likely more), perfectly sequencing traffic comes at a cost. It will no longer be perfect A-B. Expect detours, longer travels as a result of synching that traffic. At some point, congestion will be unavoidable. Which brings me back to trains...

The AV dream works if you imagine fewer cars on the road, but if you think about how many people will use such a system in 20-40 years, I really think the sheer number will quickly diminish the novelty of it. Too complicated to work IMO.

And btw; there are countries that already have a car sharing program. Quite popular too. Perhaps just not in Australia [yet]?

[Phil]

The funny thing about this debate is people take a demagogic approach constantly, their position is "if AV are not perfect in every way, at every situation, and 100% safe, they can´t be used".

Then no human should be driving, period. Even if they´re not 100% safe and perfect, they´ll be several orders of magnitude safer than 90% of humans.

Take all those questions about AV and do it to an average driver..... What will an average driver do when a bird goes directly to his windshield? Many will panic, turn the wheel sharply tryinig to dodge it, and will even go over a pedestrian if there´s any. Should we ban human drivers then?

What if a small kid jumps into the road and the driver is distracted because he only has two eyes and they´re looking at the radio or dashboard? Do we ban human drivers then?

If there's a flaw in the software or the sensors that control hundreds or thousand (if not more) of cars, the impact is a little larger than if a single person 'makes a mistake' and causes an accident. Just because one person makes a mistake, it doesn't become a question of if the humans in general are flawed.

By the sheer numbers of people driving on the road (billions) in very unpredictable circumstances and the 'relatively' low rate of accidents, I'd say humans are displaying quite a remarkable ability of safely driving cars. Most people involved in car accidents, how many accidents did they have? Across how many years of driving? How many kms driven? I'm not convinced AV could do the same, even with a 'million sensors' around the car.

Besides, if a 'machine' causes a fatality or grave injury, who is accountable? Will it feel remorse? Guilt? Say sorry for making a mistake? And learn from it?

[AJI]

...

All good points and I appreciate the time taken to answer, but I can't ignore the fact that manufacturers are forging ahead on AV development. Perhaps it's some sort of elaborate long con that I can't see, but you have to ask the question, if it's truly a dead end why bother?
The manufacturers are in business to make money, and if after tens (possibly hundreds) of billions of dollars of R&D the AV dream is abandoned what was the point?

I admire your quasi romantic view that the individual will sacrifice personal transport in favour of a public system (albeit in your version it's almost entirely down to necessity) however I also can't ignore the fact that a lower percentage of the population take public transport now than ever before...

Anyway, it's clear that the AV viability question isn't going to be solved in this thread, so I guess the both of us get to keep on driving our performance cars while we wait and see what happens?

[Phil]

I think big car manufacturers and technology companies are pursuing this for PR benefit. Bragging rights, so to speak. It's not so much about if it is really viable, but it's good PR. It's exciting. I think robotics, AI and machine learning are probably one of the next big breakthroughs we are yet to encounter. And the technology there surely can be used for other things too, even if the autonomous cars never happen because it may not be feasible for economic or social reasons.

Perhaps I could see the technology being used in more predictable circumstances. Instead of imagining a future where everyone has a self-driving-taxi, I could see the technology end up in buses that follow a controlled path (but here too, I question why it's not easier to employ a driver). Or the tech will simply be there in an assistance function. Or I could see the tech being used for agriculture. Instead of being outthere doing it yourself, you could have a 'robot' doing it for you in an environment where there is little risk to cause harm.

No doubt, if there is a breakthrough, it will be great PR. Perhaps there will even be a small market for it and being the first will ensure a big win, similarly like Tesla being the first to produce usable and desirable electric cars people actually want to have and drive.

That's my guess. Right now, lots of things are being 'invented' for the sake of having something new, exciting and fascinating. Few of these things are actually really usable. And once you have it, the novelty wears off fairly quickly. The smartphone market is a great example. Now that we are slowly reaching limits of technology (battery and processing power per square inch), companies have to find new ways to get us to 'upgrade' every year to a supposedly new and better phone. Same applies to TVs. We had SD,then HD. Then FullHD (1080p) and now we're at 4k. Supposedly the big CE manufacturers are hard at work to bring us 8k. Really? Most people can't even distinguish 4k from 1080p because it comes down to viewing distance. And not everyone wants to have a TV screen that's the size of an entire wall (which it would take to take full advantage of that kind of resolution). Doesn't stop manufacturers pursuing these things, because if they don't, what makes us want to upgrade and not just stick with what we have that is 'good enough'? Same applies to sound formats. I'm an audiophile, but who in gods name needs and wants 7 channel surround at home, let alone 9?

Yes, sorry for the OT rant...