Do you think self-driving cars will actually happen?

[mbg]

I know it's technically possible, and we're just a few innovations here and there from putting the whole thing together.

But, given the regulatory environment, and the "minor" showstoppers involved, do you think they will actually be something most people will be using in future?

And how long until it happens?

I hear people talking about how so many of the technologies to make it work already exist (lane assists, self-parking, GPS navigation, etc.) and that it's just a matter of packaging... but then you consider that lane assist only works if the lane markings are visible, and that mapping data is never perfect, and you have to kinda wonder whether there are some insurmountable "showstoppers".

I mean, what if somebody did a mapping data screwup and everyone ended up piled into each other in the middle of a construction site?

[yao416]

No
Wrong forum

[Aznsilvrboy]

http://www.youtube.com/watch?v=t9Fxp3HK6DI

That's not lane assist either. There's only pylons, not lanes.

[mbg]

That's not lane assist either. There's only pylons, not lanes.

So we'd have to install pylons everywhere? Is that the idea?

I assume it's driving between "obstacles" to make that happen.

So instead of banana in the tailpipe, you spray shaving cream on its cameras to cause mischief :)

Or maybe just dense fog -- ha.

[olssy]

If the world as we know it continues to evolve in the way it is now then we will see self-driving cars for sure. Once they become available and statistics show they are safer than letting humans drive they might even outlaw human driven cars in some countries...

[Mr. Robo]

Not for another 100 years. The way I look at it, we would still require a driver's license in case the on board computer fries and we go to manual.

[NEMESIS_2008]

It will never happen:


No more drivers license to drive= lost revenue
No more speeding tickets or dui by programming the car to never break the law = loss of $$$$$$ for municipalities.

The only good thing is it will eliminate bad drivers (once the software is flawless, if that ever happens). Id love to sleep in the back seat while the car drives to work for me though :)

[mbg]

Not for another 100 years. The way I look at it, we would still require a driver's license in case the on board computer fries and we go to manual.

That's the part I don't really get.

Would you trust the average person to take over in the event that the automatic controls fail? Not that they wouldn't be capable, but you'd have to trust them to react quickly and know exactly what to do when it happened. You are asking someone who may not have had to really drive for 5 years (because it had been automatic) to instantly take over in the event something went wrong :)

I know airline pilots do similar things, but those guys are professionals and there are additional safeguards in place such as manned traffic control. They're not the people who can't work the self-service lane at the supermarket each and every time they try to to use it for 2 years straight.

I think it'll end up being one of those "technically feasible, but just not a good idea" types of things... like flying cars.

I wonder if the Google car is being piloted in Nevada because there are so many open, long, relatively empty roads with perfect weather?

[Ottomaddox]

Never say never. At the rate technology builds upon itself, nothing is out of realm of possibility.
I'm still waiting for a computer OS that intuitive enough for me to give a few basic commands, and let it do the rest.
(kinda like SickBeard...)

[mbg]

Never say never. At the rate technology builds upon itself, nothing is out of realm of possibility.
I'm still waiting for a computer OS that intuitive enough for me to give a few basic commands, and let it do the rest.
(kinda like SickBeard...)

Yeah, I am kinda wary of saying "never", but it's a fact that certain things just didn't pan out like we expected them to, or like the narrative told us they would.

Look at everything that was predicted for nuclear energy back in the 1960s, for example, and look at where we are now. Wasn't it supposed to be a wonderful new and clean energy source that would get cheaper to build over time and solve all of our energy problems? I'm sure people were extrapolating into the future that everyone would have a nuclear reactor in their house or something like that.

Same for space travel.

Let's not forget, we still haven't figured out a good way to get off the gasoline engine :) Computers are great, but there's a limit to what you can solve by pushing pixels around on a screen. Google engineers will be the last to realize this.

[doomhammer]

Yes I see it happening but not for another maybe 50 years or so. The technology still needs to mature a lot before it can be federally approved by whatever govt regulating body to be let loose on the general public.

[Oscillator]

I know it's technically possible, and we're just a few innovations here and there from putting the whole thing together.

But, given the regulatory environment, and the "minor" showstoppers involved, do you think they will actually be something most people will be using in future?

And how long until it happens?

I hear people talking about how so many of the technologies to make it work already exist (lane assists, self-parking, GPS navigation, etc.) and that it's just a matter of packaging... but then you consider that lane assist only works if the lane markings are visible, and that mapping data is never perfect, and you have to kinda wonder whether there are some insurmountable "showstoppers".

I mean, what if somebody did a mapping data screwup and everyone ended up piled into each other in the middle of a construction site?

You can't be serious. Top computer scientists in the world have spent 30 years designing robots with AI that still cannot exceed the intelligence of fruit fly, let alone a human. The type of AI required for a car to do self-driving would have to exceed the intelligence of humans, and we are no where near the technological singularity. Perhaps 1000 years into the future where the technology for faster-than-light travel, cloaking, etc. also exist may this be a possibility.

[Syne]

It will never happen:


No more drivers license to drive= lost revenue
No more speeding tickets or dui by programming the car to never break the law = loss of $$$$$$ for municipalities.


I hope you're kidding. Public revenue is disposable revenue, and wealth protectionism as it exists in the public sector is limited to the secondary concern of union job protection, whereas private sector you have much more aggressive protection measures that are directly related to the bottom line of an existing company.

Public revenue loss won't even slow down progress in this area. To think otherwise is just silly.

[JAC]

The infrastructure investment would be prohibitive. GPS technology isn't accurate enough, so you'd need some form of positioning beacons lining every street. Then you'd need central control systems with massive processing power and redundancy. The only way to recoup this investment would be pay-per-use highways like the 407, and the rates would be rather unattractive, I think.

[mbg]

You can't be serious. Top computer scientists in the world have spent 30 years designing robots with AI that still cannot exceed the intelligence of fruit fly, let alone a human. The type of AI required for a car to do self-driving would have to exceed the intelligence of humans, and we are no where near the technological singularity. Perhaps 1000 years into the future where the technology for faster-than-light travel, cloaking, etc. also exist may this be a possibility.

This is a good point.

But, you know, there are "a lot of smart people" who expect this kind of thing to be available very soon. Personally, I'm skeptical, but the "we've got technology, dude" crowd are pretty confident about it.

[mbg]

The infrastructure investment would be prohibitive. GPS technology isn't accurate enough, so you'd need some form of positioning beacons lining every street. Then you'd need central control systems with massive processing power and redundancy. The only way to recoup this investment would be pay-per-use highways like the 407, and the rates would be rather unattractive, I think.

This is my thinking, too... it basically means you couldn't go anywhere where the necessary beacons weren't. I mean, what if you turned onto some rural road without the necessary objects, or even just a lack of clear delineation between road and ditch? And what about when everything's covered in snow, or if there's dense fog?

That's why I wonder if they didn't choose Nevada specifically to test this out and inflate their "miles driven" figure. It's like testing an electric car in California where the conditions are ideal for the battery to give stated mileage.

As already discussed, it's one thing to provide driver assistance features, but entirely another to encourage people to be so lazy and hands-off that they'd probably be either asleep, making macaroni & cheese, or having sex in the backseat when the car required them to take over the manual controls because it couldn't deal with something ahead, which might happen so infrequently that they don't even remember what to do anymore.

I can see why people used to playing in virtual worlds, where you can buy experience and build new roads by shifting pixels around the screen would think it's easily do-able, but the real world is something different.

[sandikosh]

You can't be serious. Top computer scientists in the world have spent 30 years designing robots with AI that still cannot exceed the intelligence of fruit fly, let alone a human. The type of AI required for a car to do self-driving would have to exceed the intelligence of humans, and we are no where near the technological singularity. Perhaps 1000 years into the future where the technology for faster-than-light travel, cloaking, etc. also exist may this be a possibility.

They don't have to engineer a car to be smart as a fruit fly. All that is required is to have each car be able to read each and every other other signals and to react to it appropriately. It is like the avoidance systems on planes. As smart as humans are, we still have accidents.

[stealth]

No, I don't think any company would want to assume the liability if anything tragic ever happened with one of their self driving products.

[althetrainer]

Even if they made it and I could afford it I would never buy one. I don't want any computer malfunction worse yet a crash; I value my life a lot greater than technology! :D

[maasai]

I don't see it as being too far off.
1. We already have gps to map the routes, show detours and could feasibly show speed limits on each road.
2. We already have self parking cars that can navigate small spaces.
3. We already have cars that will slow down or stop based on a vehicle or object in front.
4. Also have cars with blind spot monitors.

Just need an inexpensive way to marry/sync the above to make it affordable for the general public.

[sandikosh]

I don't see it as being too far off.
1. We already have gps to map the routes, show detours and could feasibly show speed limits on each road.
2. We already have self parking cars that can navigate small spaces.
3. We already have cars that will slow down or stop based on a vehicle or object in front.
4. Also have cars with blind spot monitors.

Just need an inexpensive way to marry/sync the above to make it affordable for the general public.

Not only that but cars are becoming more electrical and less mechanical. Electronics are already replacing the need for power assisted steering, brakes and throttle.

[dealseeker2011]

I wish I could still see this happen. I guess, it will happen. Nothing is impossible. Human mind is so great that we won't even know how far could it take us.

[king_george]

OK so this car ran acceptably in the states in a fairly low-density area.

Let's see this thing travel around Brampton at rush hour.:D It would need a language module that could swear in several languages and another module that turns off the signal lights during a lane change. Oh yes and a module for honking the horn every few seconds.

I'd pay to watch that.

[DearSummer]

Even if they made it and I could afford it I would never buy one. I don't want any computer malfunction worse yet a crash; I value my life a lot greater than technology! :D

Because cars driven by humans are so safe... ;)

[Oscillator]

They don't have to engineer a car to be smart as a fruit fly. All that is required is to have each car be able to read each and every other other signals and to react to it appropriately. It is like the avoidance systems on planes. As smart as humans are, we still have accidents.

You realize how much infrastructure that would require? Even then, the most advanced signal processing designs still fail on a daily basis, like traffic lights for example and that's just the tip of the iceberg. It would just be hopelessly infeasible.

[uber_shnitz]

If it happened, the entire road system would have to be automated in which case, why not simply use some adapted form of automated public transit?

[MrKap]

I mean, what if somebody did a mapping data screwup and everyone ended up piled into each other in the middle of a construction site?

Sure. There is a lot of smoke and mirrors, but if they do get it right, it will eventually go through, only they will have a tough time selling the concept to local governments who would have to approve that sort of thing, correct?

Considering they basically mapped the entire planet with street view, anything seems possible with them, however they might have a reputation for escaping any liability or responsibility for everything they do as well, so if you ask me, that will be the show stopper for them.

Maybe in theme parks and such first, or private communities, ect... as for letting it loose on the general public as some sort of taxi or public transit system, I don't know, but it probably wouldn't happen overnight.


If anything, why not just some sort of extension to cruise control? That's been around forever. So in a sense cars already, "self drive" as for complete and total autonomy, probably won't happen.



However, let's say they do get it right? Who is liable for an accident and how does that effect insurance policies? That will be the really interesting part. Jmo...



This is my thinking, too... it basically means you couldn't go anywhere where the necessary beacons weren't. I mean, what if you turned onto some rural road without the necessary objects, or even just a lack of clear delineation between road and ditch? And what about when everything's covered in snow, or if there's dense fog?

Well, let's say they couldn't use infrared lasers to determine distance. I don't know how they do it, but what about eco-location?

http://en.wikipedia.org/wiki/Human_echolocation


In the end, the visibility available to humans, is more or less available to technology, with extra capabilities as well.


You guys have seen Big Dog, right?


http://www.youtube.com/watch?v=W1czBcnX1Ww


It is hard to look at that, and think that it isn't one of the most organic moving computers, ever. It's pretty smart looking.