Autonomous vehicles

[Moriarte]

Little bit of background - I consider myself a trader/investor. Been doing it on and off over the last 10 years, lately it became full time job and something I still feel passionate about today. The way I do it is I try to find market leaders based on one important metric - technological breakthrough. Indeed, I look at many other things in a company: quarterly and annual financial statements, popularity among hedge-fund managers, membership in a sizzling-hot industrial group/economic sector, I look at their leadership, how open-minded and passionate and professional and educated they are. Finally, my signal for entering a position is what Jessy Livermore called "breaking the price resistance level on high volume of trade".

But the absolute foundation and what keeps me going is finding "the new thing", be it a product or service, which is bound to become popular. One day it was Microsoft, then Google, then Apple returned in spades, now it is Nvidia, shares of which I bought in March 2016. What caught my eye initially with Nvidia is the absence of serious competition in their GPU branch. What seems even more interesting is the latest push towards providing technology for automated vehicles. So, I'm here to ask intelligent people at CFC whether automated vehicles is the new thing worth puting money in. What do you think? Are we going to live like they shown us in minority report, the movie, in a couple years? Is it more difficult than it seems and bound to take decades? Is it a complete fluke or will remain a niche thing, like Oculus?

Opinions welcome!

https://www.bloomberg.com/news/arti...ates-a-computer-for-fully-autonomous-vehicles

 

[GenMarshall]

No. I'd rather drive the car manually than let a computer/robot take the wheels. Too many issues in regards to vulnerabilities on being hacked into and the "Trolley Problem" (An ethics thought experiment of flipping a switch to divert a train to save one person vs. five people).

Ideally, I see it more of a niche thing.

 

[Ryika]

No. I'd rather drive the car manually than let a computer/robot take the wheels.

The interesting question is: How long will you be allowed to do this? ^^

 

[tuckerkao]

The interesting question is: How long will you be allowed to do this? ^^

Safety is #1, as long as the vehicles know how to avoid the crashing accidents, the auto features will keep going.

 

[Ryika]

Safety is #1, as long as the vehicles know how to avoid the crashing accidents, the auto features will keep going.

No, I mean how long he'll be able to drive manually.

That's under the assumption of course that automated vehicles are the safer than manual drivers (which they seem to be), and supplemented by the idea that automated vehicles work better when there are just other automated vehicles around. They could be a way to make traffic flow better, and manual drivers, as forces that disrupt this flow, could be phased out at some point.

 

[EgonSpengler]

I'm been following the development of driverless cars off and on for a while. One thing to keep an eye on, whether you're an investor, or looking to buy one, or just digging the tech, is the legal questions/issues, which frequently drag behind the technology. California is in the process of moving forward on fully-autonomous (e.g. no human driver as a backup) vehicles as we speak.

Personally, I'm pretty excited about the potential of driverless vehicles. I think they won't just replace drivers in the same roles and for the same purposes we use vehicles for now, I believe they could open up new possibilities. The analogy that gets used a lot is the automobile replacing horses and animal-drawn wagons. Of course there are some questions that haven't been answered yet, the technology isn't anywhere close to "done", but stalling the process won't provide those answers (does it ever?). But when it comes to, say, the aforementioned "trolley problem", an overwhelming majority of traffic accidents are caused by human error, not circumstance. The human drivers in the analogy aren't the person throwing the switch, they're the tracks guiding the trolley into the crowd of people.

I would say that autonomous vehicles are a great long-term investment, but I wouldn't be able to say anything about the short-term. Which company to back, for example? No idea. Since this is a gaming site, people here might be interested to hear the NVIDIA is a big player in autonomous car technology (well, "big" player might be an overstatement, but they're in there). Does that mean one should invest in NVIDIA? Don't ask me, but at least one magazine says, "Detroit is stomping Silicon Valley in the self-driving car race" so buying into good ol' Ford Motor Company may be the high-percentage play (and when they say "Detroit", they really mean "established car companies", because they cite Renault-Nissan, Daimler, and Volkswagen as being ahead of Google and Tesla, in addition to Ford and General Motors).

That's under the assumption of course that automated vehicles are the safer than manual drivers (which they seem to be), and supplemented by the idea that automated vehicles work better when there are just other automated vehicles around. They could be a way to make traffic flow better, and manual drivers, as forces that disrupt this flow, could be phased out at some point.

As I was Googling around for that article about NVIDIA, I passed by a headline that said the biggest trouble for self-driving cars right now is the other cars driven by humans. It's not hard to imagine "AI-only" lanes on highways, but surface streets would be a more tangled knot. The simplest solution might eventually be to make human-driven cars illegal on public roads.

Some cities are already putting restrictions on motor vehicles. New York City considered a restriction similar to the City of London's, for the southern tip of Manhattan, but backed off. I think China and some European countries are working on eliminating gas-powered vehicles entirely.

I wonder, can one just ride a horse through a town or city today, if one wants to? Or was the drastic reduction in horses just due to the fact that people didn't want them anymore, once automobiles were cheaper?

 

[GoodEnoughForMe]

We're still decades away. We haven't even really figured out how insurance and fault will work if/when an injury or death occurs.

That said it is coming. It would save 10s of thousands of lives a year in the US alone. Autonomous vehicles would drastically cut down deaths due to distracted driving, falling asleep at the wheel, drunk driving, etc.

 

[warpus]

Autonomous vehicles are already (on average) safer than humans. But we are emotional creatures so the technology is going to take some time to be accepted as normal or wanted.

Eventually I forsee legislation in some places that will see autonomous-only car lanes. Once people get used to that, the scope will expand, and eventually I see a time when the majority of the cars on the road are autonomous. The U.S. is probably going to be fairly slow in terms of western countries to mandate a phasing out of non-autonomous vehicles. Having said that, we will always probably see the need for vehicles driven by humans.

It's like other things that have been automated over time. Look at photography. Even though digital cameras exist and allow you take a photo and see the photo right away, and even print it right away, there are still people out there who use film and take the time to develop their photos the oldschool way. They cling to the know-how of doing all that for various reasons. Being able to drive a car by yourself is a good feeling that a lot of people enjoy, so while the convenience and the "I don't have to think about it/humans are lazy" factor will eventually overtake that, I don't see human drivers becoming extinct. They will always probably exist, but eventually they are going to be a small niche, and in a lot of places you probably will not even be able to drive a car, for reasons of safety and efficiency.

The problem is that it is hard to predict whether any of what I said will take 5 years, 15, 25, 50, or 70 years.

It's also sort of impossible to predict where technology is going to end up. It's easy enough to see where it is headed, like I just did, but that's not always where things end up. Most people who try to predict technology end up being wrong. If it were easy, people like OP would be able to predict the future easier and invest in the right technologies and retire early.

 

[Lexicus]

That said it is coming. It would save 10s of thousands of lives a year in the US alone. Autonomous vehicles would drastically cut down deaths due to distracted driving, falling asleep at the wheel, drunk driving, etc.

A lot of person-hours saved as well, both from drastically reduced traffic problems and being able to do other stuff while you're "driving".

Eventually I forsee legislation in some places that will see autonomous-only car lanes. Once people get used to that, the scope will expand, and eventually I see a time when the majority of the cars on the road are autonomous. The U.S. is probably going to be fairly slow in terms of western countries to mandate a phasing out of non-autonomous vehicles. Having said that, we will always probably see the need for vehicles driven by humans.

My prediction is that insurance will make it prohibitively expensive to have a flesh-and-blood driver, and probably sooner than we all think. Driving cars will end up like riding horses, something rich people do for fun, rather than a serious option for ordinary people looking to get around.

 

[Valka D'Ur]

Safety is #1, as long as the vehicles know how to avoid the crashing accidents, the auto features will keep going.

How about the skidding accidents, the sliding accidents, and the blowing-into-the-ditch accidents, all of which are common in Canada in the winter, when a vehicle unexpectedly encounters black ice, particularly when there are wind gusts?

Seriously, how many of these driverless cars are tested anywhere that has real weather?


Oh, and how about the running-into-a-moose accidents? The cars are, I assume, programmed not to hit humans. Are they programmed to know what to do to avoid hitting a moose or deer, without having to swerve into oncoming traffic in the opposite lane? Thing is, when car meets moose, the results usually end up with either the human or the animal being killed or at least severely injured.

 

[Lexicus]

How about the skidding accidents, the sliding accidents, and the blowing-into-the-ditch accidents, all of which are common in Canada in the winter, when a vehicle unexpectedly encounters black ice, particularly when there are wind gusts?

Seriously, how many of these driverless cars are tested anywhere that has real weather?

An AI will be guaran-damn-teed to be better than humans at responding to these things. Much faster reaction time and it never does the wrong thing under pressure.

Oh, and how about the running-into-a-moose accidents? The cars are, I assume, programmed not to hit humans. Are they programmed to know what to do to avoid hitting a moose or deer, without having to swerve into oncoming traffic in the opposite lane? Thing is, when car meets moose, the results usually end up with either the human or the animal being killed or at least severely injured.

Any autonomous car worth the paper it's printed on would be programmed not to hit anything. More importantly, I think it likely that any autonomous vehicle would have to be programmed never to drive so fast that it couldn't come to a complete stop in time to avoid things that unexpectedly enter its path.

 

[Ryika]

Actually, one of the major issues when it comes to self-driving cars is whether there are situations where they should intentionally "hit" humans in a situation where it's impossible to avoid the crash, or whether they should risk the well-being of the driver by doing risky maneuvers such as crashing into a car on the side. It is completely impossible to drive in a way that you are able to react to anything, unless you're driving so slowly that it would likely be better if you just walked.

I'm sure though that self-driving cars will still yield way lower mortality rates, as they don't enter a state of "shock" when something suddenly shows up in front of you, and can thus react in a way that prevents as much damage as possible. A human just can't do that, he'll either not react at all, or just drag the car away from the moose in panic.

So really, the only issue I see is that you could hack and get control of the system. In a realistic scenario, a person could get control of your car and ramm you into a wall, in a horror scenario, Russia - yes, Russia! - creates a virus that makes all cars go wild. 8)

If that can be avoided, then in the future I am sure people will wonder why people ever wanted to drive themselves, when they could also sit in their car and do whatever while being transported.

 

[Lexicus]

If that can be avoided, then in the future I am sure people will wonder why people ever wanted to drive themselves, when they could also sit in their car and do whatever while being transported.

I can see the viral videos of people jerking off in their cars now...

It is completely impossible to drive in a way that you are able to react to anything, unless you're driving so slowly that it would likely be better if you just walked.

Or if you're an AI with a fraction of a human's reaction time. You do bring up a good point, though, which is that what to do when a crash is unavoidable is not so clear cut.

 

[civvver]

I think the harder issue here is identifying which companies you should invest in to take advantage of this tech. Ford, GM, Tesla, Google, tons of other small companies you probably can't just go invest in are all researching driverless cars. Which one is going to turn the most profits from it is probably impossible to say right now. It will happen eventually though. The idea of driving is generational. Less and less teens are getting driver's licenses every year because they don't need to go places. Their social lives are online now. Driving eventually will be seen as an inconvenience and it'll be a lot better to just have a driverless uber pull up when you need to really go somewhere.

 

[Valka D'Ur]

Any autonomous car worth the paper it's printed on would be programmed not to hit anything. More importantly, I think it likely that any autonomous vehicle would have to be programmed never to drive so fast that it couldn't come to a complete stop in time to avoid things that unexpectedly enter its path.

They're making these things out of paper?

They will definitely not have a chance against a moose. Moose, deer, and bears don't act or react like humans. Have these car makers taken this into account?

 

[Lexicus]

They're making these things out of paper?

It's just a figure of speech. To my knowledge there are no plans to make any cars, automated or not, out of paper.

Moose, deer, and bears don't act or react like humans. Have these car makers taken this into account?

That's a good question. I would hazard a guess that they probably have; it's not like cars hitting animals (particularly deer, at least around where I live) is uncommon.

 

[warpus]

How about the skidding accidents, the sliding accidents, and the blowing-into-the-ditch accidents, all of which are common in Canada in the winter, when a vehicle unexpectedly encounters black ice, particularly when there are wind gusts?

Seriously, how many of these driverless cars are tested anywhere that has real weather?


Oh, and how about the running-into-a-moose accidents? The cars are, I assume, programmed not to hit humans. Are they programmed to know what to do to avoid hitting a moose or deer, without having to swerve into oncoming traffic in the opposite lane? Thing is, when car meets moose, the results usually end up with either the human or the animal being killed or at least severely injured.

It's Tesla and Google pioneering these types of cars I believe. There's no way they're not testing for all sorts of extreme situations. An autonomous car would be useless if it only worked in ideal conditions.

 

[EgonSpengler]

One thing to bear in mind with autonomous cars is that their sensors are better than human sensors, in many ways. The trouble with black ice is not that it's slippery, it's that it's hard to spot before you drive onto it (it's called "black" ice because it's clear - it's the road under the ice that's black). Autonomous vehicles are being built with lidar, cameras, and multiple radar systems (one for far away and one for up close). The cameras take 60 frames per second and won't have to be limited to visible light, and the radar systems might be able to use reflections to 'see' around corners. I think there's already an automatic parking feature that uses sound waves, believe it or not; very short range, not enough to avoid hitting a moose on the highway, but it will prevent you from running over the toddler who's shorter than the car's hood (I think the term "sonar" only applies to underwater sound wave devices, but if that's true, I don't know what "air sonar" is called).

There's a whole vision lab at MIT doing things like teaching computers to differentiate between the image of an object reflected in a window and the same object on the other side of a window (it's interesting to think about - our brains always seem to know the difference, but how?).

 

[Lexicus]

(I think the term "sonar" only applies to underwater sound wave devices, but if that's true, I don't know what "air sonar" is called).

It probably shouldn't. Per wikipedia it was originally an acronym for sound navigation and ranging. But I think we should call it a sonic rangefinder when used in air

There's a whole vision lab at MIT doing things like teaching computers to differentiate between the image of an object reflected in a window and the same object on the other side of a window (it's interesting to think about - our brains always seem to know the difference, but how?).

There's a really great Richard Feynman quote where he talks about how utterly awesome it is that the brain can build such a rich-seeming model of the world based on a relatively very small number of photons, and such tiny variation in how they reach the eye. It really is mind-boggling if you think about it.

 

[warpus]

I think our brains use context to sort it out. So, the image will move differently depending on where it is and our brains quickly pick up on that. But that is a complete guess