How long before most automobile driving is done by computers?
Displaying poll results.28778 total votes.
Most Votes
- What's the highest dollar price will Bitcoin reach in 2024? Posted on February 28th, 2024 | 8361 votes
- Will ByteDance be forced to divest TikTok Posted on March 20th, 2024 | 2432 votes
Most Comments
- What's the highest dollar price will Bitcoin reach in 2024? Posted on March 20th, 2024 | 68 comments
- Will ByteDance be forced to divest TikTok Posted on March 20th, 2024 | 9 comments
Never gonna happen. (Score:5, Funny)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
Re:Never gonna happen. (Score:5, Insightful)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
Ironically racing is probably a better defined operating environment so easier to successfully automate.
I can't see everyday driving being automated because there will always be a scenario that isn't covered, an odd bug that no-one expected, or mechanical failure that the computer can't compensate for in a sensible way.
Commercial flight systems on aircraft must be so thoroughly tested, yet we still have pilots to take over if the machine fails. How would that work with cars? If you have to be there paying close attention in case you need to take over, doesn't that negate the purpose of the automated system in the first place?
Re:Never gonna happen. (Score:5, Insightful)
I would assert that the population of terrible, drunk/drug addled drivers on the road today are far more dangerous that the odd software error we will see. Especially since each failure of automation will result in improvements over time.
Re:Never gonna happen. (Score:5, Insightful)
Re:Never gonna happen. (Score:3, Interesting)
Building automated cars that only drive around other automated cars is 80s technology. The whole point of automating driving is getting the algorithms good enough to handle "erratic drivers" which are just one of the hazards of the road.
I would turn your sentence around and say automated cars will be the biggest threat to anyone without top notch reflexes. Unlike a human, an algorithm can control the speed and breaking distance to 1 inch from the next car during an emergency stop. The automated cars will also communicate their intentions with one another and have practically no latency in issuing the command, so they will be able to drive faster and more aggressive than a human, while still safe. This will create a dangerous environment for unaccustomed drivers, who will tend to drive "with the flow", that is: faster than the road conditions allow for a non-robotic driver.
Re:Never gonna happen. (Score:5, Funny)
Ahh! This takes me back. I remember a time around 1954 when bumpers were bumpers, and my dad and uncle were driving from Providence to Pittsburgh, playing bumper cars all the way. Fortunately this was before seat belts so we kids could climb up on the back window ledge for a good look.
Re:Never gonna happen. (Score:5, Funny)
Re:Never gonna happen. (Score:4, Interesting)
No, the biggest threat to automated driving algorithms is liability. As long as the manufacturer is liable for accidents they'll never get anywhere.
Re:Never gonna happen. (Score:3)
Re:Never gonna happen. (Score:4, Interesting)
I think we should start with the Interstate system. A vehicle can operate manually until hitting an on ramp where a computer will take over and log into the roadway system. In this manner things can happen gradually as no one will be able to take the Interstate unless their car is compatible. Others can continue to use the back roads and streets until over time the busiest ones will also be included in an automated road system. Eventually only the smaller side streets will continue to be manually operable.
Re:Never gonna happen. (Score:3)
It would be safer and more cost efficient to put cars on larger vehicles, such as trains and ships. A single large unit can be navigated safer than hundreds or thousands of smaller units.
It's been available for years. It hasn't been adopted for heavy use.
http://www.amtrak.com/auto-train [amtrak.com]
There are also truck and ship transports, but as there are so many services, it's not practical to list them. Some ship transports allow passengers. Most truckers can't allow passengers.
Re:Never gonna happen. (Score:3)
There's a reason it hasn't been adopted for heavy use: it's overly expensive, and restrictive. People want cheap, on-demand, end-to-end, and preferably private transportation. This is why people buy cars. So unless this automobile bulk transport option is cheap enough to offset the route and time restrictions, it ain't gonna fly.
Re:Never gonna happen. (Score:3)
I would like to add - what about collectors? Surely they'll never put automated systems inthem.
We'll just bounce them off the road with our MSVs.
Re:Never gonna happen. (Score:3)
I would assert that the population of terrible, drunk/drug addled drivers on the road today are far more dangerous that the odd software error we will see. Especially since each failure of automation will result in improvements over time.
One of the best things about automated cars will be the ability to get home after drinking. Not much of an issue in built up areas with public transport, but being able to get home from the pub will be great for the social lives of people in rural areas.
Re:Never gonna happen. (Score:3)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
Ironically racing is probably a better defined operating environment so easier to successfully automate.
Physically? Yes, but nobody's going to make a billion dollars on slot-cars.
Racing is the one aspect of driving that will not ever be automated, nor should it.
Re:Never gonna happen. (Score:2)
how do you know, maybe there would be competitions and/or betting on fully automated cars: which team can build the best. there are similar competitions today with robots
Re:Never gonna happen. (Score:2)
how do you know
I'm speculating, just like everyone else does in a "What If" conversation. Just like you did in the rest of your post:
maybe there would be competitions and/or betting on fully automated cars: which team can build the best. there are similar competitions today with robots
Yea, maybe. But who's going to put up the millions of dollars in front money to find out? You?
A good part of what makes auto racing entertaining is the personalities of the drivers and their varying skill levels. Take away those elements, and it all basically becomes NASCAR, but without anything entertaining happening, ever.
Re:Never gonna happen. (Score:2)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
Ironically racing is probably a better defined operating environment so easier to successfully automate.
Also ironic, is that I suspect the GP was writing that with intentional irony.
Re:Never gonna happen. (Score:5, Insightful)
If a car "autopilot" detects a problem, it can just slow down, pull off the road, and shut off. You do not have such an option in an aircraft.
In a self-driving car, you only need to "be there paying close attention" if you have reason to believe the navigation system will NOT properly detect when it is operating in conditions beyond its capabilities.
Re:Never gonna happen. (Score:3)
The thing with planes is that you usually have quite some time to deal with problems. The sky is big, your plane is small. That means if the autopilot can't handle something you usualy have a fair bit of time for the human pilots to take over (which is not to say there haven't been crashes where the pilots thought the autopilot was engaged when it wasn't).
With a car you are typically a few seconds away from a crash at any moment. If something suddenly in front of you (too close to simply brake) then you have to make a split second descision between crashing into it and swerving. Swerving carries the risk of crashing into oncoming traffic.
Can an auto-driving car tell the difference between a child running out in the road (for which swerving would be the lesser evil) and a plastic bag blowing into the road (where crashing into it would be the lesser evil)?
Re:Never gonna happen. (Score:2)
Racing is based on human skill. If you automate it then it's no longer racing.
Re:Never gonna happen. (Score:2)
Racing is based on human skill. If you automate it then it's no longer racing.
What about the skill of the programmers and engineering team? Car racing used to be about the car too, now it is so regulated that it is only the driver, but maybe that will change again.
Re:Never gonna happen. (Score:2)
The difficulty of everyday driving is the "predictability" of what will happen on the road. Too many humans are unpredictable on how they will handle a given situation.
Case 1: "the polite driver". This driver will give me the right of way when I shouldn't have it. They may stop and let me make a left turn in front of them at an unmarked crossing when they should just proceed and I will wait until a clear spot. Drives me nuts when people do that as it may cause an accident with a driver behind them not expecting what they will do.
Case 2: "the impatient driver". This guy will make be behind me and make a left turn around me because he thinks I am not moving fast enough. I saw one guy nearly lose the back meter of his car to an oncoming motorist.
Case 3: "the don't know where I am going driver". This guy will suddenly realize that his turn was 10 feet back and make it anyway.
Automation should smooth out all those kinks and make it predictable....the obvious problem is the transition period between mostly human and mostly automated....perhaps designated lanes for automation.
Re:Never gonna happen. (Score:4, Interesting)
The difficulty of everyday driving is the "predictability" of what will happen on the road. Too many humans are unpredictable on how they will handle a given situation.
Case 1: "the polite driver". This driver will give me the right of way when I shouldn't have it. They may stop and let me make a left turn in front of them at an unmarked crossing when they should just proceed and I will wait until a clear spot. Drives me nuts when people do that as it may cause an accident with a driver behind them not expecting what they will do.
Case 2: "the impatient driver". This guy will make be behind me and make a left turn around me because he thinks I am not moving fast enough. I saw one guy nearly lose the back meter of his car to an oncoming motorist.
Case 3: "the don't know where I am going driver". This guy will suddenly realize that his turn was 10 feet back and make it anyway.
Automation should smooth out all those kinks and make it predictable....the obvious problem is the transition period between mostly human and mostly automated....perhaps designated lanes for automation.
I can see all this being automated. We can have a "polite mode", a "quick mode" and a "quick route change" option.
Re:Never gonna happen. (Score:3)
I can't see everyday driving being automated because there will always be a scenario that isn't covered, an odd bug that no-one expected, or mechanical failure that the computer can't compensate for in a sensible way.
If most people thought like you do, then we won't even have people driving, as your logic applies to human drivers just as well (and pretty much every single machinery or power tool also):
"I can't see everyday driving being allowed because there will always be a scenario that isn't expected, an odd failure that no-one thought of, or mechanical failure that the driver can't compensate for in a sensible way."
Fortunately, most humans are fine with taking risks as long as the benefit is clear. And in 20 years, I will bet that automated cars will be safer than human drivers.
Re:Never gonna happen. (Score:4, Informative)
When I once was stopped at a red light and a fire truck stopped behind me and started honking I hesitated for a while even though there was no other traffic and then decided to drive despite the red light. In order for a computer to handle that kind of situation properly, there must be a legislative clarification since otherwise the computer would handle it poorly by obeying the red light.
When autonomous cars become common, emergency vehicles will simply tell vehicles in its way to make way ahead of its arrival. There are many ways to accomplish this, most simply by automatically uploading its target location and route to all major mapping services, so those services will then notify the autonomous vehicles to make room.
With the majority of the cars programmed to cooperate and communicate, the roads can be hugely more efficient than it is now.
Re:Never gonna happen. (Score:5, Informative)
Ironically racing is probably a better defined operating environment so easier to successfully automate.
I work for a company that automates vehicles (ASI). We specifically target controlled operating environments like vehicle proving grounds, mines, and commercial harvesting operations. These places all have one thing in common: ten foot fences (aka, no toddlers in the vicinity).
The biggest struggle we have had is obstacle detection; it only works at distances less than 50m. The various vision devices aren't accurate enough beyond that range (or get lost in smoke, fog, dust, shakiness, etc.) And differentiating small objects (aka, 20cm cube) from standard terrain is neigh impossible with current technologies. The algorithms used to process that information can't run in real-time on embedded hardware.
I'm excited for a lot of recent progress in electronic vehicle control. Look for your favorite auto-manufacturer to introduce electronically controlled steering, transmissions, and throttle over the next few years. The pedals, knobs, and wheels will soon be fancy computer joysticks.
Re:Never gonna happen. (Score:3)
Re:Never gonna happen. (Score:3)
Commercial flight systems on aircraft must be so thoroughly tested, yet we still have pilots to take over if the machine fails. How would that work with cars? If you have to be there paying close attention in case you need to take over, doesn't that negate the purpose of the automated system in the first place?
Cars allow for many many failures. So many so that they have a dedicated "breakdown lane" on many roads. An airplane that suffers a failure at 30,000 ft is likely going to have a bad result. Having the automated car set up in a manner that it won't drive into situations it gauges to be high risk, and it offers control to the driver or pulls over to the next safe location and stops isn't hard. Once you've programmed a self-driving car, some safety fail backs isn't hard.
Re: Never gonna happen. (Score:3)
actually no they can't.
There is no drone, or automated plane that can taxi, take off, fly and land.
There's even an open platform that can do all those things..
http://plane.ardupilot.com/ [ardupilot.com]
http://diydrones.com/ [diydrones.com]
Re:Never gonna happen. (Score:2)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
I tend to find that people who define themselves as the latter category are more or less also in the former category. Offense is not the best defense when driving.
yep...practically 'never' (Score:2)
I voted 'never'...for me the key word was 'most'
Most usually means >51%-66% at least...some say a simple majority counts as 'most'...others 2/3 majority....
Either way...I just don't see it happening unless human behavior changes in ways we cannot predict or imagine
If you count different times/types of driving, and break it out in City vs Highway & then compare 'AI' usage there would be some obvious predictable tendencies...and in that way, **maybe** in 50 years...if driving habits change dramatically and road systems are updated...them people might do enough highway driving (read: driving on AI) to balance it out...
Here are 3 other reasons I voted 'never'
> People like driving
> AI is not nearly as capable as its proponents imagine
> The legal liability issue is daunting (just think of the current auto accident liability laws...for **human** drivers & how they vary by state)
The tech will definitely get developed, but I see mostly incremental change in production models.
In cars, the real advance comes when we move off of fossil fuels.
Re:yep...practically 'never' (Score:2)
Re:yep...practically 'never' (Score:3)
The most likely early adopters are old people who have lost the ability to drive safely, but still want to get around.
Another target audience should be young males, who have to pay high insurance rates. A car that is entirely automated, with manual driving limited to speeds below 5 mph, should cost about $10/year to insure, not $1000. It seems reasonable that the entire leech-industry that is automobile insurance would be destroyed by automated cars. For that matter, licensing should go away too; another advance for freedom.
Re:yep...practically 'never' (Score:2)
Re:Never gonna happen. (Score:5, Informative)
Re:Never gonna happen. (Score:3)
Re:Never gonna happen. (Score:3)
Perhaps a few of the less competent might opt for it but us race car drivers will never stand for it.
Your carpool companions called, they want the the blood to return to their white, white knuckles.
Re: Never gonna happen. (Score:2)
I also don't see it happening. Cities and counties make truckloads of cash off of tickets and red light cameras. Plus it gives cops the ability to search people for prohibited plants. With driverless cars they wouldn't get ticket revenue, be able to use civil asset forfeiture or easily fill up the prisons with stoners.
When the control code is 100% bug free (Score:3)
Exhaustive Testing is Impossible.
Re:When the control code is 100% bug free (Score:2, Insightful)
We don't need 100% bug free and "exhaustively tested" to be better than even a very skilled human..... ....let alone an idiot that's posting to facebook while driving home from work in heavy traffic.
I still voted for "pry it from my cold, dead hand" though
Re:When the control code is 100% bug free (Score:2)
Yeah, when tttonyyy is bug free, we'll let him drive.
Re:When the control code is 100% bug free (Score:5, Insightful)
I don't think the semi-inebriated drivers showing up on the roads here in Austin for ACL weekend are have been tested exhaustively either.
One of the things about computers... if someone gets in front of an autopiloted car and randomly slams on brakes, the autopilot adjusts speed up and down. Do that to a person, after a bit, someone is going to be run off the road. Road rage isn't an issue with a computer, nor is alcohol, texting, marijuana, bath salts, Molly, LSD, PCP, Ativan, or whatever substances a person is on.
Don't forget what can be done with automated vehicles that can't be done now. Four way intersections where cars can move at speed, being slowed down or sped up slightly to get one group across before the next group hits.
Pedestrians? Punch the crosswalk signal, and the vehicles WILL stop. No slowing down and honking, no beer bottles thrown at the person crossing.
Parking? Vehicles can park themselves a ways away.
Oil change time? The vehicle goes off and takes care of it in the middle the night, and is back before work.
Moving and have two cars? Toss your crap into one car and tell it to go to your new residence, where someone there can unload it. That way, only the large furniture pieces need the U-haul truck.
There are just so many issues and bottlenecks that would be eased by removing the person out of the driving equation.
Re:When the control code is 100% bug free (Score:2)
Pedestrians? Punch the crosswalk signal, and the vehicles WILL stop. No slowing down and honking, no beer bottles thrown at the person crossing.
This won't happen until there are no people in those cars. Yes, the AI won't throw the beer bottle, but the (possibly drunk, as he is not driving) passenger might.
Also, the "press button to stop cars" system is easily abused if not carefully set (a quite long minimal time between stops), that is why it is not that common in my country.
Oil change time? The vehicle goes off and takes care of it in the middle the night, and is back before work.
Will it go to my neighbor who is a mechanic or only the Authorized Service Center so I could pay 10x the amount I would if I just went to the neighbor?
Modern cars have enough DRM as it is, no need for more. At least with my classic car, I can fix it myself or go to one of the many mechanics who can do it for me, no DRM at all (no computers too).
Moving and have two cars? Toss your crap into one car and tell it to go to your new residence, where someone there can unload it.
That sounds nice. Not for moving (that is a PITA anyway), but for going to the mechanic - no need for a second person to follow me and drive me home, I could just set my backup car to go there automatically and then I could use it to go home.
Re:When the control code is 100% bug free (Score:2)
Modern cars have enough DRM as it is, no need for more.
Agreed. Unfortunately, I think it's quite likely that corporations (and governments) will find a way to turn this technology against us in a lot of ways.
Re:When the control code is 100% bug free (Score:2)
100% bug free? That's impossible. What has 100% bug free? Prove it please.
Re:When it's mandated (Score:2)
The government doesn't care if it's 100% bug free and you can't sue them and they can protect the manufacturers from suits.
That's what Alaska thought.
And they just got sued over ignoring problems with climate change.
Never underestimate how much damage America's lawyers can do to your finely tuned theories.
Missing option: (Score:5, Insightful)
Not nearly as soon as I'd like.
Re:Missing option: (Score:2)
I thought another missing option was put railway everywhere.
More of a legal issue than a tech one. (Score:2)
The technology is getting better rapidly, but until someone can legally flop wasted into their back seat, at 4 AM, shout "Take me home!" and drunk text their ex-girlfriend like they currently can in a taxi, it's not going to get much traction.
So I'd say we're less than 10 years out from a tech perspective, and n years out from a legal one.
Re:More of a legal issue than a tech one. (Score:5, Interesting)
The technology is getting better rapidly, but until someone can legally flop wasted into their back seat, at 4 AM, shout "Take me home!" and drunk text their ex-girlfriend like they currently can in a taxi, it's not going to get much traction.
Reminds me I was once told that the horse that pulled my grandmother's and her sister's wagon when they were in their teens and twenties could do that. I thought that was cool until she mentioned it was handy on dates to be able to give the horse the reigns. That was TMI.
Re:More of a legal issue than a tech one. (Score:3)
Voted 20 (Score:3)
I'd love to see it within 10, but I think 20 is more realistic.
I remember just before I got my license (~17 years ago) I thought driving was the most awesome thing I could imagine. Now it's a mindless chore that I accept as necessary but I certainly don't like doing it. My daily commute to work is only 5 minutes so that's fine, but fairly often I take trips that consist of 4 to 12 hours of driving. I can think of nothing more amazing than just being able to put that address into my car and then chill out on my laptop for the duration.
I also this could also do a LOT to reduce drunk driving. When you can just climb into the car and tell it to take you home we'll have a lot less accidents due to intoxication.
Re:Voted 20 (Score:2)
I remember driving as looking like great fun when I was a kid. As I grew up, I found I liked to drive. I had a 74 Chevy Nova which was great fun, an 84 Pontiac Fiero, which was great fun, and a 300ZX which was also great fun. After the 300ZX, I got a Ranger pickup which was practical for doing stuff around the house, but also got a motorcycle, which was great fun. I'm on my second Ranger and my 6th motorcycle (Hayabusa) and am currently planning my trip for next year. Roughly 10,000 miles to Alaska and the Arctic Circle.
It's still great fun. :)
[John]
Consider existing cars (Score:2)
Actually, even 20 years seems overly optimistic. Some quick web searching [usatoday.com] indicates that the average age of a car in the US is 11 years. Assuming the total volume remains reasonably unchanged, that would mean that it would take around a decade* to reach 50% even if all cars sold from today were self-driving.
If we assume it takes another 10 years for fully autonomous cars to be commercially available (an legal), it is still unlikely that zero "ordinary" cars would be sold from then on. So I don't really see how any alternative lower than 30 years could be realistic. Remember, the question is when "most" driving is autonomous, not some driving.
Personally, I think we'll see a gradual transition, with an increasing degree of automation, over several decades. We are approaching the intermediate state where cars are more like horses; you still steer them, but they refuse to run off cliffs or into objects.
(* The time obviously depends on the shape of the distribution, which probably isn't uniform, but probably close enough for the sake of this argument.)
Re:Consider existing cars (Score:2)
Prying the steering wheel (Score:5, Funny)
You'll have to pry the steering wheel out of my cold, dead hands
Why would I do that when I can have the car accident investigators and the medical examiner do that for you?
A long time. (Score:2)
The technology is already on the roads. But aside from the normal amount of time necessary for technology adoption, this also faces significant legal hurdles. There's the big question of liability, of course, and we're starting to deal with that now. But the legal issues will get worse before they get better--self-driving cars are still experimental enough that they aren't a huge political battlefield yet.
Once they develop a bit more, many people will have safety and NIMBY concerns--even if they're much safer than human drivers, many won't want it around without 100% safety. Not to mention other lobby groups--cab drivers, truck drivers, and so forth are heavily unionized, and will use their political sway to oppose this technology as much as possible, since it will (eventually) take their jobs in a very real, direct sense.
Re:A long time. (Score:3)
Not to mention other lobby groups--cab drivers, truck drivers, and so forth are heavily unionized, and will use their political sway to oppose this technology as much as possible, since it will (eventually) take their jobs in a very real, direct sense.
You left out a big one - governments that depend on revenues generated through traffic enforcement.OOH, and the companies that sell red-light camera systems!
We'll get there gradually, but we'll get there. (Score:2)
When will I be able to go out and buy a Honda Civic EX with this technology? That'll be a much harder market to get into and the technology needs to be that much more robust, commodity and modular.
Re:We'll get there gradually, but we'll get there. (Score:5, Insightful)
Actually it's not Honda and Toyota et al -- it's the Insurance Companies. As soon as this technology becomes approved and certified by the NTSB here in the States and it's equivalent elsewhere, you can be sure that the insurance companies will offer steeply discounted rates to drivers who have fully automated cars since they can sue the Automobile Manufacturer or Software Manufacturer in the event of an accident by an automated vehicle rather than collecting from another insurer or individual.
When people see that they can get a good break on Insurance if they buy an automated car, the mainstream manufacturers will be tripping over themselves to get their own automated vehicles out to the mainstream market to capitalize on that and to be competitive.
Re:We'll get there gradually, but we'll get there. (Score:4, Interesting)
It may not all be the high end guys. I remember looking at a self-parking car. Then I remembered an old French brand (Citroen) that had a compact car that had two pairs of small wheels mounted sideways. A flip of a button, the car was lifted onto the wheels. A move of a slider, the vehicle moved left or right directly. The Lexus parallel parking capability is impressive, but the little French car being able to just stop, move 90 degrees into a space without requiring any movement forward/reverse and park was more practical.
Re:We'll get there gradually, but we'll get there. (Score:3)
Very true. There are some advancements which are hitting cars that were not thought of recently.
For example, Fiat has an automatically shifted manual. Instead of using a torque converter and the hydraulics, they have a fully manual transmission with clutch, all computer controlled. This wasn't possible in the past just due to computers not being advanced enough to automatically throw the clutch out, shift, pop it back in. Now, with this technology, the conventional hydraulic automatic with its loss of power is now pointless compared to the classic manual transmission.
As an aside, this technology gives the best of both worlds. Someone who knows how to shift can do it (rather than the "pseudo-shifting" that is on a lot of auto transmissions), and if not, pop it into "D" and let the computer handle the gear-grinding.
What country? (Score:5, Insightful)
Re:What country? (Score:2)
Not gonna happen (Score:3, Insightful)
I'd say never. Why? Because driving on random roads and locations in varying conditions requires *intelligence*.
Most attempts at computer-driven cars so far basically just put cars on an electronic track and throw in a little bit of sensor data and pre-programmed logic.
Yes, 99.9% of driving is just dull "stay between the lines" but it's that .1% of unexpected shit that requires a brain. (not that too many people these days have those)
What happens when some crap falls off a truck and the sensors don't quite pick it up? What happens when someone runs out in the road in front of it? Can it tell the difference between a rectangular dark black flat piece of dangerous metal on the road and dark shadow from an overhead roadway sign? Would a computer even realize you might need to drive off the road in the event of an emergency or special circumstances? There are an almost infinite number of possibilities that you just can't pre-program in or plan for.
Throwing control back to the user probably wouldn't work well in practice, because once a car can drive itself sometimes, people will expect it to drive itself ALL the time.
And when something does go wrong, who will people sue? Right now it is the individual driver of the vehicle. I think car manufacturers would prefer to avoid the risk of being sued and leave responsibility where it is.
So what if someone creates a computer that is really intelligent enough to do that? Well, I'd suggest getting out of its way as it kills all humans. :P
Re:Not gonna happen (Score:5, Insightful)
Secondly, the hypothetical situation you describe is exceedingly rare and I suspect human drivers would nominally fair equally bad at avoiding the situation.
No matter how good your automomous car is, if you're driving along and a 1000lb piece of cast iron pipe tumbles off the back of a flatbed, you're fucked. Possibly having a car with reaction times that exceed a human might save your life. Your autonomous car is going to be not tailgating the semi to begin with and will be going the speed limit or lower depending on the traffic conditions.
Thirdly, I can only imagine that these cars will be recording all the telemetry and video they possibly can. To complete my scenario, when a piece of cast iron pipe falls off the truck and lands in the road and the lidar system doesn't properly identify it and the car runs into it, the insurance company is going to analyze the telemetry and the dashboard video and they'll sue the truck driver for not properly securing their load. They possibly would go after the car maker for a faulty system, but more likely the car companies just going to want the telemetry so they can improve the system.
As for knowing the difference between a shadow and a piece of black metal, these systems are currently using lidar... so I don't see this as much of an issue.
Also your scenario of what happens when someone runs out in front of the car... Mercedes already has an automatic breaking system.
Re:Not gonna happen (Score:3)
When the car doesn't know what to do, it squaks at the driver and the person takes over. Understandably that has it's own issues.
Actually, if this limitation is not ironed out, it would seriously slow down the adoption, here's why:
1. If I buy a self driving car, I expect it to drive itself while I am doing something else (or am just drunk).
2. Even if I am sober, I would not be paying full attention to the road (because I am not driving), so when the car beeps that I should take over, my initial reaction will be really slow.
3. If I have to pay full attention to the road to be able to take over at a moments notice, I might as well be actually driving (it would be easier to pay full attention that way).
Re:Not gonna happen (Score:2)
Re:Not gonna happen (Score:3)
Current robotic cars only have trouble with this in snowy conditions;even then, if it is a large chunk that "sits up high" on the road the lidar can detect it.
Unfortunately snowy conditions lead to other huge, currently unresolved problems in the visual navigation/reaction part of the car's logic.
The Question is Ambiguous (Score:3)
"Most driving" could mean "most vehicles" or "most passengers". For example, if there are 2 human-driven cars and 1 computer-driven bus, then most vehicular passengers are being driven by a computer, yet most passenger vehicles are being driven by a human.
Such an ambiguously worded question deserves an equally ambiguous answer: "Quite a few years from now."
I picked 20 years (Score:3)
Of course there is a lot to be said if these cars can even function to begin with. Anyone who's had their Garmin(Gremlin) tell them to drive down the wrong way on a road knows that even for route planning as roads change(something the self driving car needs), there is still much to be done.
Re:I picked 20 years (Score:2)
I really hope it could work though. Man it'd be nice to not have to pay car insurance by law. That stuff is a scam.
Sorry to be the pessimist, but ... (Score:2)
I am not so sure that we are going to have the infrastructure to support automobiles about the time when the control tech is good enough for widespread deployment.
Only recently, peak oil came and went and worldwide economic growth has plateaued. Even if we cut emissions of greenhouse gases completely right now, there is still enough of it in the atmosphere to cause climate change that is going to manifest itself in famine, "act of god" disasters, economic turmoil, civil unrest and wars across the world.
It was a beautiful dream. That is what it was.
Re:Sorry to be the pessimist, but ... (Score:2)
Depends of the country (Score:2)
Who is going to take us there ... Google? HA! (Score:2)
Yea right using the latest and greatest technology they cant even get a basic map right.
I am the designer of a prototype project which requires accurate machined parts, so one day I get hooked up with this machinist that does stuff for our state tech university, does great fast and cheap work! Problem is he is about 3 hours away from our offices so I take a day trek down there to get ~25 units made up for field testing.
His shop is just outside of the college town, which is pretty sizable, and has lots of out of state traffic, so I load up a google map and off I go. Get down there and it's telling me to go on this 2 lane highway for a quarter mile and take a left. The high way is now a 4 lane divided job, there is no left, and at the quarter mile mark there is nothing.
I call the guy up and he says, no its about a mile past the on ramp, take this exit, THEN turn left blah blah blah, sends me an email with a scan out of a cheap ass gas station map, which is correct. When I got there I asked "Wane, how long has that been a 4 lane bypass" ... "oh I dont know, 96-97?"
Whats a google van gong to do, button hook me into a concrete barrier at 65Mph based on a map 17 fucking years out of date?
Nah I am good, been driving for as long as google's map has been out of date and only once been in an accident where some numbnuts twice my age rear ended me.
Re:Who is going to take us there ... Google? HA! (Score:3)
Why I picked 50+ years? Look at hybrids & the (Score:2)
Lexus LS 460 in 2006 had the self parking car. 7 years ago.
I just got a new car. I would have liked to have both these features, but both were at a price premium I could not practically afford. Mind you, there are many people that buy over what they can afford not withstanding what they can practically afford.
Still many people are not getting these features, though they are becoming more common now.
Self driving cars are just starting to come out now. Previous history would suggest that it will be 10-20 year until it becomes practical for some people to afford these cars. Lets say there is another at least 10 years until these become affordable to most people. You will get quite a few people at that 30 years that will have just got a car with out self-driving, but would have liked to have gotten them, so they won't have one until about 50 years from now, which will probably be the practical earliest point where MOST people as the question states will have self-driving cars.
You will also have people that will refuse to get a car that drives itself because they want to drive, or they do not trust the car to drive its self. Thus I see the earliest tipping point more likely being 55 years out. There are all sorts of things that can happen between now and then that could push this back further.
Already having a car, Cost, and Trust I think will be key factors that make it a long time until MOST of the people have self-driving cars.
I do think at some point most people will have self-driving cars. I think it is a long way off. Could it be never? There could be accidents that cause government to outlaw self driving cars, so I wouldn't count out the people voting never.
I do think the people voting 20 years are delusional. Time and time again we see new technology and it doesn't start being affordable until at least 10 years, more often than not 20 years. Even it it were to start becoming affordable in 10 years, it will be at least 5 years more to be affordable to most people. Even if a person only has their car for 10 years, that means most people won't have the new tech until maybe in 24 to 25 years. And taxi companies aren't going to lower fairs because they got rid of the people driving.
So next 20 years, no way. 30 years, maybe, but only if government mandates it. 40 years its possible, but doubt probable. 50 years is being kind of realistic, but more likely more than that.
Too soon unfortunately. (Score:2)
Cars already have too many *improvements* already. Self-parking, automatic transmissions, traction control, ABS, blind spot warning, so on.
Kinda removes the need to be able to drive, so when there's an emergency the driver won't know how to react.
Re:Too soon unfortunately. (Score:3)
Comment removed (Score:2)
Re:It won't be until... (Score:2)
On a side note, if you want to own a pickup truck, but don't want anybody to borrow it, get one with a manual transmission. The odds of any of your friends knowing how to drive it would be pretty low. :)
Good luck finding a full sized pickup truck with one. I have started looking and they seem to be exceedingly rare and it doesn't seem that anyone is making one (I haven't looked at all manufacturers yet so I might be wrong).
Re:It won't be until... (Score:2)
According to this article, [trucktrend.com] The only full size trucks that come with a manual option are the 2500 and 3500 series Dodge Rams.
Re:It won't be until... (Score:2)
This is true. But the practical consequence is that, since they're my friends, I end up doing the driving when they need a truck.
Re:It won't be until... (Score:2)
or they'll lie, borrow it and drive like my mother, "fully synchro-smashed in all forward gears"
Re:It won't be until... (Score:2)
Personally I'd prefer having folks start off on dirt bikes, scooters, and low cc motorcycles and work their way up to cars and trucks. It might be a bit safer for those of us who _do_ ride.
[John]
Re:It won't be until... (Score:3)
Anyone who can drive using a manual transmission can drive using an automatic transmission. That means there are fewer people who are able to use a manual transmission. Those people who can't master it are very likely to be less skillful in the mechanics of driving.
Does that make them less safe? I have no idea, I wasn't advancing any theories, simply explaining why someone might use that as a criteria. Your hypothetical also might be correct. However, your earlier statement rejecting any possible connection between the ability to learn how to use a manual transmission and how safe that person is as a driver is overly broad.
Re:It won't be until... (Score:2)
nonsense, that's only about 7% of vehicles in the USA. the big historic advantage has disappeared, the fuel economy is the same nowadays.
Re:It won't be until... (Score:2)
Re:Autonomous Driving?? (Score:2)
we have flying cars and jet packs but that little business of folding into a briefcase after the commute is still in R&D
Re:I work to much (Score:2)
Is that like a Sorny or Panaphonic TV?
No, a Porshe is a deluxe, high end car, with a safety glass windshield to prevent blowaway. And don't get my wrong, those TVs are all superior machines, but a better TV analogy would be a Carnivale.
Re:I work to much (Score:2)
Re:Technology and acceptance (Score:2)
Doesn't matter how fast the technology advances, nor how safe it is - the government makes money from issuing moving violations, a cash cow that automated cars would effectively destroy.
In other words, they aren't going to kill the goose that lays golden eggs.
Re:Technology and acceptance (Score:2)
Re:Young Drivers (Score:2)
How much worse could a computer driven car be? If anything, it would be better at judging speeds and distances; far better than a human.
The computer would probably be very good at anything a human would be bad at (precise speeds, distances, reaction time etc), but it may be bad at stuff the humans are good at (creative solutions for problems - let's say you just found out that your brakes no longer work, you are currently going at 60km/h and would really like to stop, what do you do?).
Also, when a bug causes an accident, it would most likely be such that any human could have avoided it (and not avoided the accidents that the computer avoided), it could also be that if some bug causes an accident there would be many accidents (caused by cars with the same software), though then all of them could be patched.
Re:Young Drivers (Score:2)
How much worse could a computer driven car be? If anything, it would be better at judging speeds and distances; far better than a human.
The computer would probably be very good at anything a human would be bad at (precise speeds, distances, reaction time etc), but it may be bad at stuff the humans are good at (creative solutions for problems - let's say you just found out that your brakes no longer work, you are currently going at 60km/h and would really like to stop, what do you do?).
I pulled over a little so my right tires were crunching through the snow on the shoulder of the road, which slowed me down quite nicely. Then pulled into a parking lot, turned around, and limped a mile back home to park it until I could fix it.
Re:Young Drivers (Score:2)
And if there is no snow you can downshift (engine braking) and maybe even turn off the engine to stop.
The question is, will the AI figure out how to slow down and stop in this case?
Re:Young Drivers (Score:2)
Oddly enough, that's what I did when the brakes went out on the car I had before the first one I mentioned. But then I was 30 miles from home, in a medium-size city. Had to drive home using the hand brake.
Thankfully, my cars since then haven't been crumbling death traps. Oh wait, I drove a POS Ford Taurus for a year. The back bumper fell off one day when my mom backed into a vehicle at church. We replaced it with a wood plank, spray painted silver to match the grey car. That time the brakes worked well, but Mom's legs didn't. :^)
I think I just found the real danger of self-driving cars. They will have a short life-span, until needed systems are not reliable enough to support AI control.
Re:"Some"? 10 years. "A high minority?" 20 years. (Score:2)