PDA

View Full Version : Imagine Computer-driven Vehicles



Pages : [1] 2 3

Inclusa
2012-Mar-29, 03:54 AM
Due to slow reaction time, I refuse to be a human driver; still, North Americans often place high regards on ability to drive, and public
transit is not exactly well-developed.

The only way I may "drive" may be a computer-driven vehicle, but I'm not too sure how far are we from it.

danscope
2012-Mar-29, 04:11 AM
You can not delegate absolute authority and responsibility for operating a motor vehicle . YOU are the pilot in command.There is no computer which
can replace you. Period.

Celestial Mechanic
2012-Mar-29, 04:33 AM
And which operating system would you trust to drive your vehicle for you? Microsoft? It would bring a whole new meaning to the phrase "Blue Screen of Death". :eek:

Inclusa
2012-Mar-29, 06:33 AM
Ok, the realistic alternative is still taking public transit? Many North Americans often don't realize "pushing" everyone to drive is dangerous.

WaxRubiks
2012-Mar-29, 06:37 AM
computer driven cars would create a revival in country pub in the UK.

Jens
2012-Mar-29, 07:58 AM
You can not delegate absolute authority and responsibility for operating a motor vehicle . YOU are the pilot in command.There is no computer which
can replace you. Period.

I agree with your skepticism regarding space travel, but in this case I don't. I don't think that computer-driven cars are all that far away. And I think they will make roads safer, and as a bonus will lead to a revival of country pubs! Seriously. I have been in computer-driven trains and I would be happy to ride in a computer driven car. Not with MS Windows at the OS, of course.

Van Rijn
2012-Mar-29, 08:16 AM
And which operating system would you trust to drive your vehicle for you? Microsoft? It would bring a whole new meaning to the phrase "Blue Screen of Death". :eek:

But people already die all the time on the road. Obviously, you wouldn't switch over to automated driving until it was shown to have no more risk than human drivers. Automation is about the only path I see with a real possibility of significantly reducing traffic deaths and injuries.

geonuc
2012-Mar-29, 08:36 AM
Ok, the realistic alternative is still taking public transit? Many North Americans often don't realize "pushing" everyone to drive is dangerous.

It is not just 'North Americans' who drive automobiles. There are a lot of dangerous things in life and a lot of ways to minimize the risks to acceptable levels. The realistic alternatives to using a computer-controlled vehicle (which is not an option, yet) are to use public transportation as you say, but also to learn to drive non-computer-controlled vehicles. You could also use employ the Sheldon Cooper option: rely on others to drive you around.

novaderrik
2012-Mar-29, 10:51 AM
i would hate not being in control of my car- it's relaxing to get in a car and go for a drive thru the countryside, not knowing where you are going, how long it will take to get there, or what you will see or what kind of people you will meet along the way..

as to why public transportation hasn't really caught on in the USA like it has in Europe- it's a matter of geography and culture. on the geography side of things, this country is BIG compared to European countries and we tend to live farther from where we work.. on the culture side of things we tend to be more independent minded and taking a train or a bus to get everywhere just doesn't fit in with that mindset.

ddubs
2012-Mar-29, 02:06 PM
The problem with humans driving is there is very little communication between vehicles on the road. You get two-way comm which is brake/turn signals (which can be easily missed) and one-way which are your mirrors, the latter being more reactionary to what is currently happening. A computer driven car would have the advantage of being on a network of sorts with other vehicles on the road. They can communicate back and forth/verify safe maneuvers/very accurately estimate distances around the entire perimeter of the vehicle. The computer NEVER takes its "eyes" off the road - unless there is a catastrophic failure of the system, however that's probably its own topic. Living in the US, I'd love to have fully computer automated driving and hopefully its not too far off.

WaxRubiks
2012-Mar-29, 03:00 PM
one problem is the driving on small roads, where you have to pull in to let the other driver by......things like that probably require an intelligent, perhaps conscious, system.

cjameshuff
2012-Mar-29, 03:28 PM
Full automation would pretty much require instrumentation of the roads themselves (with RFID tags, tracking targets, etc) so the computer knows where it is and what it can do. This might eventually be an option in and near cities and on highways, but would not be practical to do for all the roads, and maintenance costs must be considered as well as transitions between roads equipped for automatic driving and those without...you'd need things like emergency auto-parking areas for use when the human driver isn't responding. It might help with some of the major congestion problem areas once a large enough percentage of cars are equipped.

An interesting intermediate system I've seen proposed is to have vehicles automatically follow each other, forming up behind a vehicle with a professional driver (who themselves might have computer assistance for things like coordinating with traffic lights to minimize disruptions in traffic flow). Even better would be to have two drivers watching the road, switching control of the vehicle once an hour or so to minimize fatigue.

danscope
2012-Mar-29, 05:44 PM
Answer: Deer , Child , Rock , black ice etc. You have to react to these. The computer???? Shrug .

cjameshuff
2012-Mar-29, 06:08 PM
Answer: Deer , Child , Rock , black ice etc. You have to react to these. The computer???? Shrug .

Any car that could reliably follow roads and handle the various bizarre and misleading curves and intersections encountered would have everything needed to react to the above hazards better than any human driver could. Reacting to such things is pretty trivial in comparison to navigating the roads.

ddubs
2012-Mar-29, 06:13 PM
I think that's why the largest benefit to computer driven vehicles would be on expressways around crowded cities. You shouldn't run into too many of these types of obstacles. Once you are off the expressway then you can take manual control of the vehicle.

Another thought, maybe the computers will have IR camera's and tracking software to determine if an obstacle is "alive/moving" and navigate around it. That would end the "lady who swerved to miss a squirrel and hit a kid on the sidewalk" type stories.

orionjim
2012-Mar-29, 06:32 PM
Google has been working on developing a driverless car and won a two million grand prize for their work from the US Department of Defense.

Currently Nevada is the only state that allows a driverless car to be driven and they can only be driven in specified areas set up for testing. This is where Google has been testing their vehicle.

This is a link that describes Google’s work:

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

Note; if you read the article you will see that the car has driven 1,000 miles without any human intervention; and has logged about 140,000 miles with some human intervention.

Also it was involved in one accident, but a human was driving at the time.

I know that GM and Nissan have been working on driverless vehicles, and I am fairly certain other car manufactures are also working on developing them.

NEOWatcher
2012-Mar-29, 06:58 PM
Google has been working on...
I don't want to put google down, because thier work is impressive. I just think that there needs to have a few statistics put into context.

First off, this is only a test, so things definitely aren't at the point of real world situations yet.


Currently Nevada is the only state that allows a driverless car to be driven and they can only be driven in specified areas set up for testing.
A large part of Nevada is flat wide open non-populated areas.
I would be interested in knowing what these specified areas are like.
Another issue is the climate. I'd love to see what happens during rain or snow or heavy fog.



Note; if you read the article you will see that the car has driven 1,000 miles without any human intervention; and has logged about 140,000 miles with some human intervention.
I'd like to see some profiles on what that human intervention is.


Also it was involved in one accident, but a human was driving at the time.
Yes; humans are fallible. But; giving them trust in technology makes them fallible in another way.
I've heard this about airliners. Pilots start to rely on the technology and are letting thier piloting skills lapse.
This situation could have been caused because the technology was there rather than the human in complete control.

I think for the technology to work, the human factor would have to be removed from all the cars on the road.
The other thing is people can't maintain thier cars now.
And; the more technology there is, the more likely you completely lose the use of your vehicle when there is even the slightest malfunction.

danscope
2012-Mar-29, 07:03 PM
Computer driven cars are ridiculous. I think that somewhere there is an engineer who wants to make a push-broom and put an integrated circuit on it so they can monitor the amount of strokes per minute..... and sell it to the Army at $500
a piece.
If you want to use computers, use them on a monorail with a proximity sensor with back-ups. Put your personal car
on the monorail and select your exit in advance and sit back or sleep untill you get there. Now you are talking.
Charge your car off the rail while you are at it.


" There are ways of telling if one is a witch." Sir Bevedere

" Are there ? " an ignorant peasant

orionjim
2012-Mar-29, 08:17 PM
...

I'd like to see some profiles on what that human intervention is.


...

I found the following quote at:
http://www.inglewoodtodaynews.com/?p=4190


Humans won’t relinquish total control of their driverless cars. Google has a built in mechanism in the cars that switches from a human-controlled mode to self-driving. The technology guides the car and tells it where to stop. It also gathers information on its exact location and the direction to proceed. In situations best judged by a human, the car’s landing strip allows a human driver to decide acceptable parking places for the vehicle.

Solfe
2012-Mar-29, 08:23 PM
I drive almost every time my family takes a trip. When I don't drive I use public transit because I am not travelling with anyone I know. Typically I sit in the back and read a book, never looking up. Not ever.

Very rarely, my wife will drive. She hates it. One time my children found my wife driving and me in the passenger seat to be so novel, they asked:

"Dad, why aren't you driving?"

My answer was equally novel. "Because the imaginary pedals don't work."

I would hate being in a car that drives itself. I would feel that I was seconds from being an organ donor.

Ara Pacis
2012-Mar-29, 08:54 PM
I disagree with Danscope and some others. Computer systems do react faster than humans, the issue is that you just need to give it enough sensor input to make the correct decision. A visible light camera can keep an eye on road markers better than humans. This doesn't need to be a one-eye-sees-all system but could have subsystems where there is a camera at each corner of the car to take note of lane markings and alert the main computer when there are deviations, this could be enhanced with simpler optical sensors that can look straight down and can alert the computer when it passes over a lane marking with a different albedo than the road surface. GPS, a magnetic compass, and internal navigation databases (preloaded and learned) can allow the computer to discern when lane markings deviate and why. One or two main cameras can keep an eye on the road for obstructions including tail lights and headlights of oncoming traffic using optical parallax to discern distance. Near IR lasers and/or RADAR can scan the road ahead for obstructions by seeing a laser scan-line gap caused by an elevated obstruction or depression, and may be able to discern information about the hazard based on parallax. Humans can't see at night, but a computer can be fitted with near IR and thermal IR sensors for when visible light is too low or too bright. The vehicle can also be equipped with accelerometers and stress indicators to estimate lateral loads from curves/turns and winds.

Cars can also be made with signalling systems that let other cars estimate distance and aspect changes as well as actively signalling what they are doing or intending to do. This doesn't have to be as privacy-concerning as destination information, but a signal of an impending turn in x-number of feet with a countdown and could be useful. Of course, this could be a target of off-road hackers, so making it an limited angle signal instead of omnidirectional radio might be better, and it should be over-ridden by optical road and traffic hazard scanning, in case of on-road hacked cars.

Multilane freeways might be more difficult instead of less difficult, due to the ability to determine appropriate lane changes and how to anticipate lane changes of other vehicles. This might be resolved by a freeway management system external to the car or special lanes.

Inclement weather, like snow, might be problematic for computers, but it's just as problematic for humans. The computer should be smart enough to recognize when it has visibility problems and alert the driver to take control or pull over and wait for conditions to improve or for the human driver to take control. Then, the computer may be able to assist the driver using all its sensors so that human driver takes agency over decisions to drive in bad conditions, for legal purposes. The computer should also be smart enough to remind the human to clean the optics coverings when road-grime makes it less able to see adequately, and refuse to take control until this, or other mechanical issues are resolved.

See, it's not that difficult to conceptualize if you just think about it.

cjameshuff
2012-Mar-29, 09:15 PM
A visible light camera can keep an eye on road markers better than humans.

And near-IR cameras would see better through fog, dust, etc. Cameras and other sensors could quickly detect obstacles and have the car reacting to them before the human driver has even registered their presence...this is another area that I have read about research into, systems to enhance safety of cars with human drivers.



GPS, a magnetic compass, and internal navigation databases (preloaded and learned) can allow the computer to discern when lane markings deviate and why.

Lane markings are often absent, worn, buried under snow, obsolete, or just plain incorrect, etc. They can not be relied on for navigation, which is why I assumed instrumentation of the roads.



Of course, this could be a target of off-road hackers, so making it an limited angle signal instead of omnidirectional radio might be better, and it should be over-ridden by optical road and traffic hazard scanning, in case of on-road hacked cars.

The bit rate required is not great, a straightforward solution would be to use IR LEDs in the vehicle lights to emit signals that the navigation cameras can pick up. This would make it quite clear which car is transmitting which signal and make interference difficult.

orionjim
2012-Mar-29, 11:44 PM
I just saw this on Google News:

"googles self driving car takes blind man to taco bell"

http://content.usatoday.com/communities/ondeadline/post/2012/03/googles-self-driving-car-takes-blind-man-to-taco-bell/1?csp=34news

danscope
2012-Mar-30, 01:01 AM
Here's a fun scenario: There you are, driving down the street, minding your own business , when here comes..... no, not
a 'CellPhone Zombie' crossing the solid yellow line to take you out head-on; oh no...... it's somebody's bloody
computer-controled car !!!!!!!! You think that engine service light on your dash panel is a pain in the posterior? Just wait untill you are in a full body cast with no feeling from the waist down because of a failed computer.

"You won't get it !" Patrick McGoohan

Jens
2012-Mar-30, 01:27 AM
I've heard this about airliners. Pilots start to rely on the technology and are letting thier piloting skills lapse.


That may be true, but I don't think the number of airline accidents is increasing. I think there were probably more accidents before the advent of autopilots.

Jens
2012-Mar-30, 01:38 AM
one problem is the driving on small roads, where you have to pull in to let the other driver by......things like that probably require an intelligent, perhaps conscious, system.

Any intelligent vehicle would have to have a way for you to instruct it to pull over. For example, suppose a wasp gets inside the car and you need to stop the car to let it out. There would have to be something like an emergency button that would cause the car to stop as quickly as safely possible.

Similarly about the comment about a leisurely drive. Though typing in a destination would undoubtedly be a major way of using it, I assume there would also be controls for "go right at the next light" and things like that.

cjameshuff
2012-Mar-30, 01:46 AM
Here's a fun scenario: There you are, driving down the street, minding your own business , when here comes..... no, not
a 'CellPhone Zombie' crossing the solid yellow line to take you out head-on; oh no...... it's somebody's bloody
computer-controled car !!!!!!!! You think that engine service light on your dash panel is a pain in the posterior? Just wait untill you are in a full body cast with no feeling from the waist down because of a failed computer.

Human drivers in the US had 10.8 million accidents in 2009, with 35.9 thousand deaths (http://www.census.gov/ (http://www.census.gov/compendia/statab/2012/tables/12s1103.pdf)). Engine control computer failures are rather rare in comparison, and they aren't subject to anything like the reliability requirements that would apply to the systems we're talking about. When you consider the fault detection and failsafe systems that would be in place, your scenario just isn't plausible.

Far more likely is for the car to alert the driver and stop the car because they didn't respond fast enough (which is essentially what would happen if the engine control computer failed in a vehicle with a human driver), and then get plowed into by you because you didn't react quickly enough to an unusual obstacle when a computer would have handled the situation just fine.

Van Rijn
2012-Mar-30, 06:20 AM
It is not just 'North Americans' who drive automobiles. There are a lot of dangerous things in life and a lot of ways to minimize the risks to acceptable levels. The realistic alternatives to using a computer-controlled vehicle (which is not an option, yet) are to use public transportation as you say, but also to learn to drive non-computer-controlled vehicles.


I know how to drive a car. I've taken defensive driving courses too. However, that can only help so much because there are a lot of incompetent drivers. Hypothetically, a driver's license could be made much harder to get and keep, and there could be much greater penalties given for driving without a license. However, that could have been done decades ago, but hasn't. It's doubtful that will change any time soon either. On the other hand, here is an emerging development that could get around the issue of incompetent drivers and provides more advantages than just safety. If we can develop computer controlled cars at reasonable cost that are safer than non-computer controlled cars, why not do it?

Jens
2012-Mar-30, 07:10 AM
I have kind of a simple question, but I wonder if any of the people who responded they would not trust a computer-controlled car would be willing to go into space on a rocket.

Cookie
2012-Mar-30, 07:59 AM
How many other rockets will I share the skies with?
How many of those will be within inches of my rocket?
Will they be taking any parallel, or overlapping, paths?

Jens
2012-Mar-30, 08:27 AM
So in other words, you'd feel safer in a rocket going into orbit? Impressive.

geonuc
2012-Mar-30, 09:12 AM
I know how to drive a car. I've taken defensive driving courses too. However, that can only help so much because there are a lot of incompetent drivers. Hypothetically, a driver's license could be made much harder to get and keep, and there could be much greater penalties given for driving without a license. However, that could have been done decades ago, but hasn't. It's doubtful that will change any time soon either. On the other hand, here is an emerging development that could get around the issue of incompetent drivers and provides more advantages than just safety. If we can develop computer controlled cars at reasonable cost that are safer than non-computer controlled cars, why not do it?

Nothing in my post was meant to imply we shouldn't pursue the technology vigorously.

swampyankee
2012-Mar-30, 09:25 AM
And which operating system would you trust to drive your vehicle for you? Microsoft? It would bring a whole new meaning to the phrase "Blue Screen of Death". :eek:

Embedded systems -- like digital flight controls -- don't need operating systems.

swampyankee
2012-Mar-30, 09:42 AM
Here's a fun scenario: There you are, driving down the street, minding your own business , when here comes..... no, not
a 'CellPhone Zombie' crossing the solid yellow line to take you out head-on; oh no...... it's somebody's bloody
computer-controled car !!!!!!!! You think that engine service light on your dash panel is a pain in the posterior? Just wait untill you are in a full body cast with no feeling from the waist down because of a failed computer.

"You won't get it !" Patrick McGoohan

And that is worse than what happens now in what way? Sorry, I've known several people killed or seriously injured by drivers who were neither drunk nor using cell phones. If we're collectively sane, so we have an adequate certification process, computer-controlled cars are going to be safer than human-controlled ones.

HenrikOlsen
2012-Mar-30, 09:50 AM
Computer driven cars are ridiculous.
If you stop and think hard about the situation, it's actually the human driven cars that are the ridiculous part of having computer driven cars on the road.

Computer driven cars stop being ridiculous when they're the only ones on the road.

NEOWatcher
2012-Mar-30, 12:28 PM
That may be true, but I don't think the number of airline accidents is increasing. I think there were probably more accidents before the advent of autopilots.
True; but pilots need continual updates on thier training, the equipment is meticulously maintained, and there are people watching every one of them while they are flying.

Personal vehicles (as they are now) don't have any of that. I just wonder how many people drive around with thier engine light on and don't care.
You also have lax restrictions such as Indiana where renewal is online. No eye exam, no new picture, just a click and pay.

cjameshuff
2012-Mar-30, 12:41 PM
How many other rockets will I share the skies with?
How many of those will be within inches of my rocket?
Will they be taking any parallel, or overlapping, paths?

How do these things make a computer, which can react faster and more precisely than any human, coordinate constantly with the other vehicles in the vicinity, and do so reliably without being subject to conditions like fatigue, distraction, illness, intoxication, inexperience, emotional state, or sheer incompetence, less desirable?

Solfe
2012-Mar-30, 01:23 PM
Now that I think more deeply about the computer driven cars, I suspect there would be huge social issues. I suspect that a properly programmed car could make one immune to tickets. If the car can't break the established programming, then as long as it is in "automatic" mode, you shouldn't be able to do something ticket-able.

I suppose that there could be some very positive social changes enable by automatic cars. Imagine if your personal car could be designated as an ambulance. That would very efficient for getting you to a hospital.

What if instead of an alarm, the car could pilot itself away from someone trying to steal a radio or the car itself. There would have to be a lot of safe guards in place, but it is a neat idea. "Sir, I am sorry. I had to move so that I was not stolen. I will pick you up in five minutes."

Traffic stops could be an option event: "Place your car in self drive mode or pull over." Much safer for police, and if you complied perhaps no ticket for you.

No tipping a valet, the car parks itself. What if a firefighter could order your car out of a no parking zone or make the zone much bigger? Maybe you wouldn't need to designate no parking zones at all. The same for handicapped parking, the "driver" gets out at the door and the car parks itself anywhere. The owner would "summon" it on the way to the door and doesn't have to travel across the parking lot at all.

I have been blueskying a lot. :)

cjameshuff
2012-Mar-30, 01:56 PM
Coordinate with other vehicles on the nearby roads to minimize stops at lights, route around blocked or particularly crowded roads, etc. This is something that wouldn't require full automation.

Collision detection and avoidance, coordinating multiple vehicles, etc are not major obstacles, the big challenge is terrain navigation...recognizing where the vehicle can safely and appropriately drive. There's a lot of unmarked, poorly maintained roads and driveways, pedestrian walkways and access paths that look like roads, etc. It's unreasonable to expect a vehicle to work out where it can go on its own when dealing with routes that aren't in any database and lack any unambiguous indications. A workaround might be to have the driver manually drive the route and some alternate routes for the car to store for later reference, so the car "learns" the roads its owner needs it to drive on (and perhaps uploads information about the roads to a common database, though you'd need a way to handle things such as private drives...maybe just require a certain number of unique travelers in a given period of time before a path is considered a public road).

danscope
2012-Mar-30, 03:21 PM
Why do helicopters crash? Answer: too many moving parts.
To suggest that you can dumb-down the driving public with your infatuation with computer chips is a non-starter. It is more likely that we will reduce the number of drivers and remove some of the riffraff that never should have a license in the first place. That's all.

cjameshuff
2012-Mar-30, 05:34 PM
Why do helicopters crash? Answer: too many moving parts.

They quite often don't crash. And when crashes do happen, the majority of the time it is due to pilot error.

They don't even have that many moving parts compared to vehicles with piston engines, their mechanical failure rate has more to do with the requirements for low weight and the high stresses imposed on and performance required from the components. Besides, computers involve few or no moving parts. Plus, automobiles already have them. A computer in control of a vehicle has the same ultimate fallback the engine control computer has if it experiences failure...stop the vehicle. It in fact has the additional backup of a human who can take over if it encounters problems.



To suggest that you can dumb-down the driving public with your infatuation with computer chips is a non-starter. It is more likely that we will reduce the number of drivers and remove some of the riffraff that never should have a license in the first place. That's all.

You have yet to give any facts or reasoning supporting your position, just uninformed ranting and appeals to emotion.

Ara Pacis
2012-Mar-30, 06:38 PM
Lane markings are often absent, worn, buried under snow, obsolete, or just plain incorrect, etc. They can not be relied on for navigation, which is why I assumed instrumentation of the roads.

True, but if a road is so poorly maintained or low state (gravel) that its lane markings are absent or worn out then it may not be a candidate for instrumentation, unless the instruments are cheaper than road paint per unit length of road. If we use a system where the instrumentation is few and far between (except on curves and intersections) then maybe it would be cheaper than paint. Maybe it would be a solar-powered module attached to road signs.

I still think passive markings are just as good, if not better, because they have a different failure mode and a longer mean-time-between-failures. Not only can it be markings on the road but on the backs of other signs on the road, perhaps with NIR reflecting or fluorescing material so that it's invisible to humans. The best practice would be to use both. Of course, any active or passive marking risks the same hazard as regular visible elevated signs - they attract bullets.


The bit rate required is not great, a straightforward solution would be to use IR LEDs in the vehicle lights to emit signals that the navigation cameras can pick up. This would make it quite clear which car is transmitting which signal and make interference difficult.

I was thinking NIR LEDs would be the perfect candidate tech. I was thinking about having it separate from the other indicator lamps, but with the change to color LED instead of incandescents, IR blooming won't be a problem. However, having a set distance between the various LEDs would help a computer discern distance, speed and heading. But the system is not immune to accidental or intentional spoofing (sunrise/sunset), and people may customize their cars to remove or obscure the NIR signal LEDs. (some people already do that with Daytime Running Lamps, and some people like to put running lights on trucks smaller than 3/4 ton just for kicks). It might be a good idea to give a computer the ability to use shadows and silhouettes to make guesses about the size and location of vehicles that are not instrumented properly, and make guesses about intentions based on aspect changes in the outline of that vehicle. It might even be useful for it to have a database of known vehicle outlines so that the computer can use those dimensions to more clearly discern distance, speed and heading. On the other hand, maybe it's not that important as using visible taillights for aspect discernment may be sufficiently accurate when combined with an adequate safety margin.

A database of vehicle outlines might also be useful for police in case a certain vehicle needs to be identified or if there is a search for a certain type of vehicle due to a crime (e.g. an Amber Alert - child abduction emergency), in which case the computer can then attempt to capture an image of the license plate for comparison. (I know that might sound like a privacy issue, but perhaps the car should ask the driver if it should keep an eye out in the event of an alert in their SAME and alert the driver if the vehicle is found and ask if it should contact police, so that the human always maintains agency, which might help prevent any Big Brother abuse.)

Ara Pacis
2012-Mar-30, 06:54 PM
Traffic stops could be an option event: "Place your car in self drive mode or pull over." Much safer for police, and if you complied perhaps no ticket for you.

Can people hear police car loudspeakers when at highway speeds? They might use some a radio or light signal to make your computer notice when it's your car specifically being asked to do something (as opposed to a general move over and let the police car or ambulance or fire truck pass). Unfortunately, some people might try to spoof the signal to cause cars to pull over for kicks or for criminal intent. So, a solution might be to have the police car use some sort of coded or encrypted protocol by which to identify itself that your car can use to query an online or stored database (regularly updated).


No tipping a valet, the car parks itself. What if a firefighter could order your car out of a no parking zone or make the zone much bigger? Maybe you wouldn't need to designate no parking zones at all. The same for handicapped parking, the "driver" gets out at the door and the car parks itself anywhere. The owner would "summon" it on the way to the door and doesn't have to travel across the parking lot at all.

Cars might also be able to park closer together and fit more in. When one wants to park or leave, it signals the other cars and they all shuffle a bit so that there is room to maneuver.

NEOWatcher
2012-Mar-30, 07:01 PM
Cars might also be able to park closer together and fit more in. When one wants to park or leave, it signals the other cars and they all shuffle a bit so that there is room to maneuver.
Having my car start everytime for everyone around me parking or leaving.... Thanks for killing my battery.

We can dream up schemes and solutions to all these problems, but the reality is that any system needs to be maintained, needs to be adopted for all applications. Cities have a hard time with thier infrastructure now. I see plenty of missing signs, burned out traffic lights, crumbled roads, etc all around me. If they don't have the time and money to maintain what they have now, how do you think they can do the same for a more sophisticated system?

Besides, technology can solve a lot of things, but it also causes the issue that when something goes wrong, it goes horribly wrong.

cjameshuff
2012-Mar-30, 07:10 PM
True, but if a road is so poorly maintained or low state (gravel) that its lane markings are absent or worn out then it may not be a candidate for instrumentation, unless the instruments are cheaper than road paint per unit length of road. If we use a system where the instrumentation is few and far between (except on curves and intersections) then maybe it would be cheaper than paint. Maybe it would be a solar-powered module attached to road signs.

What I had in mind was RFID tags embedded in the pavement. Almost certainly would be cheaper than paint, installation would be the great bulk of the cost. And you mainly need them at turns and intersections and where speed limit changes...more or less where you'd need a painted marking.

And maybe the roads are better maintained where you live...there's some heavily traveled roads in this area that get much of the way to gravel before getting fixed up, especially up in Michigan (combination of thinly populated areas with poor funding for road maintenance and fall-spring weather that makes roads rapidly disintegrate).



I still think passive markings are just as good, if not better, because they have a different failure mode and a longer mean-time-between-failures. Not only can it be markings on the road but on the backs of other signs on the road, perhaps with NIR reflecting or fluorescing material so that it's invisible to humans. The best practice would be to use both. Of course, any active or passive marking risks the same hazard as regular visible elevated signs - they attract bullets.

Markings of any sort, even signs, tend to get obscured in snow (or flattened by plows/wayward vehicles) during the winter or vegetation during the summer around here. It's barely a suitable approach with human drivers.

Ara Pacis
2012-Mar-30, 08:51 PM
What I had in mind was RFID tags embedded in the pavement. Almost certainly would be cheaper than paint, installation would be the great bulk of the cost. And you mainly need them at turns and intersections and where speed limit changes...more or less where you'd need a painted marking.Sounds workable.


And maybe the roads are better maintained where you live...there's some heavily traveled roads in this area that get much of the way to gravel before getting fixed up, especially up in Michigan (combination of thinly populated areas with poor funding for road maintenance and fall-spring weather that makes roads rapidly disintegrate).Sounds like we're at the same latitude. We get road replacements fairly frequently due to the multiple freeze-thaw cycles each winter, which means roads go bad quickly, but they also get fixed quickly.


Markings of any sort, even signs, tend to get obscured in snow (or flattened by plows/wayward vehicles) during the winter or vegetation during the summer around here. It's barely a suitable approach with human drivers.But they are continuous and for the most part, contiguous. What would be the spacing on RFID tags? I'm guessing that depends on power source and that depends on size and that depends on ruggedness for a roadbed.

I ask because if the wheels are slightly out of alignment or the wind is blowing or the roadbed is slanted, the car may drift out of position. A lane marker spotting system will notice this continuously. GPS can be fairly precise, if there are enough satellites visible, and can tell you where you are along a line with a map as guide, but it may not be accurate enough to keep you from drifting onto the shoulder or into the next lane.

The computer will need good optical recognition anyway, to avoid obstacles too and know the position of the car in relation to the optics to know how to clear an obstacle, so it may as well read lane markings.

Ara Pacis
2012-Mar-30, 08:58 PM
Having my car start everytime for everyone around me parking or leaving.... Thanks for killing my battery.By the time computers are driving cars, you're powertrain is likely to be mostly battery.


We can dream up schemes and solutions to all these problems, but the reality is that any system needs to be maintained, needs to be adopted for all applications. Cities have a hard time with thier infrastructure now. I see plenty of missing signs, burned out traffic lights, crumbled roads, etc all around me. If they don't have the time and money to maintain what they have now, how do you think they can do the same for a more sophisticated system?How is this different from now? Roads, signs, traffic lights all need to be maintained in the current paradigm. If a lot of the system is based not on the roadway but on vehicles, then they will naturally take over as newer vehicles replace older vehicles. It might make the newer cars more expensive, although a lot of the increasing computer use of cars is getting more affordable through economies of scale, but as accident rates decrease, insurance premiums may also decrease on such cars allowing it to pay for itself.


Besides, technology can solve a lot of things, but it also causes the issue that when something goes wrong, it goes horribly wrong.That's why you should have redundancy and alternate methods in a system. Human drivers currently have no redundant systems, backseat drivers not included.

danscope
2012-Mar-31, 03:11 AM
And that is worse than what happens now in what way? Sorry, I've known several people killed or seriously injured by drivers who were neither drunk nor using cell phones. If we're collectively sane, so we have an adequate certification process, computer-controlled cars are going to be safer than human-controlled ones.

Hi Swampy, The whole problem with this concept is failure mode. Now, if your engine konks out, it's bad enough, but AAA will get to you and tow you( better get out of the car and safe on the side ). But to suggest that we abandon our responsibility to operate an motor vehicle and cede that responsibility to a computer begs credulity. Living on the edge of a surprise is no way to go through life. The failure mode is death .End of game.
No, the technological infatuation of putting computers to work on every device known to man is fraught with dangers beyond reason. Talk about computer crash !!!!!!!!!!!!
Your life hangs by a thread with such a scheme. There are many applications where computers serve us well. This is not one of those applications.
Frankly, I would rather have an old volkswagen bug with a simple electrical system that was bullet-proof than these computer infested behemoths which are deliberately designed to
get into your wallet....in a big way. We want more reliability, not more vulnerability. At least..... that is how I believe and think. I would hate to have to reply when someone asks .....
" Say, where is your family ? " that " They died in a head-on crash from a failed computer.
I can rest easy,...knowing full well that the insurance companies won't buy into such a naive scheme. Even they have money to lose. We have only our lives.

Best regards,
Dan

Inclusa
2012-Mar-31, 03:14 AM
Ok, let's first do the mixed first: an automobile that will stop you from dangerously crossing lanes and other dangerous actions, rather than a fully computer-driven automobile first.
Institutional reforms have been proposed long time ago, but how effective are them?

danscope
2012-Mar-31, 05:47 AM
Ok, a deer jumps out at night (happens all too often) . Your lane is blocked . You have 2 seconds to act. BUT......
you have to over-ride this stupid computer which insists on driving you through the windshield and totaling your car.....
so you won't change lanes. No. People have to grow up, think responsibly and act .

DoggerDan
2012-Mar-31, 07:57 AM
Ships are largely computer-driven, particularly tankers and cargo vessels. We override as required, but it's usually not required.

cjameshuff
2012-Mar-31, 04:04 PM
Ok, a deer jumps out at night (happens all too often) . Your lane is blocked . You have 2 seconds to act. BUT......
you have to over-ride this stupid computer which insists on driving you through the windshield and totaling your car.....
so you won't change lanes. No. People have to grow up, think responsibly and act .

Still no reason or facts, just nightmare scenarios based on your own ignorance and meant to provoke an emotional response.

This is exactly the situation a computer is better equipped to handle than a human. More likely is that you swerve into oncoming traffic, lose control of the vehicle, or fail to react sensibly in time to avoid the collision, when a computer driver could have stopped the car while avoiding oncoming traffic, or in the worst case when the collision is unavoidable, reducing the speed at time of impact to something lower than a human driver could achieve. A computer would be taking evasive action and checking for other hazards while a human is still registering there's something amiss.

The truth that you keep ignoring is that humans are terrible drivers, and it's quite likely computers can be better, safer ones. We do not have a "responsibility" to ensure there is someone to blame when accidents happen. I'll take the risk of getting injured or killed by a wayward computer driver quite happily if it's lower than the risk of the same happening with a wayward human driver.

swampyankee
2012-Mar-31, 04:35 PM
Ok, a deer jumps out at night (happens all too often) . Your lane is blocked . You have 2 seconds to act. BUT......
you have to over-ride this stupid computer which insists on driving you through the windshield and totaling your car.....
so you won't change lanes. No. People have to grow up, think responsibly and act .

In the second or so that it would take you to react, the computer calculates the current, straight line path of the deer, the current, straight-line path of the car, applies the brakes (and, like it or not, ABS is much, much better at controlling brakes than are humans), checks the other lane for oncoming traffic, and either a) swerves to miss the deer or b) determines that hitting the deer is less risk than hitting that truck coming down the other lane or that large tree to the right. This could all happen in about the time it takes for you to actually take your foot off the accelerator and move it to the brakes -- studies have shown this takes an alert human driver at least a second just to do that when confronted with an unexpected event -- and it makes it very difficult to see where it's better to disable the system: it will probably handle this sort of situation just fine.

Chances are very good that a computer driver will be better than most people really are, as opposed to how good most people think they are. Do consider that almost all drivers significantly overestimate their skill level and their ability to react in an emergency. Some of the worst exaggeration of skills are among people who consider themselves to be "very good" drivers.

danscope
2012-Mar-31, 05:06 PM
I said "Failure mode". You are driving along on two lane blacktop and your 'cruise'control ' fails,sending you head-on into
an oil truck. Done. Do you see ?

Dan

HenrikOlsen
2012-Mar-31, 05:17 PM
I said "Failure mode". You are driving along on two lane blacktop and your 'cruise'control ' fails,sending you head-on into an oil truck. Done. Do you see ?

Dan
You are driving along on two lane blacktop and you fall asleep, sending you head-on into an oil truck. Done. Do you see ?

Or rather, do you see why this is a silly way to argue?

swampyankee
2012-Mar-31, 05:34 PM
I said "Failure mode". You are driving along on two lane blacktop and your 'cruise'control ' fails,sending you head-on into
an oil truck. Done. Do you see ?

Dan

Or the brakes on the poorly-maintained truck fails, plowing into several cars and killing about a half-dozen people? It's the sort of risk we already have.

danscope
2012-Mar-31, 07:38 PM
And shall we compound our risks with electronics? Clearly no.

We had some good people on the Thresher. Some engineer decided to put an angle limit switch into the scram system.
When they hit the emergency blow system, and forward blew first, it scrammed the reactor. Ran out of steam and slid back down. Another case of electronics installed in a bad place.

I say put the money and research into better batteries. Now you are talking. The primary mission is transportation.

cjameshuff
2012-Mar-31, 08:55 PM
I said "Failure mode". You are driving along on two lane blacktop and your 'cruise'control ' fails,sending you head-on into an oil truck. Done. Do you see ?

You're ignoring the inconvenient facts that human drivers do this all the time, rather more often than existing car computers fail on the road, and that (as I've mentioned several times), there's a failsafe behavior in simply stopping the vehicle. That's not entirely safe, but mainly because the guy behind you might not have his car on automatic and so might not react fast enough to avoid running into you.



And shall we compound our risks with electronics? Clearly no.

We would not be compounding our risks, a self driving car is not subject to the risks that result from having a human driver. We would be replacing a known-unreliable part with one that is designed to do the job more safely and reliably. You have done nothing to demonstrate that computers would be inherently worse than human drivers.



We had some good people on the Thresher. Some engineer decided to put an angle limit switch into the scram system.
When they hit the emergency blow system, and forward blew first, it scrammed the reactor. Ran out of steam and slid back down. Another case of electronics installed in a bad place.

So an abnormal situation triggered a safety system in a submarine's nuclear reactor. How is this in any way relevant?

This is just another example of you trying to argue with something intended to provoke an emotional reaction, rather than actual facts.

orionjim
2012-Mar-31, 09:10 PM
And shall we compound our risks with electronics? Clearly no.

We had some good people on the Thresher. Some engineer decided to put an angle limit switch into the scram system.
When they hit the emergency blow system, and forward blew first, it scrammed the reactor. Ran out of steam and slid back down. Another case of electronics installed in a bad place.

I say put the money and research into better batteries. Now you are talking. The primary mission is transportation.

Dan are you aware that today’s automobiles have anywhere from 40 to 50 computers in them and the same amount of computing power in the Thresher’s time would have probably required half of the Navy’s fleet just to carry them?

The Thresher went down in 1961, over 50 years ago. Yes, things have gotten more complex, but our ability to handle and deal with complexity has also changed.

A driverless car is complicated but the only thing I can see that will stand in the way from it happening are not engineering and design issues but rather legal and liability issues.

To understand the real problem you need to watch this video of Sebastian Thrun who is the brains behind Googles driverless car. Sebastian’s goal is to reduce the number of auto related deaths by one million a year:

http://www.youtube.com/watch?v=bp9KBrH8H04

Today there are about 1.2 million auto related deaths a year. If Sebastian and the automakers together are able to reduce the deaths by one million a year think of what they have accomplished. But can you see where the problem is? The automakers have taken over the responsibility of the remaining 200 thousand deaths.

Should the automakers not proceed? That seems to be your suggestion.

Fortunately Google and the automakers are working on the technology, but someone is going to need to step up and deal with the legal issues.

danscope
2012-Apr-01, 12:31 AM
The advocates for driverless computerized cars have not had to repair too many things in life. We aren't talking about inconvenience. We are talking death. Sell it to the insurance companies and the clamdiggers back home.
Keep it simple .
Question: would you like to drive a car with a steering collum made of a flourescent bulb tube? Probably not, for when it fails you will have no connection to the steering at all. What you are asking is to abandon a solid mechanical conection
for drive-by-wire. I simply don't like it at all.
Those who want it should try it on the alpine rally. Good luck.

DonM435
2012-Apr-01, 01:04 AM
If every square mile of town were controlled by a dedicated computer that could keep track of every vehicle, every pedestrian and every hazard, said computer could shuffle all the traffic along at optimal speeds prior to turning it over to the next computer. Maybe that would work. But anything less than that, I'm not too comfortable with. The first fatality clearly attributed to a computer will probably scullte billions of dollars of research.

swampyankee
2012-Apr-01, 01:16 AM
On the other hand, some of us have worked with very complex systems -- much more complex than automobiles -- and do not find the idea of computer-driven cars frightening. I will, quite cheerfully, get on board aircraft or ships where there is no direct connection between power levers (real vehicles don't have throttleable engines) and engines or directional controls (columns, wheels, pedals) and the control surfaces.

Ara Pacis
2012-Apr-01, 02:02 AM
If every square mile of town were controlled by a dedicated computer that could keep track of every vehicle, every pedestrian and every hazard, said computer could shuffle all the traffic along at optimal speeds prior to turning it over to the next computer. Maybe that would work. But anything less than that, I'm not too comfortable with. The first fatality clearly attributed to a computer will probably scullte billions of dollars of research.

Now that sounds dangerous. It's too much responsibility with what appears to be a single point of failure. Now, if that computer was like a traffic cop and directed cars that had computers that were able to ignore such direction when necessary (tree branch in the road, etc), then we're getting somewhere.

As for the deer mentioned earlier, someone should remind Dan that humans don't see in the dark. This gets even more difficult when headlights are blinding. Computers, on the other hand can not only look several places at once, they can see in the dark using thermal and don't have to be blinded by headlights.

cjameshuff
2012-Apr-01, 02:29 AM
The advocates for driverless computerized cars have not had to repair too many things in life.

You have absolutely no grounds for making that claim. Your emotionally based arguments clearly show you have nothing behind your position and now you resort to argumentum ad hominem.



We aren't talking about inconvenience. We are talking death. Sell it to the insurance companies and the clamdiggers back home.

Yes, we're talking death. Human drivers are constantly getting themselves and other people killed due to stupid mistakes and poor reactions, and self driving cars and driving assistance systems could greatly reduce the death rate.



Question: would you like to drive a car with a steering collum made of a flourescent bulb tube?

Of course not, such a vehicle was clearly designed by an idiot. Fortunately, we aren't discussing such vehicles.



Probably not, for when it fails you will have no connection to the steering at all. What you are asking is to abandon a solid mechanical conection
for drive-by-wire. I simply don't like it at all.

We aren't, actually. Computer driving doesn't fundamentally require steer-by-wire, though the human driver would have to take some care around the steering wheel on the vehicles that don't use it. And yes, I would drive a car that was steer-by-wire. Most vehicles today already have power steering anyway, a mechanically complex system that can wreak havoc when it fails just from the vehicle no longer responding as it should. The risk of the steering failing totally with no warning is quite low in comparison to the risk from a mistake by myself or other human drivers.



Those who want it should try it on the alpine rally. Good luck.

Yet another totally irrelevant attempt at distraction from your lack of support for your arguments...

cjameshuff
2012-Apr-01, 02:37 AM
Now that sounds dangerous. It's too much responsibility with what appears to be a single point of failure. Now, if that computer was like a traffic cop and directed cars that had computers that were able to ignore such direction when necessary (tree branch in the road, etc), then we're getting somewhere.

Indeed, and there's further complexities with handing things off from one city to the next, and who controls traffic in between? Plus the demands of constant long range communication between the vehicles and controlling system, and issues of interference and obstruction of signal. A distributed system centered in the vehicles themselves is far more reasonable. City-wide traffic control is still a possibility, but interacting with the cars via stoplights, publishing advisories of traffic levels on different routes, etc, with the cars handling the actual driving themselves.

cjameshuff
2012-Apr-01, 02:46 AM
Besides, technology can solve a lot of things, but it also causes the issue that when something goes wrong, it goes horribly wrong.

It can, but it also quite often does the opposite. Look at what MEMS technology allowed auto makers to do with airbags, or the difference made by automatic braking systems.

In this case, when something goes wrong there's also the possibility of sounding an alarm and falling back to human control or just stopping the vehicle if the system detects a fault with the computer or the computer has problems controlling the vehicle or making sense of its surroundings. The ultimate fallback if the driving computer completely and suddenly fails could be quite simple and self contained...all it needs is an accelerometer/gyro to keep the car going in a straight line as it stops (or perhaps keep the rate of turn constant). (It doesn't strictly require even that, but it might be preferable to have some active directional control as the car stops.)

swampyankee
2012-Apr-01, 02:53 AM
Probably the absolute easiest place to get a computer-driven vehicle to work would be drag racing. Likely the only reason it hasn't been done is because it's probably against the rules. I'll bet a team of three or four EE and ME undergraduates could get a computer-controlled dragster working as a two semester project.

Inclusa
2012-Apr-01, 03:29 AM
The arguments seem to be between:
1)Liability for deaths
2)Computer decision making vs human decision making
3)Safety

There are also quite a few professions at stake: 1)drivers of various types 2)driving instructors 3)traffic police
But why should we keep the status quo when we know there are better alternatives?

danscope
2012-Apr-01, 04:19 AM
@ James, You mean well, but don't comprehend the problems and severe risks associated with this scheme.
When you have put your life on the line for a living, you tend to evaluate these kind of schemes with an eye to
practical safety. If you love computers that much, I wish you joy of your research. Good luck with the insurance companies. They are quite clear-eyed when it comes to safety. And..... don't believe everything you read in
Popular Science.

cjameshuff
2012-Apr-01, 04:39 AM
There are also quite a few professions at stake: 1)drivers of various types 2)driving instructors 3)traffic police
But why should we keep the status quo when we know there are better alternatives?

I don't think it's feasible to make cars so automatic that you'll never need to take them to manual, so people will still need to learn to drive. And even if all cars on the roads are capable of automatic driving, some people will use manual if it's an option, even if just (or perhaps especially) in an attempt to game the system. Plus there's still breakdowns and such. And drivers are in some cases needed to supervise the vehicle and its contents, especially if something goes wrong with it. Demand for some professions might shift, and some of them will change a bit in nature, but I don't see any of these going away.



@ James, You mean well, but don't comprehend the problems and severe risks associated with this scheme.
When you have put your life on the line for a living, you tend to evaluate these kind of schemes with an eye to
practical safety. If you love computers that much, I wish you joy of your research. Good luck with the insurance companies. They are quite clear-eyed when it comes to safety. And..... don't believe everything you read in Popular Science.

I do understand the problems and risks. All your arguments have been based on incorrect and nonsensical comparisons and nightmare scenarios that exist only in your own mind. You have utterly failed to demonstrate any fundamental safety issues, or even any understanding of the actual difficulties and hazards involved.

(And I gave up on Popular Science years ago.)

DonM435
2012-Apr-01, 05:07 AM
Yeah, it would seem that all vehicles and all people would require sensors. But what about the dogs, the deer, the fallen tree or misplaced garbage can that ends up in the street?

Ara Pacis
2012-Apr-01, 06:52 AM
Yeah, it would seem that all vehicles and all people would require sensors. But what about the dogs, the deer, the fallen tree or misplaced garbage can that ends up in the street?Sensored garbage?


When you have put your life on the line for a living, you tend to evaluate these kind of schemes with an eye to
practical safety. If you love computers that much, I wish you joy of your research. Good luck with the insurance companies. They are quite clear-eyed when it comes to safety.

We've addressed each and every one of your concerns with straightforward engineering solutions. Maybe the idea was silly decades ago, but now we know how to actually do these things and have systems in place (GPS, digital mapping, optical recognition, RFID, etc) that make it actually feasible now. But let me ask you this, are you the only one who ever drive or do you ever let someone else drive a car you ride in? I ask because it's possible their vision is worse than yours, their reaction time is slower than yours, their ability to navigate is worse than yours and their ability to control the car is worse than yours. Now, even if their faculties are not worse than yours, they are certainly worse than a machine's, by an order of magnitude.

Solfe
2012-Apr-01, 08:01 AM
Can people hear police car loudspeakers when at highway speeds? They might use some a radio or light signal to make your computer notice when it's your car specifically being asked to do something (as opposed to a general move over and let the police car or ambulance or fire truck pass). Unfortunately, some people might try to spoof the signal to cause cars to pull over for kicks or for criminal intent. So, a solution might be to have the police car use some sort of coded or encrypted protocol by which to identify itself that your car can use to query an online or stored database (regularly updated).

I was picturing a coded radio system as opposed to a loudspeaker.

orionjim
2012-Apr-01, 04:58 PM
Well, here it is… Google announced today they were entering a driverless car into the Daytona 500.

See: http://www.youtube.com/watch?v=hlsW3BHhtwA&feature=player_embedded#!

Imagine that; today!

swampyankee
2012-Apr-01, 05:28 PM
Well, all the roundy-round racers have to do is stand on it and turn left.

Solfe
2012-Apr-01, 05:52 PM
I would think the first step to all of this would be to place technology in cars that is not related to entertainment. My mother-in-laws car can change the radio station on voice command, dial the phone, and connect to phones and ipods automagically. Good technology, but wrongly purposed. Failure in these sorts of things add to driver distraction. "Why won't the phone dial???"

Some cars already have HUD's, cameras and radar to eliminate blindspots. Ford Focuses have detectors to check the blindspots for people/objects on the rear and sides of the car. GPS systems are nice too, with the hands off controls. The prices on these device need to come down and stop being add on options. They are very to highly reliable as far as they go; after you have a significant portion of the public using them the idea of turning a car over to a computer becomes less of a hurdle.

Ara Pacis
2012-Apr-01, 07:11 PM
Also, imagine all the savings on gas if the computer drives a lower speeds per law. We might save the planet while we're saving lives.

Wait a tick, if more people live, and more people who can't drive cars now start being driven by them in the future, then we'll just have more cars on the road causing more traffic and using more gas/electricity. And since people don't have to pay attention to the road, they may get frisky in the backseat thereby creating more people that will become if not future drivers then future auto-car riders. Oh, great! there goes the planet.

I gest, but it might need to be clear that an auto-car must have a licensed driver in it just in case. Otherwise, we'll end up with youngsters and oldsters on the road where they don't belong. 'Cause you just know that some upper-middle-class parent is going to get one to ferry her kids to school while she stays home.

ShinAce
2012-Apr-01, 07:44 PM
Google needs to work harder on their april fool's jokes.

I would be fine with computer driven cars. We already have traction control, cruise control which slows down the car when it closes the gap on the car in front.

Think about the benefits. Women will be able to legally do their makeup in the car. Thats a multi billion dolar value to our economy.

publiusr
2012-Apr-01, 07:45 PM
Earlier I posted--I forgot where--a blurb where intersections have no more lanes stopping. The computer has two flows of traffic passing through each other at speed in the spaces between cars.

Now that is fine in theory, the problem is that the car and its plumbing don't have the fine control of the computer itself. We saw that with some of Carmacks landers--the computer is faster than the plumbing. Now we have all seen that footage of kinetic Kill vehicles hovering over that net remaining very stable. That precision is rare.

So if a car is not properly maintained and its engine stumbles while passing through an intersection--it throws everything off and causes a horrific wreck that the computer actually makes worse since it keeps things at speed.

The computer has a faster reaction time than even the most sober of us--but it is of no matter since the momentum is still there and cannot be instantly stopped, unless the traffic is very slow, as what you see when cars creep up to the light.

So the standard practice of stopping at the intersection may have to continue...

Ara Pacis
2012-Apr-01, 07:55 PM
Earlier I posted--I forgot where--a blurb where intersections have no more lanes stopping. The computer has two flows of traffic passing through each other at speed in the spaces between cars.

Now that is fine in theory, the problem is that the car and its plumbing don't have the fine control of the computer itself. We saw that with some of Carmacks landers--the computer is faster than the plumbing. Now we have all seen that footage of kinetic Kill vehicles hovering over that net remaining very stable. That precision is rare.

So if a car is not properly maintained and its engine stumbles while passing through an intersection--it throws everything off and causes a horrific wreck that the computer actually makes worse since it keeps things at speed.

The computer has a faster reaction time than even the most sober of us--but it is of no matter since the momentum is still there and cannot be instantly stopped, unless the traffic is very slow, as what you see when cars creep up to the light.

So the standard practice of stopping at the intersection may have to continue...

I agree. The computer-driven car shouldn't be allowed or expected to perform maneuvers that a human wouldn't be able to perform. "Right of Way" is too basic a concept to any traffic system to give up for an increase in speed at intersections. On multi-lane highways, on the other hand, the cars would be stationary with respect to each other because they'd be driving in formation.

caveman1917
2012-Apr-01, 08:56 PM
I wonder about the impact of this technology on taxi's. Right now one of the biggest things in making taxi rides so expensive is the payment of the driver, what if this would be, by using driverless cars, eliminated? My best guess would be that this changes the economics of transportation such that for a significant part of the population it would become more economical to rely on taxi rides rather than own their own car. So even if the benefits in terms of safety would be marginal (which they probably wouldn't), the benefits on traffic congestion and other economic aspects might be worth the effort by itself.

cjameshuff
2012-Apr-01, 09:03 PM
So if a car is not properly maintained and its engine stumbles while passing through an intersection--it throws everything off and causes a horrific wreck that the computer actually makes worse since it keeps things at speed.

I have trouble seeing exactly how you see this happening. Cars certainly wouldn't be slipping through the intersection so closely that low friction paint jobs have an impact on your mileage. It'd be pretty straightforward to not enter an intersection if another car's stuck in the middle of it, and to ensure that the intersection will be clear when the car attempts to go through it. Above a certain traffic level, there wouldn't be enough room to do this...and so vehicles would stop and take turns rather than smash into each other trying to go through at the same time.

ShinAce
2012-Apr-01, 10:06 PM
My car had a problem that had me scared of being rear ended. I would be going down the highway and it would kill the cruise control and feel like i hit a half inch of sand. Fly by wire was not taking throttle commands. Yet i would pull over, restart the car and be fine. Finally the throttle body gave out and we found thr problem.

I learned a few things:
You can get into serious trouble from the simplest failures. Gas pedal getting stuck, losing power from the engine, not having working brakes.
The most sophisticated computer in the shop can only properly diagnose a car problem x% of the time.
Sometimes the fix that a computer applies is wrong. My car would go into a limp simply because the engine wasnt giving enough power. If that happened to me on ice i would be writing a letter asking why a computer prevented me from avoiding an accident.

A human must be able to drive the car without restrictions. Things like traction control and limp modes neeed to disabled sometimes. Making a car handle like an iphone is just not a good idea.

However, if we all had the system it would be great for the highway.

LotusExcelle
2012-Apr-01, 11:23 PM
I would have no problem at all turning over control of my car to a computer. The tech isn't there just yet but it is a matter of time - and not much at that. Computers WILL be able to handle almost every driving situation AND be able to handle those situations better than most humans.

And that is the point I'd like to stress. The vast majority of drivers are mediocre at best. Accidents occur mostly from human error. Do you really trust humans to drive *better* than a fully developed computer system? I don't.

LotusExcelle
2012-Apr-01, 11:26 PM
My car had a problem that had me scared of being rear ended. I would be going down the highway and it would kill the cruise control and feel like i hit a half inch of sand. Fly by wire was not taking throttle commands. Yet i would pull over, restart the car and be fine. Finally the throttle body gave out and we found thr problem.

I learned a few things:
You can get into serious trouble from the simplest failures. Gas pedal getting stuck, losing power from the engine, not having working brakes.
The most sophisticated computer in the shop can only properly diagnose a car problem x% of the time.
Sometimes the fix that a computer applies is wrong. My car would go into a limp simply because the engine wasnt giving enough power.

I'd like to clarify something. A computer does not diagnose a problem with the car. The technician working on the vehicle does. If the technician fails to correctly diagnose the problem it is not the fault of any of the computers involved. No computer I'm aware of "applies a fix" to any car system other than updating programming. I also believe your understanding of what was happening to your car and what a limp-mode is is a bit off.

ShinAce
2012-Apr-01, 11:49 PM
The computer killed throttle input. The butterfly on the throttle body was likely sticking. Power was available, but not used.

Imagine you're towing a boat up a hill when the computer decides that the engine will overheat as kills gas to one cylinder. You'll be cursing.

Computets can drive fine in plenty of situations, but not all.

I still think the driver should always have access to the controls. Its going to be hard to implement the system when some dont use it.

There are cars that can self parallel park, but arent you still liable when something goes wrong. Ill pay for my mistakes, but not a computer's.

LotusExcelle
2012-Apr-01, 11:53 PM
I would rather save the multiple thousands on the engine and let the computer keep it safe and cool. The butterfly on all throttle-by-wire throttle bodies has a strong spring which defaults it to shut. (Incidentally this is why the Toyota runaway car syndrome was found to be erroneous) Power also is not the correct term - if the throttle cannot open, for any reason, then the car is idling. Throttle cables on carbs used to snap too. The throttle by wire system offers a much wider range of control by the ECM.

swampyankee
2012-Apr-02, 12:01 AM
I'd like to clarify something. A computer does not diagnose a problem with the car. The technician working on the vehicle does. If the technician fails to correctly diagnose the problem it is not the fault of any of the computers involved. No computer I'm aware of "applies a fix" to any car system other than updating programming. I also believe your understanding of what was happening to your car and what a limp-mode is is a bit off.

I know that one of the earliest applications of AI was diagnosis (the APACHE system (http://www.ncbi.nlm.nih.gov/pubmed/3928249)). Also, there are quite active projects to provide automated diagnoses of complex systems, a category into which automobiles are only recently beginning to enter.

HenrikOlsen
2012-Apr-02, 01:55 AM
My car had a problem that had me scared of being rear ended. I would be going down the highway and it would kill the cruise control and feel like i hit a half inch of sand. Fly by wire was not taking throttle commands. Yet i would pull over, restart the car and be fine. Finally the throttle body gave out and we found thr problem.

I learned a few things:
You can get into serious trouble from the simplest failures. Gas pedal getting stuck, losing power from the engine, not having working brakes.
The most sophisticated computer in the shop can only properly diagnose a car problem x% of the time.
Sometimes the fix that a computer applies is wrong. My car would go into a limp simply because the engine wasnt giving enough power. If that happened to me on ice i would be writing a letter asking why a computer prevented me from avoiding an accident.

A human must be able to drive the car without restrictions. Things like traction control and limp modes neeed to disabled sometimes. Making a car handle like an iphone is just not a good idea.

However, if we all had the system it would be great for the highway.
I would take the time to be slightly nasty and claim that this is not an example of computer error but rather an example of the dangerously erratic behavior of a human driver.
Ignoring an error condition by overriding the computer control is exactly what makes the ability to override dangerous.
Refusing to continue to drive the car after detecting a potentially hazardous error in the engine control system would be part of the elementary safety features of the computer driven car.

And yes, you'd have the inconvenience of having a car that refused to drive until it was repaired because it couldn't guarantee complete throttle control, you'd also not have to be afraid of being rear ended because you "fixed" the problem by resetting the computer.
And no, you don't know better than the computer if it's safe to continue.

HenrikOlsen
2012-Apr-02, 04:24 AM
So if a car is not properly maintained and its engine stumbles while passing through an intersection--it throws everything off and causes a horrific wreck that the computer actually makes worse since it keeps things at speed.
How is this different from an engine that stumbles just when the human driver pulls into traffic?

Except that once the engine was detected as stumbling, the computer would have the option of signaling to the other cars that control was erratic, something the human driver can't, and has the option to get out of traffic as fast as possible and schedule a checkup, something a human driver probably wouldn't bother doing unless he got a real scare from a near accident.

Jens
2012-Apr-02, 06:49 AM
I wonder about the impact of this technology on taxi's. Right now one of the biggest things in making taxi rides so expensive is the payment of the driver, what if this would be, by using driverless cars, eliminated? My best guess would be that this changes the economics of transportation such that for a significant part of the population it would become more economical to rely on taxi rides rather than own their own car.

And if that were the case, it would be possible to have companies that basically have a membership fees that allow you to use one of the cars. You would just call, and the nearest car would come to your home. I imagine that pure personal vehicles would become much rarer than they are today.

swampyankee
2012-Apr-02, 09:46 AM
And if that were the case, it would be possible to have companies that basically have a membership fees that allow you to use one of the cars. You would just call, and the nearest car would come to your home. I imagine that pure personal vehicles would become much rarer than they are today.

I've always wanted a car that would come when it was called. I don't mind walking, empty-handed, to the store, but carting multiple bags to the farthest corners of a parking lot is another thing.

NEOWatcher
2012-Apr-02, 02:14 PM
By the time computers are driving cars, you're powertrain is likely to be mostly battery.
Ok, then thank you for using up my battery charge to shuffle it around.


How is this different from now? Roads, signs, traffic lights all need to be maintained in the current paradigm.
Complexity.

Signs: there's no real issues except that the person alert level drops somewhat, there are still plenty of visual clues. And what are the failure modes of a sign? Not much.
Traffic lights: they do fail. When they do, traffic can be controlled by someone controlling the traffic immediately. If an electronic controller fails, then you have to wait for the repairman, and the time it takes to repair. In the meantime what's controlling the traffic through that intersection?
Roads: I drive on plenty of deteriorated roads. It does not, in any way, inhibit my navigation.

Besides, you did not address my issue about budgets. Most current road conditions ARE NOT taken care of until the budget allows them to be. Only those vital items are addressed in a timely manner. In the automated system ALL repairs are vital and need to be addressed immediately.


If a lot of the system is based not on the roadway but on vehicles, then they will naturally take over as newer vehicles replace older vehicles. It might make the newer cars more expensive, although a lot of the increasing computer use of cars is getting more affordable through economies of scale, but as accident rates decrease, insurance premiums may also decrease on such cars allowing it to pay for itself.
But; if the vehicle requires roads to be easily recognized and maintained so they are usable by automated cars, then we still have the problem.


That's why you should have redundancy and alternate methods in a system. Human drivers currently have no redundant systems, backseat drivers not included.
I consider that an invalid comparison. We have an added system called "judgement".


In this case, when something goes wrong there's also the possibility of sounding an alarm and falling back to human control or just stopping the vehicle if the system detects a fault with the computer or the computer has problems controlling the vehicle or making sense of its surroundings.
Returning to human control can work provided the computer knows there is a failure and is able to relinquish control.
The other issue is in today's system, there are multiple mechanical systems still useable as something fails. I don't see that in computerized systems.

Something as simple as finding a safe place to stop is not something that I see a computer doing. In the past few years I had two problems where I needed to make that judgement. One was when I had a fuel line break and the engine was completely dead which caused the power steering and brakes not to operate. The other was a flat tire.
In both cases, I was able to judge how far I would be able to stumble or coast and was able to limp easily into a gas station. I don't see a computer being able to do any more than pull off to the side. This would be fine where there is an ample shoulder, but most roads don't have that.

I'm not saying that any of this can't work SOMEDAY. I just think that this sudden "ooh cool, do it now" mentality is way too premature. Let it evolve, let's see where the problems are first. People can be dealt with individually. An unforseen issue that senses something as a "turn right now" can affect millions at one time.

Ara Pacis
2012-Apr-02, 08:03 PM
Ok, then thank you for using up my battery charge to shuffle it around. Yes. If you don't like it, don't park on a parallel parking zone that requires shuffling. It's not like there won't be other options. Remember, driving is a privilege, parking even more so.



Complexity.

Signs: there's no real issues except that the person alert level drops somewhat, there are still plenty of visual clues. And what are the failure modes of a sign? Not much.
Traffic lights: they do fail. When they do, traffic can be controlled by someone controlling the traffic immediately. If an electronic controller fails, then you have to wait for the repairman, and the time it takes to repair. In the meantime what's controlling the traffic through that intersection?
Roads: I drive on plenty of deteriorated roads. It does not, in any way, inhibit my navigation.You'll have to talk to CJamesHuff about that, I'm suggesting cars that can read human-readable signs. A road reading computer system for a car would be able to drive the same roads you do, only better. As for failure modes, signs can be stolen, be defaced by paint or bullet holes or unfolded in the case of temporary signs that are deployed when a traffic light is out. But a computer with GPS might have an internal map that tells it where the traffic signs are supposed to be, while you might only do the same thing if you are familiar with the area.


Besides, you did not address my issue about budgets. Most current road conditions ARE NOT taken care of until the budget allows them to be. Only those vital items are addressed in a timely manner. In the automated system ALL repairs are vital and need to be addressed immediately. In an automated traffic system, perhaps, but not in a system that uses automated cars that can guide themselves.


But; if the vehicle requires roads to be easily recognized and maintained so they are usable by automated cars, then we still have the problem.Not in my proposal. Cars with a computer and sensors would be able to see the road better than a human in all conditions. So, if a road is so poorly recognized/maintained that a computer can't drive it, then a human shouldn't be driving it either.


I consider that an invalid comparison. We have an added system called "judgement".This comment does not make sense. Can you clarify?


Returning to human control can work provided the computer knows there is a failure and is able to relinquish control. The ability to relinquish control would be a sine qua non. There should be a dedicated "Disengage Autopilot" switch as well.


The other issue is in today's system, there are multiple mechanical systems still useable as something fails. I don't see that in computerized systems.In case it's not clear, the navigation and driving computer would be just that, it would not be the locomotion systems operation computer. In other words, it would drive the car but it would not run the car. It provides oversight of the simpler systems. It replaces the driver, not the ECM.


Something as simple as finding a safe place to stop is not something that I see a computer doing. In the past few years I had two problems where I needed to make that judgement. One was when I had a fuel line break and the engine was completely dead which caused the power steering and brakes not to operate. The other was a flat tire.
In both cases, I was able to judge how far I would be able to stumble or coast and was able to limp easily into a gas station. I don't see a computer being able to do any more than pull off to the side. This would be fine where there is an ample shoulder, but most roads don't have that.On the contrary, a computer that has a map that includes gas stations, road-shoulders and emergency pull-over areas would be much more capable of choosing a proper emergency stopping point than a human who is limited to line of sight.


I'm not saying that any of this can't work SOMEDAY. I just think that this sudden "ooh cool, do it now" mentality is way too premature. Let it evolve, let's see where the problems are first. People can be dealt with individually. An unforseen issue that senses something as a "turn right now" can affect millions at one time.The systems have evolved. The capability is current. This is illustrated by the fact that for every problem you conceive of, I have a workable answer that does not rely on future discoveries/inventions or on untested engineering concepts.

Ara Pacis
2012-Apr-02, 08:14 PM
I wonder about the impact of this technology on taxi's. Right now one of the biggest things in making taxi rides so expensive is the payment of the driver, what if this would be, by using driverless cars, eliminated? My best guess would be that this changes the economics of transportation such that for a significant part of the population it would become more economical to rely on taxi rides rather than own their own car. So even if the benefits in terms of safety would be marginal (which they probably wouldn't), the benefits on traffic congestion and other economic aspects might be worth the effort by itself.

That depends no what you mean by "taxi". A driving service/limo service that responds to calls might not need a human driver, but a taxi that roams the street looking for people to flag it down might still work best with a driver. Taxi systems are most common in large cities, and so traffic and ownership issues could be more affected by mass transit, but that's a different debate.

NEOWatcher
2012-Apr-02, 08:31 PM
The systems have evolved. The capability is current. This is illustrated by the fact that for every problem you conceive of, I have a workable answer that does not rely on future discoveries/inventions or on untested engineering concepts.
Yes; it has evolved. Yes; the capability is there. But; it is not refined, it is not proven, it has not shown to work in many situations except limited areas or off road.
It introduces maintenance issues and added costs.

Ara Pacis
2012-Apr-02, 08:31 PM
My car had a problem that had me scared of being rear ended. I would be going down the highway and it would kill the cruise control and feel like i hit a half inch of sand. Fly by wire was not taking throttle commands. Yet i would pull over, restart the car and be fine. Finally the throttle body gave out and we found thr problem.

I learned a few things:
You can get into serious trouble from the simplest failures. Gas pedal getting stuck, losing power from the engine, not having working brakes.
The most sophisticated computer in the shop can only properly diagnose a car problem x% of the time.
Sometimes the fix that a computer applies is wrong. My car would go into a limp simply because the engine wasnt giving enough power. If that happened to me on ice i would be writing a letter asking why a computer prevented me from avoiding an accident.

A human must be able to drive the car without restrictions. Things like traction control and limp modes neeed to disabled sometimes. Making a car handle like an iphone is just not a good idea.

However, if we all had the system it would be great for the highway.

I'm confused. Why would you expect a driving and navigation computer to act as a diagnostic computer? I get that it might be useful to the driver, but it's ultimately not necessary to the problem at hand. The critical issue for the driving computer is to be able to respond safely and appropriately to maneuver the car off the road so that you can evaluate the problem an attempt a repair or call for one. That's what human do currently. I'm not familiar with many cases of people popping the hood or crawling underneath to repair an engine problem while driving down the road at highway speeds.

Ara Pacis
2012-Apr-02, 08:40 PM
Yes; it has evolved. Yes; the capability is there.Thank you for agreeing with me.


But; it is not refined, it is not proven, it has not shown to work in many situations except limited areas or off road.Depends on what you're referring to. Some factories and warehouses use robotic forklifts and other mobile machinery that use optical systems to read painted lane markings. Other types of automatic driving systems have been demonstrated on-road. We don't need new tech, we just need to adapt existing tech to a different vehicle category with different performance requirements.

It introduces maintenance issues and added costs.It might costs more, that doesn't mean it'll be prohibitive. What sort of maintenance issues are you thinking of, cleaning the optics? You have to do that now.

ShinAce
2012-Apr-02, 10:27 PM
And no, you don't know better than the computer if it's safe to continue.

I'm sure the crew of Apollo 12 would love to hear that.

No one is a perfect driver, but at least when things go wrong, we're able to act. A computer won't give me CPR, put pressure to stop the bleeding, etc...

ShinAce
2012-Apr-02, 10:38 PM
I'm confused. Why would you expect a driving and navigation computer to act as a diagnostic computer? I get that it might be useful to the driver, but it's ultimately not necessary to the problem at hand. The critical issue for the driving computer is to be able to respond safely and appropriately to maneuver the car off the road so that you can evaluate the problem an attempt a repair or call for one. That's what human do currently. I'm not familiar with many cases of people popping the hood or crawling underneath to repair an engine problem while driving down the road at highway speeds.

That's exactly it, I feel like working on a hot, combusting, engine.

The problem I personally had was that the generic service light came on, but after restarting the engine, would go away. No error code saved. It was probably a 'fuel mixture too lean' error. Nonetheless, I thought it was hilarious that a computer would completely cut throttle to a car that's travelling at 70mph, then not retain an error code. The dealership shrugged it off, and I had the pleasure of having a car randomly die on the highway. That let me know that I cannot depend on a computer in the least bit when it comes to roadworthiness.

Star Trek TNG even did an episode where the simulation said they couldn't fly out. They made it out, without a navigation computer.

I'm also of the attitude "It's too expensive and too early".

Soon we'll have a computer to make love for you too.

swampyankee
2012-Apr-02, 10:49 PM
That's exactly it, I feel like working on a hot, combusting, engine.

The problem I personally had was that the generic service light came on, but after restarting the engine, would go away. No error code saved. It was probably a 'fuel mixture too lean' error. Nonetheless, I thought it was hilarious that a computer would completely cut throttle to a car that's travelling at 70mph, then not retain an error code. The dealership shrugged it off, and I had the pleasure of having a car randomly die on the highway. That let me know that I cannot depend on a computer in the least bit when it comes to roadworthiness.

Star Trek TNG even did an episode where the simulation said they couldn't fly out. They made it out, without a navigation computer.

I'm also of the attitude "It's too expensive and too early".

Soon we'll have a computer to make love for you too.


Considering something remarkably similar happened to many people I knew who drove cars that had no electronic engine controls, I've got to say "this ain't new..."

publiusr
2012-Apr-02, 10:55 PM
Now if an engine stumbles while you are pulling out--and the coast is clear--it is of no matter. The point is that traffic streams passing through each other are timed very carefully. If a valve should give a problem, loss of power--that throws off the calculations enough that a car that was going to pass your car just behind it but just ahead of the car behind you---well, it no longer cleanly misses, but clips the back of your car. That slows it down where the lane next to you hits it.

Chain-reaction city-johnson...

ShinAce
2012-Apr-02, 11:01 PM
Exactly. It's not new in the least bit.
Mechanically, cars will fail. Personally, people will lapse. Inevitably, any program is incomplete. With all the progress that has happened in the industry, why do some of us still prefer stick shift? I enjoy engine braking and double clutching. Manual is cheap and works. I don't want the fun taken our of my ride. And until we all go automated, enjoy paying top dollar for dedicated roads/lanes.

The liability alone would likely scare any car manufacturer away from making a fully automated car.

LotusExcelle
2012-Apr-02, 11:13 PM
The problem I personally had was that the generic service light came on, but after restarting the engine, would go away. No error code saved. It was probably a 'fuel mixture too lean' error. Nonetheless, I thought it was hilarious that a computer would completely cut throttle to a car that's travelling at 70mph, then not retain an error code. The dealership shrugged it off, and I had the pleasure of having a car randomly die on the highway. That let me know that I cannot depend on a computer in the least bit when it comes to roadworthiness.

Star Trek TNG even did an episode where the simulation said they couldn't fly out. They made it out, without a navigation computer.


I'm not really sure where to begin with this. I'll start with some facts, though. A check engine light cannot come on WITHOUT retaining failure data. You dealership is either lying to you or they don't know what they are doing. The fact that they light went out with a keystroke means nothing - the error went away after a key cycle so the code went inactive. The freeze frame data and historic code are still retained. Under no circumstances would a throttle by wire system cause a lean code (i.e. the accelerator pedal position sensor and the throttle body). There are many other things that would cause that code and that is why a human being must diagnose the cause of such a thing - NOT A COMPUTER.

The only thing I can picture causing the ECM to relinquish control of the throttle body would be a faulty APS or a faulty throttle body. Again if it tripped the light it would store that code and freeze frame data. And again the ECM would only indicate the fault code - it would not in and of itself be the diagnosis as to what is wrong. This is why so many people throw parts at a car and still do not fix the problem. The error in the case you have described is human. The computer did exactly as it is designed to do and is geared towards safely getting your car out of harms way (rather than, say, jamming the throttle wide open in a school zone). A snapped or stuck throttle cable (both of which I have seen) is in one case equally annoying and in the other insanely dangerous. I would much rather have an ECM in between the pedal and the throttle than a cable.

As for your last quoted sentence... there is not a military jet in the sky (that I can think of) that is NOT throttle by wire. They literally cannot fly without computers.

ShinAce
2012-Apr-02, 11:39 PM
I never gave a make and model. Are you telling me you know every detail of every computer for every car ever made? My brother in law owns the dealership. I had the mechanics drive the car three days in a row to reproduce the error. There was no error saved!!!! Yet the day before I'm on the shoulder with an engine bucking wildly. One suggestion I got was leave the car running until I could get a scanner on it. Would you leave an engine that's bucking, running? It amazed me how everyone thought that if we could get the code, the problem would be found. Easiest way to find a problem is to reproduce it, whenever possible.

We worked on the car. The problem is that the manufacturer says that if it starts without any error codes on it, it's fine. No matter what the driver(human) and mechanic(human) feel like. Are you going to throw parts at it? Not me. Who said computers should diagnose? People shouldn't diagnose over the internet either.

LotusExcelle
2012-Apr-03, 12:13 AM
If it is new enough to have throttle by wire - then it is at least an OBDII car. (Or at least I'm assuming it is NOT one of the fairly rare pre-OBDII throttle by wire cars).
Every OBDII (or EOBD depending on which country you are in) has a set standard that is literally universal for every single engine computer. Manufacturers MAY use their own specific codes for additional things but the OBDII standard runs across the board. These computers differ by manufacturer but their fault codes are standardized and their function related to those fault codes is also standardized. I do not have to know which car you own to tell you what P0401 means. It is the same code definition for every car.
That standard also applies to what happens when the CEL comes on. Every OBDII (or, again, EOBD) computer WILL store the fault code. If it goes inactive for any reason it will remain in that computer's history for review *by a qualified technician*. Again freeze frame data should also be available (basically to give you a snapshot of what was going on when that code was triggered... very valuable when doing a correct-process diagnosis).
If your computer is NOT OBDII (being throttle by wire I'd honestly be amazed if it wasn't) then you have a rare vehicle indeed.

swampyankee
2012-Apr-03, 01:27 AM
Very few large aircraft have had a direct connection between the control stick/pedals and aerodynamic control surfaces for decades, even before fly-by-wire. These aircraft had irreversible control systems, where the controls pulled cables which opened hydraulic valves; any "feel" was provided by springs, weights, cams, and other mechanisms. This included all supersonic aircraft, and the only two large aircraft of which I'm aware which did not have irreversible hydraulic controls were early models of the B-52 and the 707.

It was not a huge leap to fly-by-wire. As an aside, have you ever been on a modern naval vessel? I got a tour of a Perry-class frigate. The ship's wheel was about the size of a telephone dial (remember those?), and the only connection between it and the hydraulics moving the rudder were electrical wires.

ShinAce
2012-Apr-03, 01:54 AM
That's all fine and dandy, for a computer to accept human inputs. Even as far back as the 70's you have the F-117 with some serious avionics. Without the computer, no human can fly it. But that doesn't mean we should fly them without pilots either. Just for the sake of not losing it to the enemy you would want a pilot.

Look at surgical mechanical hands. It's fly-by-wire surgery with a medical robot. Now tell me, how many operations are done without a surgeon? If there's a place where spending that kind of money would have returns, surgery would be it.

Even military applications can give more lessons. Look at laser guided bombs. Why do we even need a human to paint the target with a laser when computers can do all of that while we sip margueritas.

Who else thinks that the Matrix and Terminator did not have happy endings?

Jens
2012-Apr-03, 03:50 AM
Who else thinks that the Matrix and Terminator did not have happy endings?

You brought up Star Trek and now two other movies apparently as evidence, but I can't really see what relevance fictional stories have to this. Can I use ET as evidence that bicycles can fly?

swampyankee
2012-Apr-03, 10:04 AM
That's all fine and dandy, for a computer to accept human inputs. Even as far back as the 70's you have the F-117 with some serious avionics. Without the computer, no human can fly it. But that doesn't mean we should fly them without pilots either. Just for the sake of not losing it to the enemy you would want a pilot.

Look at surgical mechanical hands. It's fly-by-wire surgery with a medical robot. Now tell me, how many operations are done without a surgeon? If there's a place where spending that kind of money would have returns, surgery would be it.

Even military applications can give more lessons. Look at laser guided bombs. Why do we even need a human to paint the target with a laser when computers can do all of that while we sip margueritas.

Who else thinks that the Matrix and Terminator did not have happy endings?

The US military has a very active program to develop pilotless combat aircraft: I don't mean RPVs, like the Predator drones, I mean truly pilotless. They're more worried about losing a pilot than they are about losing an aircraft (for one thing, the families of lost pilots can vote, write anti-war letters to Congress, and otherwise complain). As for aircraft with digital FBW systems, the pilot does not really give the computer orders: they give it requests, which are mediated by the computer to prevent pilots from flying outside the envelope (aircraft have envelopes for good reasons, mostly involved with not dying). Similarly, teleoperated systems are going to have some mediation so the system will ignore small amplitude vibrations (tremors) or rapid large-amplitude excursions (oops! I sneezed!) or even be disabled at the operator's command (so the hands at other end don't move).

NEOWatcher
2012-Apr-03, 12:04 PM
On the contrary, a computer that has a map that includes gas stations, road-shoulders and emergency pull-over areas would be much more capable of choosing a proper emergency stopping point than a human who is limited to line of sight.
Really? How often do you use a navigation system? The coordinates may be accurate, but the information tied to them is not. I have had the system miss the destination by hundreds of feet. I have had the system tell me to turn left (not exit, not bear...turn) on a highway 15 miles from any exit. Not even near any bridge.


Thank you for agreeing with me.
That's an acknowledgement of factors into the issue, not an agreement.


Depends on what you're referring to. Some factories and warehouses use robotic forklifts and other mobile machinery that use optical systems to read painted lane markings.
There's a huge difference between a beeping, flashing, walking speed, machine in an isolated, controlled environment than an autonomous high speed vehicle in an ever changing environment.


Other types of automatic driving systems have been demonstrated on-road. We don't need new tech, we just need to adapt existing tech to a different vehicle category with different performance requirements.
"Demonstrated" is the key word. Until they are "demonstrated" in all kinds of conditions repeatedly, then it's not a matter of just adapting.


It might costs more, that doesn't mean it'll be prohibitive.
We don't know that yet.

What sort of maintenance issues are you thinking of, cleaning the optics? You have to do that now.
No I don't. Although I like to keep my car clean, I don't have to do that to keep my headlights turning on at the right time. I don't care if they come on a bit too early or too late because the light sensor is not reading exact light conditions.
And what happens when your optics get dirty while driving? Have you ever seen a car after driving about a mile on a road in snowy conditions that has been cleared with salt? I have a hard time even seeing out of my windshield without using a lot of "blue juice".


ETA: By the way...

Some factories and warehouses use robotic forklifts and other mobile machinery that use optical systems to read painted lane markings.
The company I work for has automated forklifts. You should see all the dents in the safety rails.

ShinAce
2012-Apr-03, 01:23 PM
Jens: since when does a question make for evidence?

It's simply food for thought.

caveman1917
2012-Apr-03, 02:41 PM
That depends no what you mean by "taxi". A driving service/limo service that responds to calls might not need a human driver, but a taxi that roams the street looking for people to flag it down might still work best with a driver. Taxi systems are most common in large cities, and so traffic and ownership issues could be more affected by mass transit, but that's a different debate.

Over here we don't really have the taxis roaming the streets for people to flag it down. They either line up at railway stations and airports, or you have to call them to pick you up somewhere, so that was the system i was referring to, the driving service that responds to calls. I didn't mean so much for intra-city transport, but just consider for what reasons you now own your own car. For me it's commuting to work, doing shopping and general social use such as visiting friends and so. My point was that if the costs of the driver are taken out of the equation, the cost for a taxi company to operate is the ownership of the car itself and the gasoline for it to drive around, which are the same costs you have for owning your own car. However because the base cost of the car is spread out over several people for a taxi company, it might become cheaper for people to rely on those companies for personal transport rather than everyone owning their own car, and still leaving a margin for profit for the company.

danscope
2012-Apr-03, 05:42 PM
The wasted fuel and miles involved with transiting to the fare diminish any efficiency. Perhaps in an urban environment
that would improve. Sending a car 12 miles to run someone 5 miles? Hmmm..

HenrikOlsen
2012-Apr-03, 06:13 PM
On the other hand, sending the nearest car that's free is likely more efficient as (shock! horror! disruption of cultural values! and other calamities!) more people forgo owning a private car which means there are more cars likely to be freed at any time.

Ara Pacis
2012-Apr-03, 06:50 PM
That's exactly it, I feel like working on a hot, combusting, engine.

The problem I personally had was that the generic service light came on, but after restarting the engine, would go away. No error code saved. It was probably a 'fuel mixture too lean' error. Nonetheless, I thought it was hilarious that a computer would completely cut throttle to a car that's travelling at 70mph, then not retain an error code. The dealership shrugged it off, and I had the pleasure of having a car randomly die on the highway. That let me know that I cannot depend on a computer in the least bit when it comes to roadworthiness.

Star Trek TNG even did an episode where the simulation said they couldn't fly out. They made it out, without a navigation computer.

I'm also of the attitude "It's too expensive and too early".

Soon we'll have a computer to make love for you too.

Sounds like you need a better computer, natch.

Ara Pacis
2012-Apr-03, 07:22 PM
Really? How often do you use a navigation system? The coordinates may be accurate, but the information tied to them is not. I have had the system miss the destination by hundreds of feet. I have had the system tell me to turn left (not exit, not bear...turn) on a highway 15 miles from any exit. Not even near any bridge.And your point is that we need more detailed maps with increased accuracy and precision? I could have told you that. That does not mean the technology is not mature. Perhaps it could query Google Map's street level images for verification of data-points. Or, maybe it just means that the car needs to be able to see the road lane markings and signs to decide for itself if something is at variance from expectations like I've been saying all along. Road and roadside construction happens. This is normal.


That's an acknowledgement of factors into the issue, not an agreement.An acknowledgement that the capability is there is all I wanted.


There's a huge difference between a beeping, flashing, walking speed, machine in an isolated, controlled environment than an autonomous high speed vehicle in an ever changing environment. Yes, hence adaptation.


"Demonstrated" is the key word. Until they are "demonstrated" in all kinds of conditions repeatedly, then it's not a matter of just adapting.Perhaps we use words differently. I'm not saying we can put full-fledged cars on the road today with this capability. I'm saying we can start putting the components together today and can rapidly have something that will demonstrate this capability without waiting for advances in new technology. Getting it to market would take a while, bu we're looking more at development than research.


We don't know that yet.That's kinda the point.


No I don't. Although I like to keep my car clean, I don't have to do that to keep my headlights turning on at the right time. I don't care if they come on a bit too early or too late because the light sensor is not reading exact light conditions.
And what happens when your optics get dirty while driving? Have you ever seen a car after driving about a mile on a road in snowy conditions that has been cleared with salt? I have a hard time even seeing out of my windshield without using a lot of "blue juice".Why would the cameras be outside the windshield? A lane marking sensor might do better outside the cabin, closer to the wheel wells, but it can be more robust and take more abuse, and it is a backup to assist the main camera, or maybe it can be built into the headlight lens assembly and look forward instead of straight down. So, again, it may not be a big difference from what you normally do. If it's not good enough for the computer to see, then the car computer will tell you to take control or to clean the optics, which you should do anyways. How can one be a good primary driver if they can't be a good backup-driver/co-pilot.


ETA: By the way...

The company I work for has automated forklifts. You should see all the dents in the safety rails.Another Appeal to Authority? Am I expected to assume that those dents were caused by those automated forklifts and to further assume that they were caused by machine failures instead of human failure? I know a few people who were hit by human operated forklifts, one of which lost a leg, so you need to convince me that it's less safe than human operation.

HenrikOlsen
2012-Apr-03, 07:30 PM
Another Appeal to Authority? Am I expected to assume that those dents were caused by those automated forklifts and to further assume that they were caused by machine failures instead of human failure? I know a few people who were hit by human operated forklifts, one of which lost a leg, so you need to convince me that it's less safe than human operation.
Mandatory youtube (http://www.youtube.com/watch?v=-oB6DN5dYWo) link. Warning, German humor.

NEOWatcher
2012-Apr-03, 08:32 PM
... Perhaps we use words differently. I'm not saying we can put full-fledged cars on the road today with this capability.
That might be the problem. We perhaps use the word "mature" differently.
Yes; the technology is there, and possibly for a prototype, but nowhere near any kind of marketability or public use.


...Why would the cameras be outside the windshield?
I was using "windshield" as an example of how bad optics can be affected anywhere outside of the car in those kind of conditions.
I don't care where you place it, it's going to get grime covered, especially around the front of the car and wheel wells.

... So, again, it may not be a big difference from what you normally do. If it's not good enough for the computer to see, then the car computer will tell you to take control or to clean the optics, which you should do anyways. How can one be a good primary driver if they can't be a good backup-driver/co-pilot.
Yes; big difference. In those conditions, which happen very often around here, virtually nobody would have thier car controlled automatically and were back to square one.
About the only thing to do is to have cleaning equipment around each sensor.
Here's a picture of the type of crap (http://i53.tinypic.com/fcvb6v.jpg) we get.


Another Appeal to Authority?
Oh; for goodness sake. NO... I'm giving you an observation and a little background for why I give the opinion I'm giving.

Am I expected to assume that those dents were caused by those automated forklifts and to further assume that they were caused by machine failures instead of human failure? I know a few people who were hit by human operated forklifts, one of which lost a leg, so you need to convince me that it's less safe than human operation.
Yes; the humans are not in control in the aisles where this shows.
Plus; I am not trying to save that human operation is ultimately safer. Only that computer operation for the application of computer driving is not ready for prime time.

Ara Pacis
2012-Apr-04, 11:03 PM
I was using "windshield" as an example of how bad optics can be affected anywhere outside of the car in those kind of conditions.
I don't care where you place it, it's going to get grime covered, especially around the front of the car and wheel wells.

Yes; big difference. In those conditions, which happen very often around here, virtually nobody would have thier car controlled automatically and were back to square one.
About the only thing to do is to have cleaning equipment around each sensor.
Here's a picture of the type of crap (http://i53.tinypic.com/fcvb6v.jpg) we get. The wipers and the driver are the cleaning equipment. If conditions happen to become too poor for a computer to drive, they are probably too poor for a human to drive. At this point the computer says "blank you, I'm not driving in this, if you want to then I'm not taking responsibility for what happens."


Oh; for goodness sake. NO... I'm giving you an observation and a little background for why I give the opinion I'm giving.

Yes; the humans are not in control in the aisles where this shows.
Plus; I am not trying to save that human operation is ultimately safer. Only that computer operation for the application of computer driving is not ready for prime time.But you still haven't said what caused the collisions between the computer controlled vehicles and the obstacles they collided with. Even if a human is not in direct, real-time control of the vehicles, they often program the machines and the routes and the packing of the material on the pallets, etc.

Ara Pacis
2012-Apr-04, 11:04 PM
Mandatory youtube (http://www.youtube.com/watch?v=-oB6DN5dYWo) link. Warning, German humor.:)

NEOWatcher
2012-Apr-05, 12:30 PM
The wipers and the driver are the cleaning equipment. If conditions happen to become too poor for a computer to drive, they are probably too poor for a human to drive. At this point the computer says "blank you, I'm not driving in this, if you want to then I'm not taking responsibility for what happens."
Are you saying that in conditions where I need wipers and fluid to keep my windshield clean the computer would refuse to drive?
That's about 3 months out of the year here.


But you still haven't said what caused the collisions between the computer controlled vehicles and the obstacles they collided with. Even if a human is not in direct, real-time control of the vehicles, they often program the machines and the routes and the packing of the material on the pallets, etc.
I said "safety rails", no obstacles. The lifts are wire guided and only steering is hands off. I don't know what caused the problems, it's just an observation I thought was worth mentioning.

Ara Pacis
2012-Apr-05, 08:16 PM
Are you saying that in conditions where I need wipers and fluid to keep my windshield clean the computer would refuse to drive?
That's about 3 months out of the year here.No, I mean conditions where wipers and fluid are not adequate for safe driving.


I said "safety rails", no obstacles. The lifts are wire guided and only steering is hands off.Rails are a subset of obstacles.


I don't know what caused the problems, it's just an observation I thought was worth mentioning.Without explanation, it really isn't.

JustAFriend
2012-Apr-05, 10:18 PM
The real question is: Will they get Robert Picardo to do the voice for the Johnny Cabs?

http://media.photobucket.com/image/johnny%20cab/spacemonkey_fg/Blog%20Pictures/Robots21.jpg

danscope
2012-Apr-06, 02:51 AM
More likely they will get the voice from " Mission Impossible " . ....." The car you are about to enter may self-destruct
at any time . Good luck , Jim . "

swampyankee
2012-Apr-06, 10:23 AM
More likely they will get the voice from " Mission Impossible " . ....." The car you are about to enter may self-destruct
at any time . Good luck , Jim . "

A possibly apocryphal story is that GM would test Vegas to determine component lives. The goal was to have all components with expected life spans of 50,000 miles, neither less nor more. The obvious* conclusion was the desire was for the car to, like the wonderful one-horse shay, to just disintegrate completely.


---------------

* And wrong.

NEOWatcher
2012-Apr-06, 11:58 AM
No, I mean conditions where wipers and fluid are not adequate for safe driving.
Then in that case, I agree. It does increase the complexity and cost of a sensor system, but that's a different topic since we have not be estimating any kind of costs. We'd need heaters (frost buildup), wipers (crud), Fluid(crud streaks), and I don't know enough to say how just plain water affects certain sensors, but it is something I'd like to see.
Now I can see those people that drive around with just a little peephole scraped off on their frost covered car not even bothering to do even that. Not a problem if they don't need to interact, but it is a scary thought.


Rails are a subset of obstacles.
Then we have a different interpretation of obstacles (http://en.wikipedia.org/wiki/Obstacle). Since a safety rail is not a hinderence to the intended path of the lift, I don't consider it an obstacle.


Without explanation, it really isn't.
It is when it factors into how I feel about the situation.

swampyankee
2012-Apr-06, 04:04 PM
If we're going to compare computer-driven vs humanly driven systems, let's at least compare the real-world performance of people vs the real-world performance of humans, like fork-lift drivers. In every warehouse and factory (and it's a fair number -- a dozen or so) I've been, there have been scrapes, scratches, and dents where human-driven fork-lifts have met racks, walls, and fixtures. Everybody can see the real-world performance of human drivers of motorcycles, cars, and trucks (UK heavy-goods vehicles) on the evening news. That's what computers have to beat. Do note, too, that many of those features on the evening news are the result of people who think that they are much better drivers than they really are.

LotusExcelle
2012-Apr-06, 04:13 PM
Many sensors required to *support* or augment such a vehicle exist already. Passive and active warning systems are meant to keep a distracted (or inept) driver from drifting lanes or from rear-ending someone in traffic. They are not, by any means, perfect. Nor should they be *yet*. But the fact is that they do exist. Plus more and more vehicle systems are becoming fly-by-wire. Mercedes has had a brake by wire system for years (possibly other manufacturers). This is not a question, in my mind, of whether it is possible or probable. It is merely a question of when.

Ara Pacis
2012-Apr-06, 07:00 PM
Then we have a different interpretation of obstacles (http://en.wikipedia.org/wiki/Obstacle). Since a safety rail is not a hinderence to the intended path of the lift, I don't consider it an obstacle.Why do you keep wanting to get into a semantics argument you're going to lose instead of just answering straight-forward questions or admitting that your argument is flawed? Fine if you want to argue that something is only an obstacle when it gets in the way of where something is going then you're still wrong, because the forklift was going somewhere when it was stopped/deflected by something that was in its path, hence an obstacle. It doesn't matter where you wanted it to go, but where it was going. Your link only reinforces what I say because otherwise a tennis court net would not be an obstacle because it stops balls that are on trajectories that you don't want them to take.


It is when it factors into how I feel about the situation.How you feel about the situation has little effect on the reality of physics. If you disagree, ask a mod to take your posts here into the ATM forum.

NEOWatcher
2012-Apr-06, 07:23 PM
Why do you keep wanting to get into a semantics argument...
I don't mean it as a semantics argument. I didn't say you were wrong, I was just pointing out how I read it and why I interpreted it different. Simple as that.
Speaking of semantics argument, why do you persist on further adding semantic arguments if you don't want one?


How you feel about the situation has little effect on the reality of physics. If you disagree, ask a mod to take your posts here into the ATM forum.
No; I am stating an opinion, and I am explaining my reasons for opinion from what I have observed.

Where did I get physics wrong?

danscope
2012-Apr-06, 09:04 PM
It is difficult for those infatuated with technology to accept the possibility of misapplication of electronics in a hostile and unforgiving environment. Sometimes it's just hard to get rid of the pilot in command.

Ara Pacis
2012-Apr-06, 11:31 PM
I don't mean it as a semantics argument. I didn't say you were wrong, I was just pointing out how I read it and why I interpreted it different. Simple as that.No, you meant it as an evasion to admitting my point that your point doesn't hold water unless you can actually tell me the cause of the collissions.


Speaking of semantics argument, why do you persist on further adding semantic arguments if you don't want one?It was a continuation of an existing line of argument that shows how even your attempt at evasion is flawed, like the rest of your arguments (which you seem to have abandoned, because they were flawed, I assume).


No; I am stating an opinion, and I am explaining my reasons for opinion from what I have observed.You were stating a cause and effect as a fact of some worth. If you want to only claim it's an opinion about how you feel, that's fine, but it holds no weight as to the physical capabilities of such a system. Why can't you just admit when you're wrong and concede an argument instead of trying to change it into something like facts, which are arguable, into mere opinions/feelings, which aren't?


Where did I get physics wrong?How you feel about the applicability of the physical systems under discussion has no effect as to the actual capabilities of the system. So why mention it, except as an evasion?

Ara Pacis
2012-Apr-06, 11:33 PM
It is difficult for those infatuated with technology to accept the possibility of misapplication of electronics in a hostile and unforgiving environment. Sometimes it's just hard to get rid of the pilot in command.You mean how it's hard to get you out the driver's seat when you are infatuated with the steering wheel? :)

Seriously, all of us here are taking the issues seriously and suggesting real answers. You're the one who's relying on emotion to judge if it can or can't work.

danscope
2012-Apr-07, 01:58 AM
My answer is the mud puddle. Your sensors are gone. You die. End of game.
But you don't get it, I guess. What part of failure mode do you understand? And yes, when some person suggests that I cede the safe operation of vehicles over to robbie the robot, and see my family at severe risk, yep. I might get a little bit emotional. Little bit there.
Not for nothing, but this scheme will get you laughed out of the club house. Really .
Impractical and bloody dangerous is no way to engineer life,son.

LotusExcelle
2012-Apr-07, 02:18 AM
Humans aiming 2 ton vehicles at each other while texting is impractical as well.

swampyankee
2012-Apr-07, 04:12 AM
OK; you hit the mud puddle, your vision is gone, you die. End of game. Where's the difference? Failure modes get analyzed and worked around. It's not trivial -- this is why flight control software is very expensive -- but the failure modes are all predictable.

Van Rijn
2012-Apr-07, 07:07 AM
My answer is the mud puddle. Your sensors are gone. You die. End of game.
But you don't get it, I guess. What part of failure mode do you understand?


Okay, you're arguing safety. So what happens if you're shown that automated driving would reduce deaths by a factor of a hundred? Would you accept it then, or would you still say no?

That's the key reason people are interested in this, after all: To significantly reduce the number of people getting hurt.

danscope
2012-Apr-07, 03:32 PM
If people are interested in significantly reducing accidents, they would stop handing out driver's licenses with pounds of tea. And perhaps we could demand a lot more hours of driving instruction and some personality screening. There are clearly some people who shouldn't and ought not drive a vehicle. Put your money into that.
Drivers aren't born. It's an acquired skill and discipline. No question.
I don't believe in this robotic scheme. But if you think I am a tough sell, try the insurance companies. :)

Best regards,
Dan

profloater
2012-Apr-07, 03:46 PM
Having just driven a couple of thousand miles again, we could really do with robot intervention to stop the habit of drivers driving too close to the one in front and similarly overtaking and cutting in,. That target seems eminently feasible with car mounted radar and radar reflectors. It would be in my opinion much more effective than speed limits, which encourage bunching. The cost per vehicle would be quite small if just a warning is given and if necessary the system can store offences and "tell" on the drivers. If this system is allowed to mature, it would make the next step of full automation much easier.

LotusExcelle
2012-Apr-07, 05:10 PM
This might be the only thing I agree with you in regards to this thread, Danscope: That drivers are generally unskilled and at the very least need more training. The same holds for motorcycle licensing - I think MSF courses should be mandatory (possibly every X years). However this is, I think, a separate matter from computer control. The human element is by far the most unpredictable and dangerous part of the equation and taking the human out or mostly out of it means higher safety potential.

There is a company called Mobileye that currently builds and markets (retail and OEM) a warning system that solely uses an optical camera. Its pretty fascinating and manages to package many of the requirements that could lead to computer control into a computer warning system.

LotusExcelle
2012-Apr-07, 05:19 PM
Sorry- realize I should provide some links for the stuff I mentioned:
MSF Front Page (http://online2.msf-usa.org/msf/Default.aspx)
MSF Rider Course Info (http://msf-usa.org/index_new.cfm?spl=2&action=display&pagename=RiderCourse%20Info)
Mobileye Homepage (http://www.mobileye.com/)
Mobileye example video (youtube link) (http://www.youtube.com/watch?v=yoY5tMlrgRw)

swampyankee
2012-Apr-07, 05:28 PM
One thing that would easily improve driver safety: require all cars to have data recorders. Disabling the recorder would result in prison time, a large fine, and a permanent license revocation. Oh, and the car won't start.

Van Rijn
2012-Apr-07, 06:25 PM
Danscope, I noticed you avoided my question. Again:


So what happens if you're shown that automated driving would reduce deaths by a factor of a hundred? Would you accept it then, or would you still say no?

danscope
2012-Apr-07, 08:57 PM
Hi Van, I applaude the employment of proximity detectors as alarm sensors. They could automaticaly apply brakes and withold the vehicle from tailgating. That is doable. And I have no doubt that we could employ gps sensors which would limit the velocity of any motor vehicle as a simple governor. Do this. Provide the very real, sure and stiff penalties which have been mentioned, and yes, we NEED to cull the dangerous fools who insist on weaving through traffic at 85 MPH
just because they fancy themselves as racing drivers ....(while they gab on their cellphones or text ).
The temporary or even permanent loss of priviliges ensures respect for the public.
I live on a winding , curvy road which has a speed limit of 25 MPH for a darn good reason, especially owing to blind driveways and pedestrians, bicycles, strollers etc. Last week, we had two imbeciles racing each other. They hit the large maple tree in front of my house, left brake skid marks of 135 feet . Not good. Car is junk. Those two kids will be making a court appearance on reckless driving charges quite soon.
A gps system could sense what the posted speed is. Try that for technology. You know, that COULD work.
That makes sense. Speed and irrational thinking are the worst offenders. Work on that .

LotusExcelle
2012-Apr-07, 09:05 PM
http://en.wikipedia.org/wiki/Precrash_system

LotusExcelle
2012-Apr-07, 09:10 PM
See also "adaptive cruise", "speed limiter", etc etc. These systems already exist and have been in use for many years.

neilzero
2012-Apr-08, 03:01 AM
We have a gps with that feature. Every few minutes it says, "you are exceeding the speed limit"
I'm concerned that my automated car will go to maximum breaking, when someone cuts into my lane, ahead of me, missing me by a few meters, and I will be rear ended by the car behind. My usual reaction is to brake lightly, which could prove an error if the car that cut in front of me goes to max breaking. Neil

HenrikOlsen
2012-Apr-08, 08:38 AM
Since it by necessity has to have sensors both front and back if it's to do automated distance-keeping, that's not a scenario I can see the computer do.
You're really applying a human-style panic reaction to a computer controlled scenario here.

There is one worry I'd have though and that's someone who's decided that he hates computer aided cars deliberately targeting the computers algorithms by boxing it in until it has to used violent maneuvers to get out, then using that violent maneuver when suing in court.

swampyankee
2012-Apr-08, 01:05 PM
We have a gps with that feature. Every few minutes it says, "you are exceeding the speed limit"
I'm concerned that my automated car will go to maximum breaking, when someone cuts into my lane, ahead of me, missing me by a few meters, and I will be rear ended by the car behind. My usual reaction is to brake lightly, which could prove an error if the car that cut in front of me goes to max breaking. Neil

Just because computers are based on binary logic doesn't mean that the choice would be "slam on the brakes" or "do nothing." The software* would make a few calculations and decide what level of response is needed, in the same way that the wetware between a person's ears does it. The computer would be running code, the person would be subconsciously choosing between a bunch of options based on rules that have been learned by experience ("oops! should have braked harder that last time!"). The sensor is going to be a bit more sophisticated than the photoelectric sensor that keeps a garage door from squishing Fluffy -- it will measure distance and velocity (or at least the radial component of velocity).

As for the "my usual reaction..." Yep. People do that "max braking" bit deliberately** now, as hitting a car from behind is almost always considered prima facie evidence of following too closely. Sometimes the people who do this disable their brake lights, too, in which case you will need extra time to decide the car in front of you is braking. Remember, it takes about a second for a driver to respond to an unexpected event. Throw in a second unexpected event -- the failure of the brake lights on the car in front to work -- and the reaction time increases.


------------------

* It's not rocket science to design a system that will set the braking effort as a function of distance and closing rate. Indeed, the only reason that it couldn't be done in a completely analogue, hydro-mechanical system is probably the sensor. I can't think of a non-contact hydraulic or mechanical sensor that could get closing rate and distance at more than a few millimeters distance.

** Given the very limited amount of investigation that will be done on an accident involving no fatalities and only a few hundred dollars in damages, this is a commonly used method of picking up some loose change from insurance companies. The insurance company then gets it back (usually with usurious rates of interest) by raising the rates of the person who did the rear-ending.

adapa
2012-Apr-08, 01:39 PM
Although humans are not perfect, machines are not perfect either (and certainly not automation systems). For example, the primary cause for the the Turkish Airlines B-737 accident (http://aviation-safety.net/database/record.php?id=20090225-0) (TK1951 into Amsterdam) was that the distracted crew did not respond in a timely manner to prevent the automation from crashing the aircraft. Also, I have seen my fair share of automation and Flight Management Computer (FMC) malfunctions (as have many other pilots). These are usually no problem when the pilot reduces the level of automation, but the results would be much less benign if intervention was impossible.

Although these above examples are aviation related, I believe that they are very relevant to the topic of this thread. Also, consider that this happens with systems that have been thoroughly tested before being certified.


Another example that highlights my concern is the large number of cars that have been recalled due to software malfunctions as indicated here (http://gizmodo.com/5467388/software-bug-causes-toyota-recall-of-almost-half-a-million-new-hybrid-cars), here (http://www.geek.com/articles/geek-cetera/software-bug-stops-cruise-control-being-turned-off-jaguar-recalls-17678-x-type-cars-20111024/), and here (http://www.zdnet.com/blog/projectfailures/cadillacs-recalled-over-software-bug/1140). Now, if we want to take the human totally out of the control loop, then there needs to be adequate reason to believe that automation and navigation software are somehow immune from the bugs that can affect other types of software. So far, I see no such reason.

On another note, there have been some cases where people blindly followed their GPS directions into situations (http://www.technolog.msnbc.msn.com/technology/technolog/death-gps-could-it-happen-you-125321) that are unsavory (http://www.nodeju.com/4036/gps-directs-canadian-driver-lake.html).
These are examples of poor judgement. However, is there strong reason to believed that the automation would have done much better, especially if it uses the GPS as its primary (possibly only) navigation source?

I am not at all against technology and I believe that it certainly has its place. However, I do not believe that it can be a substitute for sound judgement and common sense.

swampyankee
2012-Apr-08, 01:57 PM
Adapa, I'm not arguing with you; I'm arguing with somebody who seems to believe that technology can't do any of these things. I've worked in aerospace (my signature is on the certification paperwork for a commercial engine, several dozen reports recommending replacement intervals for helicopter control system components, as coauthor of a couple of papers, and I've been involved, at least at the periphery, of a couple of accident investigations) and software development. I don't think I have an unrealistic view of software capabilities, nor do I have a terribly unrealistic view of human capabilities (the worst accident investigation of which I was on the periphery involved pilot error compounded by a design flaw by the airframe maker [I worked at one of their suppliers], and no, I won't go into more detail).

danscope
2012-Apr-08, 02:32 PM
But employing such technology puts the end user into a seemingly benign world of disinterest and distraction ,worse than an aircraft who still has a pilot in command and is keeping a "good watch" visually and otherwise.
"Letting the car do it" as some suggest invites trouble and worse.

Dan

cjameshuff
2012-Apr-08, 02:44 PM
Although humans are not perfect, machines are not perfect either (and certainly not automation systems). For example, the primary cause for the the Turkish Airlines B-737 accident (http://aviation-safety.net/database/record.php?id=20090225-0) (TK1951 into Amsterdam) was that the distracted crew did not respond in a timely manner to prevent the automation from crashing the aircraft. Also, I have seen my fair share of automation and Flight Management Computer (FMC) malfunctions (as have many other pilots). These are usually no problem when the pilot reduces the level of automation, but the results would be much less benign if intervention was impossible.

They don't need to be perfect, they only need to be better than human drivers. And humans are really pretty awful drivers. Not just the bad drivers, as danscope seems to think...the best human drivers have nowhere near adequate perceptions and reflexes for the job, and are prone to counterproductive panic reactions.

Automobile autopilots also have some major advantages over aircraft autopilots. They can just decide to stop if they encounter trouble, while an aircraft system has to keep operating and keep the aircraft in flight until it can get a human pilot to take over. If there's no place to pull over, this is going to annoy some people...unfortunately for danscope's argument (which is nothing but random scenarios he's cooked up where a computer driver unrealistically causes deaths), annoyance rarely kills.

Some sensors start giving suspicious/inconsistent data, it slows the car down and sounds an alarm. If the driver doesn't take control, it stops the vehicle by the side of the road or right in its lane. If the driving computer experiences a freak total failure, the fallback system (sorry danscope, I won't assume people designed the system to kill people) brings the car to a controlled stop using independent internal accelerometers and gyroscopes to keep it in its lane. If that goes too, odds are there's not enough electronics left to even keep the engine running...and we are no longer talking about a failure mode specific to computer driving systems. The closest thing to a failsafe there is probably to design the brake systems to go to full braking in such an event and hope any cars following the vehicle are on automatic so they don't rearend it.

LotusExcelle
2012-Apr-08, 03:29 PM
Some sensors start giving suspicious/inconsistent data, it slows the car down and sounds an alarm. If the driver doesn't take control, it stops the vehicle by the side of the road or right in its lane. If the driving computer experiences a freak total failure, the fallback system (sorry danscope, I won't assume people designed the system to kill people) brings the car to a controlled stop using independent internal accelerometers and gyroscopes to keep it in its lane. If that goes too, odds are there's not enough electronics left to even keep the engine running...and we are no longer talking about a failure mode specific to computer driving systems. The closest thing to a failsafe there is probably to design the brake systems to go to full braking in such an event and hope any cars following the vehicle are on automatic so they don't rearend it.
This is essentially how I'm picturing it as well. A car can report, several miles away, that it is disabled. Other cars in the area can ad-hoc the report down the line so that prior to a visual (the best a human can currently do), the *car* knows there is a disabled vehicle. It can adjust accordingly and likely without loss of speed. The car that is disabled safely brought its occupants to a stop after a serious fault and the rest of the vehicles respond before any human *could* be aware there is an issue. That type of system alone would be invaluable - each car reporting telematics to every other car. If one slips on ice, the others know to alter speed and course. The safety would, in my mind, increase exponentially. And insurance rates/cost to insurance companies would respond accordingly as well.

Perfect? Never. As cjameshuff says - it just needs to be better.

swampyankee
2012-Apr-08, 03:36 PM
But employing such technology puts the end user into a seemingly benign world of disinterest and distraction ,worse than an aircraft who still has a pilot in command and is keeping a "good watch" visually and otherwise.
"Letting the car do it" as some suggest invites trouble and worse.

Dan

The trouble is that many people seem to have the attitude of "let other people watch out," or "I'm so good, I can do anything," or "I'm always 100% alert." Do note that commercial pilots -- the ones we pay (not enough) to control large plane-loads of people do get a lot more assistive technology than drivers.

danscope
2012-Apr-08, 08:09 PM
They certainly get more and better instrumentation, as well as a co-pilot who is every bit as capable as the pilot in command.
But to rely on electronics to pilot an automobile??????? Really? It is far easier and more prudent to screen the people
applying for automotive licenses .
Like I said before: apply proximity technology and velocity reduction . That we can do without much cost and with confidence.
Like Moe said...." spread out! ".

Best regards,
Dan

Jens
2012-Apr-09, 02:40 AM
It is far easier and more prudent to screen the people
applying for automotive licenses.


I don't think so. We screen people already, and people have lots of accidents. I think that in the long-run, it will be better to use automation. I also think an added benefit is that people will have time during commuting to read or do work or whatever instead of having to concentrate on watching the road. I commute to work in a vehicle that I don't have any control over (though there is a driver), and I use the time to do stuff I want to do.

danscope
2012-Apr-09, 04:34 AM
Hi Jens, I have been around some of the most expensive electronics ever devised, and built to MILSPEC which a darn sight higher quality than you would ever see in automotive applications. Trust me. And that fails at times.
You don't have to look any further than the Space Shuttle. Quality hardware and software. And..... just how many
stalled launches because of sensor failures etc etc etc etc.....
When a rocket fails, we can execute an abort command. When your misguided car fails , you are going to take out
a bunch of cars and people.
I have zero faith in this scheme... for many reasons.
Best regards,
Dan

Jens
2012-Apr-09, 06:00 AM
When a rocket fails, we can execute an abort command. When your misguided car fails , you are going to take out
a bunch of cars and people.


Not necessarily. With a car, if you have a technical failure it's fairly easy to have a backup system that will stop the car and bring it to the side of the road to avoid harming other cars. With a rocket, the options are much more difficult. When you have a technical failure during a launch, in many cases it is not possible to abort. So no, you are not going to "take out" a bunch of people.

Just in case you are wondering, I can almost assure that you that nobody arguing that such a system is worth exploring would say that we should put one into place without testing. If it ends up being safer than human drivers, go with it, and if it ends up being more dangerous, don't.

Van Rijn
2012-Apr-09, 07:26 AM
Hi Van, I applaude the employment of proximity detectors as alarm sensors.


That's not an answer to my question. You have avoided answering this simple question twice:



So what happens if you're shown that automated driving would reduce deaths by a factor of a hundred? Would you accept it then, or would you still say no?


So why won't you answer? Is it because the answer is "NO"? Even if it were shown to you that it was much safer, you still wouldn't accept it, would you?

NEOWatcher
2012-Apr-09, 12:26 PM
No, you meant it as an evasion to admitting my point that your point doesn't hold water unless you can actually tell me the cause of the collissions.
Fine. I have a feeling, it's based on unsubstantiated observations. But; it's a feeling based on many experiences in my lifetime, and I won't go into detail to what those are.


How you feel about the applicability of the physical systems under discussion has no effect as to the actual capabilities of the system. So why mention it, except as an evasion?
You said physics. Not application of physics.

How you feel about the situation has little effect on the reality of physics. If you disagree, ask a mod to take your posts here into the ATM forum.
And before you dismiss this as a semantic argument, you also framed it with the idea of ATM.
Man made systems fail, it's a fact. Not always, but it does happen. We can do what we can to avoid failures, and many times the effort is successful enough to where the failures are miniscule or eliminated with enough backup modes.

I have repeated again and again in this thread. I'm not saying the idea can not work. I am saying that I am not convinced it is ready.

adapa
2012-Apr-09, 04:07 PM
Adapa, I'm not arguing with you; I'm arguing with somebody who seems to believe that technology can't do any of these things. I've worked in aerospace (my signature is on the certification paperwork for a commercial engine, several dozen reports recommending replacement intervals for helicopter control system components, as coauthor of a couple of papers, and I've been involved, at least at the periphery, of a couple of accident investigations) and software development. I don't think I have an unrealistic view of software capabilities, nor do I have a terribly unrealistic view of human capabilities (the worst accident investigation of which I was on the periphery involved pilot error compounded by a design flaw by the airframe maker [I worked at one of their suppliers], and no, I won't go into more detail).
Actually, I did not notice your post until I made mine. I recently saw this thread and decided to put in my $0.02 worth.

danscope
2012-Apr-09, 08:03 PM
I answered. You apparently do not appreciate the grave danger involved with failure mode in a non-benign environment.
My background precludes any faith in such a scheme. We simply differ in our value structure in the absolute relying on electronics to somehow build a marshmallow world where we replace responsibility with plastic robots and let George Jetson's dog do the driving. But like I said; it's the insurance companies you will have to sell, not just me.
In my opinion, they will look at this and quote the disenchanted casting director with his shallow....'Thank you' .
I doubt that Astro will get the part.
Best regards,
Dan

ShinAce
2012-Apr-09, 08:34 PM
There is some good and bad in automation. Like it or not, cars will continue to integrate electronics. Imagine a collision detection system that inflates air bags early and gently. Back up sensors will eventually have access to the brakes. Step by step we will lose control of our vehicles.

But total automation is like cold fusion. Always 10 years away.

Humans suffer from lack of focus. Electronics suffer from interference. Alone, each will fail eventually. But if both can be used, the system can be made safer. So how much control does the driver need? How much control can you afford to give without comprimising automation. There is no perfect middle ground.

Inclusa
2012-Apr-10, 08:00 AM
There is some good and bad in automation. Like it or not, cars will continue to integrate electronics. Imagine a collision detection system that inflates air bags early and gently. Back up sensors will eventually have access to the brakes. Step by step we will lose control of our vehicles.

But total automation is like cold fusion. Always 10 years away.

Humans suffer from lack of focus. Electronics suffer from interference. Alone, each will fail eventually. But if both can be used, the system can be made safer. So how much control does the driver need? How much control can you afford to give without comprimising automation. There is no perfect middle ground.



As a Linux user, someone has complained about the inability of certain distros to login as a root user; while there are many benefits of being a root user, an over-confident user may wreck the own system.
Although there are limitless chances of reinstallations for Linux, this can result in loss of time and productivity.
How much control should an user have? For prestige and pride, humans have a tendency to overestimate their capacities, though.
Does anyone remember the Costa accident? We may mention that any competent captains will not make such a grave mistake, but we should always remember: even the least incompetent Homo sapiens show moments of incompetency.
Machines and computers fail, but the chance is generally way lower than that of humans. (in general).

danscope
2012-Apr-11, 12:59 AM
It wasn't incompetence. It was arrogance....in a big way. He'd probably had a few drinks as well. Commanding big ships and alcohol is never a good idea. We went through this with the Exon Valdez. :(

Ara Pacis
2012-Apr-12, 09:21 AM
danscope, how does being milspec affect the quality of the electronics? IIRC, they are more robust and thereby reliable for certain demanding application, but does that mean they are the fastest or smallest or newest?

Shinace, why is total automation in a car always 10 years away? The idea of a smartphone seemed like that too, and then Steve Jobs got the iPhone produced. Small evolutionary changes, like the backup sensors or the automatic braking or self-parking that are incremental are just that, adaptations to serve the market. Maybe if they just sat down and tried to build a fully automatic car from the start, it'd get made.

swampyankee
2012-Apr-12, 03:36 PM
Right now, I don't think there is any market pressure for car with sufficient automation to drive itself, partly because of the overhyped thrill of driving (anybody here drive on I-95 in western Fairfield County or the metro NYC area?). The concept of a computer -- a few grams of carefully modified silicon -- and associated sensors could do better is anathema to that idea, whether or not it's achievable.

I happen to think it's achievable, albeit difficult, as the required algorithms and programs may not be trivial, but they are not impossible. Indeed, some of the pieces have been created. For example, swarming behavior and collision avoidance in three dimensions has been demonstrated in small flocks of robots (see http://www.ted.com/talks/lang/en/vijay_kumar_robots_that_fly_and_cooperate.html). DARPA has also funded a driverless car program, where cars have driven around an unknown course.

Jens
2012-Apr-13, 12:08 AM
Yesterday in Kyoto there was a driver who had a seizure and ended up killing 9 pedestrians. Even if total automation will take some time, it shouldn't be impossible to implement some kind of system like the ATS system in trains, which would force cars to stop at red lights, or a kind of "dead man's switch". It may be that automatically stopping a car on a highway would be dangerous, but having a car on a highway with an incapacitated driver is surely more so.

danscope
2012-Apr-13, 04:43 AM
Or an incapacitated computer control modules or any nefarious weakest link in such a complex system.
The safe operation of every car hangs by a thread, a sort of "sword of Damocles" .

Ara Pacis
2012-Apr-13, 08:42 AM
Or an incapacitated computer control modules or any nefarious weakest link in such a complex system.
The safe operation of every car hangs by a thread, a sort of "sword of Damocles" .

And this means what exactly? So you think automation is like a chain, only as strong as its weakest link and any failure must be catastrophic? Or perhaps you're espousing a more reasonable proposition that a system suffering a malfunction is simple less than safe. If the latter, then you must realize that many systems in current automobiles operate at less than optimal. Brakes wear down, tires wear down, motor oil breaks down, lights go out. You know what happens when a headlight goes out? You drive with one headlight instead of two. Kinda like how when you lose the use of one of your hands, you drive with one hand instead of two.

Or you could tell us that the data reveals that computer control module failures occur more often than human failures and/or result in more severe consequences. If the data supports that. Does it?

swampyankee
2012-Apr-13, 09:51 AM
One of the reasons that commercial aviation is safer now than it was decades ago is the computer aids that pilots get, which include ground proximity warning systems, collision avoidance systems, stability augmentation, autothrottles, flight directors, an autopilots. Indeed, Category III landings cannot be made without autopilots (see http://www.airliners.net/aviation-forums/tech_ops/read.main/183032/).

I feel (yes, emotionally statement) that much of the objection to computers replacing drivers is emotional, so it's impossible to overcome this by logical argument. Similarly, people are adamantly opposed to "black-box" recorders in cars (I'm not referring to real-time telemetry, but only to data recorded and kept with the car).

adapa
2012-Apr-13, 02:36 PM
When we use data, it is important to know how it is collected and interpreted. This applies to accident and incident statistics that involve operators and automation. I will point out cases with aviation again because it has some of the most extensive uses of automation.

1. If the pilot does something wrong that causes an accident, then the accident is blamed on pilot error (as we should expect).

2. If the automation fails and the pilot failed to prevent it from crashing the aircraft, the accident is blamed on pilot error.

3. If the automation fails and the pilot prevents it from crashing the aircraft but it exceeds acceptable limits (aircraft or airspace), it is now an incident which will be blamed on pilot error.

4. If the automation fails and the pilot prevents it from crashing the aircraft or exceeding any limits, then there is no incident or accident to report and therefore nothing gets reported (although the pilot will probably make an entry into the maintenance log book).


It is also interesting to note that the majority of professional pilots (myself included) have lengthy careers without any incidents or accidents, yet one would be hard pressed to find an experienced pilot who has not experienced several automation malfunctions over the length of his/her career. It is very likely that many people overestimate the reliability of machines and underestimate the reliability of responsible people.


On a similar note, many people tout the impressive flying display of the quadrotor flying robots. However, fewer people mention the bloopers (http://www.youtube.com/watch?v=TVrxvqYlCDs).





Indeed, Category III landings cannot be made without autopilots (see http://www.airliners.net/aviation-forums/tech_ops/read.main/183032/).
Actually, this is inaccurate. In the airline that I work for, Category III approaches and landings in the Boeing 737-800 are hand flown using the Heads Up Display. Also, please read the link that you posted again (the person implied most cases but not all). For aircraft that use autoland, one pilot is required to have his/her hands on the flight controls and be ready to immediately take the aircraft from the autopilot if needed. Otherwise, any resulting accident or incident would be blamed on pilot error (see above).

While I am on the topic, I may as well dispel the myth (believed by the general populace) that the automation can always do a better landing than a human operator, because that is not necessarily the case. Every once in a while, an aircraft (such as a B767) will need to demonstrate an autoland in order to maintain its Category III certification. Although the automation landed the aircraft safely (well within safe parameter limits), in many cases I would be embarrassed if I did a landing like what the automation demonstrated.

danscope
2012-Apr-13, 04:42 PM
Yes, but those systems aren't in control of steering and braking autonomously. A driver in command can apply an emergency brake. You still have steering even when the power steering fails or engine stops. But computer driving?
shrug.
By the way: @ Adapa; I salute you and your profession. You serve us well every day.
Best regards always,
Dan

swampyankee
2012-Apr-14, 03:13 AM
Adapa, like I said, I was in aerospace, not aviation, so I'll admit error about Cat III landings.

Very few accidents are the result of just one problem, for example bad ergonomics is frequently a factor which contributes to pilot error, and that includes badly laid out controls and instrumentation and poorly planned procedures.

Van Rijn
2012-Apr-14, 05:41 AM
I answered.


You evaded the question three times.


You apparently do not appreciate the grave danger involved with failure mode in a non-benign environment.
My background precludes any faith in such a scheme.

No faith is required. I asked if you would accept it if you were SHOWN it would reduce deaths by a factor of a hundred? Any such determination would be done through testing. It's also a conservative estimate: When you remove driving under the influence, distracted driving, driving error, limits to human response time, and so on, leaving just the occasional animal running accross the road, rock slides or other events not caused by people, there aren't many accidents left to deal with. This is something that can lead to radical reduction in auto accident rates. I expect that after a decade or so for it to become established, attempting in-city manual driving would get stiff penalties and possible jail time.

swampyankee
2012-Apr-14, 12:41 PM
It's interesting that the one transportation mode which has the least access to computer assistance for vehicle operators has the highest accident rate.

Commercial pilots get collision avoidance and ground proximity warning indicates; all pilots of powered aircraft have the potential to get autopilots, which can handle many aspects of routine aircraft operation.

On many (most?) systems, trains stop if an operator ignores a stop signal.

Even sailboats have self-steering systems.

Cars and trucks? Usually ABS, occasionally traction control, rarely, and only recently, anything more, like adaptive cruise control


Incidentally, I'm ignoring such long standing automation as automatic spark advance, automatic transmission, electronic fuel injection, and speed-sensitive power steering. Even more, I'm ignoring the automation that would make car operation impossible. Do you really think anybody can automatically actuate valves? Automation has never required computers -- just look at a Brown & Sharpe screw machine ;)

Solfe
2012-Apr-14, 02:16 PM
I would offer the opinion that "automated cars" would need to have all of their computerisation and automated systems re-evaluated and retooled for human reading and intervention. I would hate to see a very expensive vehicle permanently parked for a "check engine" light.

My reason for this opinion is my wonderful car that refuses to pass inspection due to the emissions system. The check engine light came on one day and since the actual problem is the whole system not reporting when it is connected to a computer, I am down to replacing components at random. This is actually so expensive, there is a special waiver for it. My car hasn't had a normal inspection in years.

Silly way to build anything: building an interface to test the thing that tests the sensors monitoring the system. Direct output from the sensors is far better when in diagnostic mode. I don't think you can easily do that with cars today.

DonM435
2012-Apr-16, 01:58 AM
The Check Engine light must be the worst invention of the 20th century.

danscope
2012-Apr-16, 02:16 PM
Hi Van, Currently on vacation on Cape Cod. You keep bringing up " Shown" as in "Proven" that such a scheme is infallible.
I keep saying and will maintain that every electronic system, and it's failure mode is a function of it's wiring harness, it's individual components( of which there are many many many parts, along with printed circuit boards (of which I have repaired and serviced) along with the myriad connectors( ever solder an80 pin amphenol connector) ; and all of that along with servo and feed back systems makes for an
inordinately complex and as such prone to failure system, which when such happens will KILL YOU and other innocent victims in the process.
As an aside, we are moving into a world where fewer and fewer people can afford a new car or even relatively new car, and you would have us double the cost of and maintenmance of these "wundercars" that can magically drive us about without a care in a world.
This scheme, in my opinion has no merit. It would work on a monorail for it works in one dimension only. It's position on rail and proximity to the next car and next station or mid-station is simple enough to detect and alaso to back up. No question. That is the way to go.


Best regards,
Dan

Ara Pacis
2012-Apr-16, 09:17 PM
I keep saying and will maintain that every electronic system, and it's failure mode is a function of it's wiring harness, it's individual components( of which there are many many many parts, along with printed circuit boards (of which I have repaired and serviced) along with the myriad connectors( ever solder an80 pin amphenol connector) ; and all of that along with servo and feed back systems makes for an
inordinately complex and as such prone to failure system, which when such happens will KILL YOU and other innocent victims in the process.No other failure possibility, eh? Why do you assume that any type of failure is catastrophic? Or are you assuming that everything fails at once, as if hit by an EMP? I mean, even if it did cause the automated car to commit suicide/homicide, you're suggesting that it would also somehow prevent seatbelts and airbags from working too!


This scheme, in my opinion has no merit. It would work on a monorail for it works in one dimension only. It's position on rail and proximity to the next car and next station or mid-station is simple enough to detect and alaso to back up. No question. That is the way to go.Do you really think a monorail highway system stands a better chance of being adopted by the public than self-driving cars?

swampyankee
2012-Apr-17, 12:37 AM
Let me see...I've successfully controlled cars that have had high speed blowouts, danced around somebody who pulled in front of me from an intersection, blocking the opposite travel lane, and the person behind him pulled partway out of the intersection, stopping in front of my line of travel, leaving me a gap no more than a foot wider than my car to get through, driven in snowstorms, and got to watch people spinning off the road all around me.

I don't consider myself a great -- or even particularly good -- driver (for one thing, I can't parallel park). I think that a properly thought out computerized car system could manage any of those situations. Either I or people I've known have survived every scenario that's been brought up, and quite a few that haven't. These situations would be included in the software design, as engineers do not tend to be stupidly optimistic. Sensor failures? Expected. System failures? Redundancy; expected. If somebody can bring up some unlikely scenario, it will be thought of. Some problems, the automation can't recover from. I've had a friend lose a driveshaft at 50 miles per hour, another throw a rod at highway speeds, an acquaintance have an engine freeze solid at highway speed, a friend have their car's hood open while he was on the highway. By somebody's logic all of these would be instantly fatal.

danscope
2012-Apr-22, 02:40 AM
Yes... you were the pilot in command. Let's think about that for a bit. Hmmm.... pilot in command. Now, assume that the pilot just leaves the seat and goes back aft to shoot the breeze with the stewardess......
YES, no command . The system just stops working. YOU are just moving at 50 mph into traffic head on at 50 mph.
You dear reader are going to have a head-on collision with a rate of closure of 100 MPH. And it is going to happen so fast
that you will never know it happened. Tout finis.
" Farewell and adieu all you dear spanish ladies.
Farewell and adieu all you ladies of Spain ...
as we've recieved orders for to sail back to Boston;
and never no more shall we see you again . "

cjameshuff
2012-Apr-22, 08:37 PM
Hi Van, Currently on vacation on Cape Cod. You keep bringing up " Shown" as in "Proven" that such a scheme is infallible.

Nobody's claiming such a thing. It doesn't have to be infallible, it only has to be better than humans. That's not hard.



I keep saying and will maintain that every electronic system, and it's failure mode is a function of it's wiring harness, it's individual components( of which there are many many many parts, along with printed circuit boards (of which I have repaired and serviced) along with the myriad connectors( ever solder an80 pin amphenol connector) ; and all of that along with servo and feed back systems makes for an
inordinately complex and as such prone to failure system,

Amphenol connectors are generally crimped, though there are exceptions. And yes, I've soldered mechanically complex parts, including connectors, and through-hole and surface-mount PC boards. You haven't provided any factual information indicating that these systems are as horrendously failure prone as you claim. In the real world, most cars on the road rely on computers for proper operation, and while computer trouble can be extremely annoying when it does happen, it is very rare. In fact, the use of computers allows massive simplification of the vehicle wiring, greatly reducing the number of opportunities for one of the most common types of failure. The computer is indeed complex, but not failure prone, and can replace other systems that are failure prone. Including the direct cause of most vehicle crashes and thousands of fatalities each day, the human driver.



which when such happens will KILL YOU and other innocent victims in the process.

You have yet to actually show this. Repeatedly making up nightmare scenarios and asserting that any failure will lead to catastrophe and death does not support your argument. Every time you do it, it only makes it more clear you have no grounds for your objections.



As an aside, we are moving into a world where fewer and fewer people can afford a new car or even relatively new car, and you would have us double the cost of and maintenmance of these "wundercars" that can magically drive us about without a care in a world.

I'm surprised you bring this up, given that what you've suggested would simply make it impossible for most people to drive, regardless of how cheap cars are. But guess what, computers and sensors are getting more capable and less expensive as technology progresses. Your reactions, senses, and coordination are not.

And it's not magic, it's technology. An assortment of quite well-understood engineering problems, in fact.



Yes... you were the pilot in command. Let's think about that for a bit. Hmmm.... pilot in command. Now, assume that the pilot just leaves the seat and goes back aft to shoot the breeze with the stewardess......
YES, no command . The system just stops working. YOU are just moving at 50 mph into traffic head on at 50 mph.
You dear reader are going to have a head-on collision with a rate of closure of 100 MPH.

It stops working. My car sounds an alarm and I take over. Uh-oh, nobody died.
It stops working and I don't respond in time. The fallback system brings the vehicle to a controlled stop. Sorry, still nobody died.
A highly skilled driver with 90th percentile reflexes, perfect vision, and healthy sleep patterns sneezes while driving into the afternoon sun and swerves into oncoming traffic. My car takes evasive action before I could even register what's happening and turns a head on collision with fatalities in both vehicles into a glancing one resulting only in minor injuries and a severe scare. Still nobody dead, in spite of a human driver being involved!

You assume that any failure at all will immediately send the car at full speed into oncoming traffic, and that such a failure is more likely than human failure, which kills thousands of people a day. You have yet to give any reason why this is fundamentally more likely to happen with machine drivers than with human ones. Why won't fallback systems work? Why will computer drivers be mysteriously more failure prone than anything else in the vehicle?

danscope
2012-Apr-22, 09:16 PM
Hi James, You wrote..."You have yet to actually show this. Repeatedly making up nightmare scenarios and asserting that any failure will lead to catastrophe and death does not support your argument. Every time you do it, it only makes it more clear you have no grounds for your objections."
I volunteer You to test this scheme for twenty years. :) When it kills you, we shall place a small plaque near the site where your glorious trial of a roboticaly operated car failed while you were sleeping in the car. Wile E. Coyote always comes back as if by magic. And James? shrug...
You are betting much with this scheme,Sir.

Van Rijn
2012-Apr-22, 09:44 PM
Hi Van, Currently on vacation on Cape Cod. You keep bringing up " Shown" as in "Proven" that such a scheme is infallible.


No, I did not. Here's the text of my question again, bolding added:




So what happens if you're shown that automated driving would reduce deaths by a factor of a hundred? Would you accept it then, or would you still say no?


So you can see your statement above is wrong. All I'm asking is what happens if you're shown it would substantially reduce deaths? You've been arguing safety, but when confronted with this question, you seem to be doing everything you can to avoid answering it, which strongly suggests that safety isn't your real concern.

danscope
2012-Apr-22, 10:21 PM
Hi Van, In order to "Show" us all that this scheme is "safer" and more "Foolproof" than driving as we do now, you would have us take what I suggest is an extraordinary risk in experimentation . You can point to statistics all you like.
But you only have to look at and try to fix your "engine service light" to see the error of this scheme.
We agree to disagree. Like I said, sell it to the insurance companies and the government. Even the biggest long-shot Louie at Hialea wouldn't bet a dollar on this scheme.

Best regards,
Dan

LotusExcelle
2012-Apr-22, 10:30 PM
Your inability to understand your own check engine light does not effect its usefulness to the entire transportation industry.

Van Rijn
2012-Apr-22, 11:49 PM
. . . and the question is still avoided.

swampyankee
2012-Apr-22, 11:56 PM
Every transport system where there has been a significant application of automation has had fatality rates reduced because of the level of insulation against human error that automation provided.

There is no rational basis to believe that this would differ for over-the-road vehicles.

danscope
2012-Apr-23, 01:41 AM
When you have a minor failure with an engine, your car rolls to a stop. If you get rear-ended before you can exit the car, it's too bad, but you do have a chance.
If your automated steering sends you into head-on traffic, you are dead. kaput. Finis. Expired. You are a non-living person. An ex-driver, devoid of life and run down the curtain and joined the choir invisible... a late person. See?
You people have educations. And you drive each day. Suggesting this scheme is a bad joke. Or perhaps you live in a world where nothing ever breaks or fails? Is that your experience so far? Doubt.
Try selling this to the shark tank. LOL.

Dan

cjameshuff
2012-Apr-23, 01:58 AM
I volunteer You to test this scheme for twenty years. :) When it kills you, we shall place a small plaque near the site where your glorious trial of a roboticaly operated car failed while you were sleeping in the car.

There we go again, "When it kills you". Given how poor humans perform at driving, what makes you sure I'd end up killed by the computer? You haven't given any justification for your assertions, you simply seem to have some irrational fear of computers that makes you think they'll kill people whenever given the chance.



Wile E. Coyote always comes back as if by magic. And James? shrug...
You are betting much with this scheme,Sir.

Everybody is already betting their lives every time they go on the roads. And the vast majority of the risk is not from mechanical or computer system failures, it is from themselves and other drivers. There are aspects to driving that make humans fundamentally unsuited to the task: repetitive activities performed over long periods of time dull the sharpest reactions, information overload and distractions lead to less attention being paid to important things, adaptation effects interfere with perception of speed, time, and distance, and even in the best conditions a human can only look in one direction at a time and simply can not react fast enough. This is not opinion, this is fact demonstrated by the constant stream of people dying in auto accidents.

Probability of any computer failure is low. Probability of such a massive systems failure that the vehicle isn't even capable of a controlled stop are extremely low...it does not take much to perform such a simple fallback action. A failure that somehow doesn't trigger any of the failsafes and leads to your nightmare scenario practically requires criminally incompetent design or malicious sabotage...compared to a sleepy/drug impaired/momentarily distracted/simply too slow driver the likes of which kill about three thousand people a day.

adapa
2012-Apr-23, 02:16 AM
So what happens if you're shown that automated driving would reduce deaths by a factor of a hundred? Would you accept it then, or would you still say no?
I will answer this. I would gladly accept it. Now here is my question to you:

If I show real world evidence of the opposite (that automation overuse can compromise safety), would you be also willing to change your position?

Here is a paper written by a PhD from MIT about The Role of Software in Recent Aerospace Accidents (http://sunnyday.mit.edu/accidents/issc01.pdf). It has a lot of good information from which many people here could learn. In fact, here is one excerpt:

"Some of the technical inadequacies in high-tech
aircraft system design stem from lack of
confidence in the human and overconfidence in
the automation."

Also, here is a training video (http://www.youtube.com/watch?v=h3kREPMzMLk) that was part of the response to the high number of accidents and incidents that resulted from automation overuse.



On another note, many here seem unable to discern between instrumentation and automation thus incorrectly crediting automation with the improvements in safety resulting from improved instrumentation. I encourage everyone here to please learn the difference in order to prevent arguing without sufficient knowledge of the subject.

cjameshuff
2012-Apr-23, 02:19 AM
When you have a minor failure with an engine, your car rolls to a stop. If you get rear-ended before you can exit the car, it's too bad, but you do have a chance.
If your automated steering sends you into head-on traffic, you are dead. kaput. Finis. Expired. You are a non-living person. An ex-driver, devoid of life and run down the curtain and joined the choir invisible... a late person. See?

If it does that. Why would it do that?
Human drivers drive into oncoming traffic and get themselves and others killed in horrible accidents. What reason do you have to think computers will do the same?



You people have educations. And you drive each day. Suggesting this scheme is a bad joke. Or perhaps you live in a world where nothing ever breaks or fails? Is that your experience so far? Doubt.

I have an education and drive quite often if not every day. I live in the real world where things do break or fail, but where complex systems can and are engineered to handle failure, and where humans are the overwhelmingly most common failure point in auto accidents. This is not a joke...self driving cars could save millions of lives a year.

swampyankee
2012-Apr-23, 02:24 AM
When you have a minor failure with an engine, your car rolls to a stop. If you get rear-ended before you can exit the car, it's too bad, but you do have a chance.
If your automated steering sends you into head-on traffic, you are dead. kaput. Finis. Expired. You are a non-living person. An ex-driver, devoid of life and run down the curtain and joined the choir invisible... a late person. See?
You people have educations. And you drive each day. Suggesting this scheme is a bad joke. Or perhaps you live in a world where nothing ever breaks or fails? Is that your experience so far? Doubt.
Try selling this to the shark tank. LOL.

Dan

I'm getting past the point where I'm going to lose my temper. You keep bringing up fantasies of what may happen in the event of automated vehicles. You have not bothered to accept the simple fact that in every transportation system, the accident and fatality rates have gone down with the introduction of automation. Give us some reasons that automated over-the-road vehicles will be markedly different from other transportation media in this regard, especially considering that system failures in any of them are equally or more likely to cause fatalities than in cars.

cjameshuff
2012-Apr-23, 02:36 AM
Here is a paper written by a PhD from MIT about The Role of Software in Recent Aerospace Accidents (http://sunnyday.mit.edu/accidents/issc01.pdf). It has a lot of good information from which many people here could learn. In fact, here is one excerpt:

Aerospace systems are not very comparable. There's one huge difference...a car can always just stop. This is not an option for aircraft...an automated system must do what it can until a human takes over. Safeguards against human error can impede that in some situations, but the fact is that human error needs to be guarded against.

Spacecraft that experience a failure commonly do something that danscope seems convinced is impossible...go into a failsafe mode. They can't stop either, but there's often time to perform remote diagnosis and fix or work around the issue. System failures do not generally cause the spacecraft to explode or careen into a city.

Jens
2012-Apr-23, 02:58 AM
If your automated steering sends you into head-on traffic, you are dead. kaput. Finis. Expired. You are a non-living person. An ex-driver, devoid of life and run down the curtain and joined the choir invisible... a late person. See?


A person faints at the wheel, sending the car into head-on traffic. You are dead, kaput, finis, expired. A late person. See? Letting humans drive cars is a bad joke. So according to your logic, we must ban cars in toto.

danscope
2012-Apr-23, 03:00 AM
Dear james, you conveniently forget that untill impact, a space craft has the luxury of time to sort out it's problems or surrender control to the human on board..... or if no human is on board, entertain the loss of the mission.
"These things happen" is written on a very small plaque in the back of the JPL control room, face down and out of the way. But they do happen. A car in two-way traffic has milliseconds before severe impact, a fact that is lost on the advocates for this scheme.

Jens
2012-Apr-23, 03:52 AM
A car in two-way traffic has milliseconds before severe impact, a fact that is lost on the advocates for this scheme.

But then it must be lost on advocates of the scheme of human driving as well, because a human cannot react within milliseconds either, a fact that is quite well demonstrated by the fact that many people die in car crashes with human drivers.

cjameshuff
2012-Apr-23, 05:24 AM
Dear james, you conveniently forget that untill impact, a space craft has the luxury of time to sort out it's problems or surrender control to the human on board..... or if no human is on board, entertain the loss of the mission.

I did not forget that, I specifically pointed to that trait as something that made them closer to automotive systems. An automobile similarly has the option to just stop until humans can do something about the problem. That's right, it doesn't have to swerve into oncoming traffic!



"These things happen" is written on a very small plaque in the back of the JPL control room, face down and out of the way. But they do happen. A car in two-way traffic has milliseconds before severe impact, a fact that is lost on the advocates for this scheme.

Are you now claiming that computers have inferior reaction speed to humans?

Here's reality: it takes about a second under good conditions for a human to recognize and react to an unexpected event, varying widely due to unavoidable lapses in attention. Add in some fatigue or distraction and the reaction time goes way up. Computers on the other hand actually can react on millisecond timescales, and they can do so with precisely applied steering and braking while taking obstacles on all sides into account instead of a panic reaction that's likely to itself cause further trouble. So explain, please, how exactly are the fast reaction times required "lost on the advocates for this scheme"?

Van Rijn
2012-Apr-23, 05:37 AM
I will answer this. I would gladly accept it. Now here is my question to you:

If I show real world evidence of the opposite (that automation overuse can compromise safety), would you be also willing to change your position?


If you showed that a particular system was riskier than human driving, certainly I wouldn't accept it. I would only be interested in a system that could be shown to have lower risk than human drivers.

NEOWatcher
2012-Apr-23, 01:29 PM
If you showed that a particular system was riskier than human driving, certainly I wouldn't accept it. I would only be interested in a system that could be shown to have lower risk than human drivers.
That sounds like you saying that we assume the new system is safer until we can prove otherwise. Yikes.

I want to see it in real life before I will trust it.
Although I understand that it only needs to be better than human drivers, there are non-technical issues that still need to be addressed before I consider it safer than humans.

To expand on Adapa's comment:

"Some of the technical inadequacies in high-tech
aircraft system design stem from lack of
confidence in the human and overconfidence in
the automation."
And that's with highly trained operators.

Cost is an issue. Sure; mass production will bring the cost down, but there are huge political issues if it needs to be adopted by everyone.

Depending on the failsafe modes, we can really have a traffic mess on our hands if they are too sensitive or can't get the vehicle out of traffic.

It's not the problems we design for, but the problems that we don't see. I see things like the mysterious Toyota accelerator problem, and wonder how that would scale up to an automated system. Yes; there hasn't been a large number of injuries with that (compared to people who confuse pedals), but it's the potential that's there. You can take that bad driver out of the picture, but it's harder to address the millions of possibly defective systems.

Think of the Apollo 1 hearing portrayed in FTETTM. "It [the problem] was caused by a lack of imagination".

ETA:
Perhaps some kind of phased in approach would help. I've heard of some car free zones in urban areas. Maybe that idea can be extended to congested areas as automation only zones.

danscope
2012-Apr-23, 01:43 PM
Hi Neowatcher, You make the point of employing "some" of the technology, as I had stated that speed govenors (velocity limiting as a function of gps location detecting speed limit ) and a good proximity sensor would keep the tail gaters at bay. Yes, I can see that sort of thing saving us all a lot of headaches. Of course, the police are going to stop writing speeding tickets and have to find their revenue somewhere else.
Best regards,
Dan

NEOWatcher
2012-Apr-23, 02:24 PM
Of course, the police are going to stop writing speeding tickets and have to find their revenue somewhere else.
They'll find something... burned out light, obstructed plate, obstructed view...
Of course they'll have to start looking instead of having a radar beep wake them up.

adapa
2012-Apr-23, 03:27 PM
Aerospace systems are not very comparable. There's one huge difference...a car can always just stop. This is not an option for aircraft...an automated system must do what it can until a human takes over.
Actually, if you read the source that you just critiqued, then you would realize that its relevance extends far beyond aerospace applications. Please read sources before critiquing them. Also, fail safe modes apply to recognized internal failures. In many cases, two or more systems will function as advertised but their interaction can still create a hazard. This means that there is technically no failure to detect, but it can still present a danger. Here is an excerpt:

"We are starting to see an
increase in system accidents that result from
dysfunctional interactions among components,
not from individual component failure. Each of
the components may actually have operated
according to its specification (as is true for most
software involved in accidents), but the
combined behavior led to a hazardous system
state."



There is also the case where the designers cannot foresee every possible scenario as mentioned by the following excerpt:

"Instead the issue is
whether software can be constructed that will
exhibit correct appropriate behavior under every
foreseeable and unforseeable situation"

For example, you wrote that the car can always stop. However, braking is not always appropriate especially if the anomaly was caused by driving on an ice patch.


On another note, I notice that the proponents of total automation seem to focus on reaction as opposed to planning and anticipation. Any reasonably competent driver knows that a big part of road safety involves anticipating potential hazards and taking steps to avoid/mitigate them instead of just depending on reflexes. For example, a decent driver who sees a potential ice patch will slow down and/or avoid it long before he/she reaches it.



If you showed that a particular system was riskier than human driving, certainly I wouldn't accept it. I would only be interested in a system that could be shown to have lower risk than human drivers.
The question which you chose to answer is not the question which I asked. I did not ask about a particular system, but about automation overuse in general. Here is the question again:

If I show real world evidence of the opposite (that automation overuse can compromise safety), would you be also willing to change your position?

Also, the sources that I presented were not about any particular system. Both were about automation overuse in general and the paper has relevance which extends far beyond the aerospace environment.




in every transportation system, the accident and fatality rates have gone down with the introduction of automation.
I have avoided correcting you, but you keep making this wrong statement over and over again. The improvements in aviation safety resulted from improvements in Crew Resources Management combined with improvements in instrumentation -not automation. Please learn the difference between the two. In fact, the video which I posted in post#194 was a response to the accidents caused by automation overuse.

ShinAce
2012-Apr-23, 03:29 PM
I'd like to see the cost of building and maintaining roads versus actual revenue from moving violations. Actual revenue must take into account the cost of giving tickets, paying court costs, and collecting the revenue. I doubt that traffic tickets are keeping the road transportation system in the black. I feel that law enforcement is getting poor representation here.

Also, I want to point out that automation does not reduce the burden on the pilot. It actually increases it. Not only should I to study the rules of the road, and learn defensive driving, but now you want me to learn(to the level of instinct) several hundred pages of autopilot manuals. Yet I've been driving for over 10 years without a single accident or traffic ticket. Show me one automated car that has gone on public roads, in snowstorms, freezing rain, with obstacles occasionally flying across the road, and has accomplished over 100,000 miles without a single problem! The best part is, I have violated plenty of the rules of the road. Just last night, a police officer pulled someone over on the side of the road. I crossed the double yellow to pass safely. A computer would be programmed to not cross the double yellow. Meanwhile, I think the officer appreciated my courtesy to him.

Commercial planes have an immense investment in the plane itself, so adding the cost of an autopilot isn't that much. Even then, do commercial planes have collision avoidance? I don't mean an alarm that sounds when a collision is imminent. We've been handing the controls back to the pilot for this and saying "Heading for crash! Have fun!". I mean an autopilot that detects a possible collision and actively avoids it.

Full automation is just not a good investment. Let it take its course. Anti-lock brakes took time to trickle down. Are there cars without ABS nowadays? The direction of progress is always in favor of automation. Automation is usually an attempt to save costs. I don't get paid to drive, so it would save me nothing. Why should I pay more to have less?

I have a friend who is a neurosurgeon. When on call, he's not necessarily in the hospital. If he gets called in for emergency surgery on a critical patient, I want him to be able to speed. If his car is capped to the speed limit, people may die. If a woman goes in to labor, does it make any sense to drive exactly the speed limit if no one else is around? When people are allowed to take control in odd situations, they sometimes accomplish amazing things. Computer's do not think outside the box.

Swift
2012-Apr-23, 04:33 PM
Actually, if you read the source that you just critiqued, then you would realize that its relevance extends far beyond aerospace applications. Please read sources before critiquing them. ...

I have avoided correcting you, but you keep making this wrong statement over and over again. The improvements in aviation safety resulted from improvements in Crew Resources Management combined with improvements in instrumentation -not automation. Please learn the difference between the two. In fact, the video which I posted in post#194 was a response to the accidents caused by automation overuse.
I'll pick on adapa's post, but this is actually a general warning to many people.

There are a lot of comments in this thread that are pushing our limits for polite discussion. For example, accusing another members of not reading a source is considered rude. Using exaggerated phrases for death or injury are not helping the discussion, even if meant as jokes. And this isn't ATM or CT: you can ask questions of each other, and it is good to answer them if one is going to participate in a discussion, but you can't make demands to answer them.

I'd like everyone to take a deep breath and pause a little bit before hitting the Post button.

Thanks,

cjameshuff
2012-Apr-23, 05:03 PM
That sounds like you saying that we assume the new system is safer until we can prove otherwise. Yikes.

How can you possibly interpret "I would only be interested in a system that could be shown to have lower risk than human drivers." to mean that?



Depending on the failsafe modes, we can really have a traffic mess on our hands if they are too sensitive or can't get the vehicle out of traffic.

How much occasional inconvenience is a million lives a year worth?



For example, you wrote that the car can always stop. However, braking is not always appropriate especially if the anomaly was caused by driving on an ice patch.

Context. Bringing the car to a stop is a failsafe for a system failure, not for loss of traction. And computers are certainly much better suited for handling the latter situation, being able to actually monitor traction of each wheel, road temperature, etc.



On another note, I notice that the proponents of total automation seem to focus on reaction as opposed to planning and anticipation. Any reasonably competent driver knows that a big part of road safety involves anticipating potential hazards and taking steps to avoid/mitigate them instead of just depending on reflexes. For example, a decent driver who sees a potential ice patch will slow down and/or avoid it long before he/she reaches it.

I actually have referred to the enhanced capability for coordination among vehicles throughout this discussion. A self driving car could signal other self driving cars not only that it is braking, but how hard it's braking and why. It could coordinate with a city traffic management system to take less heavily loaded routes and minimize queues at traffic lights. It can take into account obstacles and hazards reported by other vehicles.



I have avoided correcting you, but you keep making this wrong statement over and over again. The improvements in aviation safety resulted from improvements in Crew Resources Management combined with improvements in instrumentation -not automation. Please learn the difference between the two. In fact, the video which I posted in post#194 was a response to the accidents caused by automation overuse.

It's not clear that it's a wrong statement. Improper application and faulty automation have led to a new category of failures, but safety has still generally increased.



Also, I want to point out that automation does not reduce the burden on the pilot. It actually increases it. Not only should I to study the rules of the road, and learn defensive driving, but now you want me to learn(to the level of instinct) several hundred pages of autopilot manuals.

This is simply not true. You assume an absurdly user-unfriendly design is a fundamental requirement of a self-driving car...utter nonsense.



Yet I've been driving for over 10 years without a single accident or traffic ticket. Show me one automated car that has gone on public roads, in snowstorms, freezing rain, with obstacles occasionally flying across the road, and has accomplished over 100,000 miles without a single problem! The best part is, I have violated plenty of the rules of the road. Just last night, a police officer pulled someone over on the side of the road. I crossed the double yellow to pass safely. A computer would be programmed to not cross the double yellow. Meanwhile, I think the officer appreciated my courtesy to him.

What we have here is a shining example of argument by anecdote.
Nothing happened to you...so what? Roughly 12 million people died in traffic accidents in that time. And I fail to see how your inability to adhere to the rules of the road supports your point in any way.



Full automation is just not a good investment. Let it take its course. Anti-lock brakes took time to trickle down. Are there cars without ABS nowadays? The direction of progress is always in favor of automation. Automation is usually an attempt to save costs. I don't get paid to drive, so it would save me nothing. Why should I pay more to have less?

Are you really saying you place no value on your life and health and that of others?

You are a bad driver. You are prone to lapses in attention and bad judgement under the best of conditions, and your reaction time is completely inadequate to the task of handling unexpected situations on the road. I know this because you are human...I don't need to know anything more about you. That nothing has happened to you yet means nothing...this is an expected result for some individuals on the road. Many others, including better drivers than you, have gotten injured or killed in that same timeframe.



I have a friend who is a neurosurgeon. When on call, he's not necessarily in the hospital. If he gets called in for emergency surgery on a critical patient, I want him to be able to speed. If his car is capped to the speed limit, people may die. If a woman goes in to labor, does it make any sense to drive exactly the speed limit if no one else is around? When people are allowed to take control in odd situations, they sometimes accomplish amazing things. Computer's do not think outside the box.

And if he gets himself and a family of five killed in a collision, or simply gets pulled over and delayed, how does the patient benefit?
Besides, prioritization of emergency traffic is something that has been brought up already as a possible benefit of automation, and even without that, a self driving car could take traffic hazards and loading on alternate routes into account and not just avoid traffic jams, but prevent them from happening in many situations. Rather more useful and less hazardous than letting your surgeon friend speed.

ShinAce
2012-Apr-23, 05:48 PM
I'm all for safety. But let's recall that some of us used to smoke indoors and yet we are entitled to say "safety first!". We must tread carefully when accusing anyone of being a hazard. Sometimes I struggle to understand if we are proposing automated driving for certain situations or driverless cars. To make all cars completely driverless is a tremendous task that seems out of reach. However, that would be necessary since I am liable for anything that may happen. If I'm liable, I do very well need to know how it operates. That's the increase in burden that automation yields. You need a system that's 100% automated and shown to be 99.99999...% safe before the driver can safely relax.

If we overextend this type of argument, then we can say that all hunters are bad people. Why would you risk all those hunting accidents when meat can be safely grown on farms? There's something humbling about hunting, and getting involved with nature.

If safety is the only concern, why don't we just make the licensing exam very difficult? That will weed out a lot of traffic since public transportation becomes the only option for many. We'll also have the best of the best drivers on the road at all times.

The truth is, we depend immensely on our cars, as evidenced by the majority of cars containing a single occupant most times. Right now we have a mix of automated tasks and driving. Traffic lights themselves are forms of transportation automation. They have road sensors, and use traffic algorithms to plan accordingly. There is the beauty of them, that the algorithms are about anticipation. That's also the beauty of a human driver, that they can anticipate, for free!

How many accidents are caused by alcohol and speeding? Why don't all cars have breathalyzers built-in to the ignition? The most obvious safety concerns seem overlooked when we say that automation is the key to safe roads.

swampyankee
2012-Apr-23, 05:49 PM
Numerous aircraft, not least the Shuttle, were completely uncontrollable by human beings without extensive computer mediation. For these, the failure of the flight control computers doesn't involve a few minutes or seconds for the pilot to sort things out; it involves irretrievable loss of control and death because people can't react within small fractions of a second. The same is true of the F-16, the B-2, the F-117, and every fourth and fifth generation fighter aircraft: no computer, airplane crashes.

Again, getting a fully automated car into service would not be trivial, but the legal and sociological barriers may be greater than the technical ones. The objections being brought up may have limited validity, but there is no reason to expect automated systems to do any worse in these highly predictable scenarios than do human drivers. I've still to see a coherent explanation of why a computer would do any worse in any of these scenarios than would a person. I suspect that's because there isn't one, but I'm willing to entertain the idea.

Van Rijn
2012-Apr-23, 07:38 PM
That sounds like you saying that we assume the new system is safer until we can prove otherwise. Yikes.


I'm puzzled how you would come to that conclusion. I see nothing in my statement that would suggest that. Anyway, I've specifically mentioned testing in other posts, and I'd think that would be an obvious requirement.

Van Rijn
2012-Apr-23, 08:31 PM
The question which you chose to answer is not the question which I asked. I did not ask about a particular system, but about automation overuse in general. Here is the question again:

If I show real world evidence of the opposite (that automation overuse can compromise safety), would you be also willing to change your position?


Change my position on what? Automation overuse for what? What exactly are you discussing (since it doesn't seem to be the topic of the thread) and what position do you think I have on it?

You can discuss the suitability of a particular system for an application, measure the limits of the system, and for driving automation, you can test whether it is less risky than human drivers, or under what conditions a particular system should be running, and when a human should be driving instead.

Generalization, however, would be very difficult. If you find a fault with one driving system, that doesn't establish that the fault must apply to all driving systems. And if you're making statements about different systems used for substantially different applications, you'd have to establish how they were applicable to a specific driving automation system.

danscope
2012-Apr-23, 09:25 PM
Hi Swamp, Say.... you live in New England. Ever hit a no joke pot hole??? Hmmmm..... That will set off your airbag,
maybe break your neck. You could steer around it, but motomarvel won't .

Dan

swampyankee
2012-Apr-23, 10:06 PM
Hi Swamp, Say.... you live in New England. Ever hit a no joke pot hole??? Hmmmm..... That will set off your airbag,
maybe break your neck. You could steer around it, but motomarvel won't .

Dan


I've hit that pothole, because I couldn't steer around it. My airbag didn't deploy, and my neck wasn't broken (nor was it when my daughter drove into the front of a city bus). And you've still not provided any evidence that this is not a situation that could be dealt with by automation: it's predictable scenario.

cjameshuff
2012-Apr-23, 10:25 PM
I've lived in Michigan, where Lake Michigan's effect on winter weather + thin population = poorly funded pothole-filled roads. I've hit several potholes that left me wondering if my vehicle had been damaged. None set off the airbag or broke my neck, and likely all would have been avoidable by a computer that doesn't have to split its attention between watching for holes in the road and watching the surrounding traffic, looking for road signs, etc. And while human drivers regularly lose control of their vehicles while swerving to avoid an obstacle, a computer is not subject to startle reactions and can reliably take controlled actions to avoid obstacles.

ShinAce
2012-Apr-23, 11:45 PM
Again, getting a fully automated car into service would not be trivial, but the legal and sociological barriers may be greater than the technical ones. The objections being brought up may have limited validity, but there is no reason to expect automated systems to do any worse in these highly predictable scenarios than do human drivers. I've still to see a coherent explanation of why a computer would do any worse in any of these scenarios than would a person. I suspect that's because there isn't one, but I'm willing to entertain the idea.

You should look at the DARPA 2007 urban challenge outcome. It's kind of funny because a collision happened when both vehicles were going under 15 mph. Plenty of close calls. Lots of erratic driving. All at low speeds. The collision happened between two of the six cars that finished the race! Only six of the 11 finalists finished the race. If that's better than human driving, then god save the queen! Last I checked, human drivers were doing over 200 mph on formula 1 courses, coming to within inches of other vehicles, all while looking so cool and collected.

Jens
2012-Apr-24, 12:06 AM
That sounds like you saying that we assume the new system is safer until we can prove otherwise. Yikes.


Like others, I'm really surprised by this statement. I don't think anybody is arguing that we should put driving automation systems into place without testing them first. I don't think anybody is saying that we can automatically assume that automation will do better. But I think many of us believe that it will do better, and are arguing that this idea should be taken seriously and tested, and put into place IF it is found to be better (which I believe , not assume, it will be shown to be).

ShinAce
2012-Apr-24, 02:18 AM
But 100% automation has a very poor track record. Those cars aren't cheap either. Meanwhile, most 12 year old virgin drivers would do better. That's why there are no driverless cars on the road.

How long are we supposed to wait? 5 years? 10 years? 25 years? Indefinitely? Does it come with a cold fusion engine? Or are we betting that google will put driverless taxis on the roads within 24 months?

swampyankee
2012-Apr-24, 02:29 AM
You should look at the DARPA 2007 urban challenge outcome. It's kind of funny because a collision happened when both vehicles were going under 15 mph. Plenty of close calls. Lots of erratic driving. All at low speeds. The collision happened between two of the six cars that finished the race! Only six of the 11 finalists finished the race. If that's better than human driving, then god save the queen! Last I checked, human drivers were doing over 200 mph on formula 1 courses, coming to within inches of other vehicles, all while looking so cool and collected.

I don't think anybody is arguing that computers are better drivers now; the argument is that they will be better drivers in the near future. Some people are arguing that this is impossible.

ShinAce
2012-Apr-24, 03:18 AM
What is the near future? I'm 29. Will my unborn kids have to get a driver's license or not? I think they will have to pass the same test I did. Is more than 20 years still considered the near future?

A driverless car for sale in 10 years? Maybe. Laws requiring cars to be driverless in less than 20 years. Place your bets. There are several models of electric cars for sale, yet I've never crossed one on the road.

Solfe
2012-Apr-24, 03:35 AM
Hi Swamp, Say.... you live in New England. Ever hit a no joke pot hole??? Hmmmm..... That will set off your airbag,
maybe break your neck. You could steer around it, but motomarvel won't .

Dan

That isn't very likely scenario. I hit an open manhole at speed, on a motorcycle. I was about 1 car length behind the car in front of me and had only enough time to level the bike after an aborted swerve. I consider this to be a worst case.

It was either the manhole or a woman then parked car or head on in to bumper to bumper traffic. Part of the problem was the cover was in the street next to the manhole and it was raining. I couldn't tell which object was the hole.

I didn't go down immediately, the bike went all squirrely for about fifty feet and I ended up dropping it.

Van Rijn
2012-Apr-24, 03:45 AM
That will set off your airbag,
maybe break your neck. You could steer around it, but motomarvel won't .

Dan

Why not?

Van Rijn
2012-Apr-24, 04:17 AM
If safety is the only concern, why don't we just make the licensing exam very difficult? That will weed out a lot of traffic since public transportation becomes the only option for many. We'll also have the best of the best drivers on the road at all times.


You answer your own question in the next paragraph:


The truth is, we depend immensely on our cars, as evidenced by the majority of cars containing a single occupant most times.


The dependence is so strong there is immense resistance to anything that can restrict that. My nephew was killed by a drunk driver, and my sister has since been doing what she can to reduce drunk driving accidents. She's managed to get a couple things done, but the reality is that they will have a minor effect at best, and there was substantial resistance to both of the things she worked on. One example: She found there was a forgotten law on the books that can delay giving a license back to people convicted of accidental killing while driving drunk. People would regularly get out of prison and get right back in a car. Now they might have to wait awhile, but they usually still get their license back eventually. Looking into it, you'd see a pattern: It takes a lot to cause a permanent loss of a driving license here, and driving without a license usually doesn't cause that much of a penalty anyway. Realistically, conventional approaches aren't going to do much more to reduce accidents, injury and death on the road.

Van Rijn
2012-Apr-24, 05:16 AM
You should look at the DARPA 2007 urban challenge outcome. It's kind of funny because a collision happened when both vehicles were going under 15 mph. Plenty of close calls. Lots of erratic driving. All at low speeds.


That was 2007. You think they haven't improved since then?

DARPA did the first Grand Challenge in 2004 where they had a prize for a winning vehicle. For that, the vehicles were to go around a course in the country, but they didn't need to interact. No vehicle completed the course or won the prize.

They repeated the challenge in 2005. All but one of 23 vehicles exceeded the best distance record in the 2004 challenge, and five completed the test.

Having established vehicles could meet that challenge, they changed it: Now vehicles had to work in an urban setting and interact with each other. Six teams completed the course, so DARPA didn't repeat it. However, there's no doubt that there has been continued work on driverless vehicles for DARPA, and elsewhere other examples, like Google's design, have moved well beyond what they were using in the 2007 challenge.

NEOWatcher
2012-Apr-24, 04:59 PM
How can you possibly interpret "I would only be interested in a system that could be shown to have lower risk than human drivers." to mean that?
Because that's not what you said, nor what I quoted....
"If you showed that a particular system was riskier than human driving"
You didn't say that it was shown to be safer than a human in the first place which is why I questioned it. I'm hoping that was in your mind when you said that.



Like others, I'm really surprised by this statement.
What's wrong with "sounds like"? It sounded like that to me, I questioned it, and now people are jumping down my throat about it.



How much occasional inconvenience is a million lives a year worth?
Where did I say it was an issue. I was mearly making an added comment about what things could be like.

How about we start with those countries that make a significant contribution to that million lives? Once we see the system help there, then we can start moving to the less risky countries.

The U.S. is a few magnitudes less than many countries (http://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate). (per vehicle... I'd like to see per mile, but theres' not enough info there.

danscope
2012-Apr-24, 06:36 PM
" Why not automate the cars and cut down the surplus population? " Ebenezer Scrooge or someone like him. :)

cjameshuff
2012-Apr-24, 07:07 PM
Because that's not what you said, nor what I quoted....
"If you showed that a particular system was riskier than human driving"
You didn't say that it was shown to be safer than a human in the first place which is why I questioned it. I'm hoping that was in your mind when you said that.

It's not what I said, no. Van Rijn said it. It is what you quoted, I copied it straight from your quote!



What's wrong with "sounds like"? It sounded like that to me, I questioned it, and now people are jumping down my throat about it.

It doesn't "sound like" it at all. There's nothing subjective about it, you're attributing a position to Van Rijn that is explicitly contradicted by the very words you quoted. Of course people are going to react negatively!

NEOWatcher
2012-Apr-24, 07:35 PM
It's not what I said, no. Van Rijn said it. It is what you quoted, I copied it straight from your quote!
Yes; sorry, your response to what I said made me think I quoted you.
But; Why did you reword what I quoted? I quoted a specific sentence and you came back and attributed my comment on different wording.


It doesn't "sound like" it at all. There's nothing subjective about it, you're attributing a position to Van Rijn that is explicitly contradicted by the very words you quoted. Of course people are going to react negatively!
Then we are reading it differently. I re-quoted it. Let me requote it again.
"If you showed that a particular system was riskier than human driving".
Now; I know what you are saying, and I believe I agree. If this system is shown as being safer than human driving, then there's no issue. But; that's not how I read it. I read that statement as the proof having to be on the part of the person saying that the system was riskier. The proof should first be on the person that is saying the system is safer because it's something new and un-proven, we have no data. We do have data on human driving.

cjameshuff
2012-Apr-24, 07:35 PM
The dependence is so strong there is immense resistance to anything that can restrict that. My nephew was killed by a drunk driver, and my sister has since been doing what she can to reduce drunk driving accidents. She's managed to get a couple things done, but the reality is that they will have a minor effect at best, and there was substantial resistance to both of the things she worked on. One example: She found there was a forgotten law on the books that can delay giving a license back to people convicted of accidental killing while driving drunk. People would regularly get out of prison and get right back in a car. Now they might have to wait awhile, but they usually still get their license back eventually. Looking into it, you'd see a pattern: It takes a lot to cause a permanent loss of a driving license here, and driving without a license usually doesn't cause that much of a penalty anyway. Realistically, conventional approaches aren't going to do much more to reduce accidents, injury and death on the road.

And there are many people dependent on private vehicles, without any access to public transportation. If we took the approach of making licenses drastically more difficult to obtain as a few have suggested, we might get the worst drivers off the road and reduce congestion a bit, but many people will be denied access to any sort of transportation apart from what they can get from friends and family that can drive. This practically means no work, unless they can find a job on a route shared by someone they can carpool with. Or they get a waiver or just ignore the law and get on the road anyway...

And even otherwise healthy and capable drivers with experience and clean records get injured or ill occasionally. Would you walk a mile down the road to ask your neighbor if they'll drive you to the drugstore to get decongestants for your cold? Call an ambulance?

cjameshuff
2012-Apr-24, 07:39 PM
Yes; sorry, your response to what I said made me think I quoted you.
But; Why did you reword what I quoted? I quoted a specific sentence and you came back and attributed my comment on different wording.

I did not reword it. It's still right there in your post (http://www.bautforum.com/showthread.php/130029-Imagine-Computer-driven-Vehicles?p=2010704#post2010704).



Then we are reading it differently. I re-quoted it. Let me requote it again.
"If you showed that a particular system was riskier than human driving".

", certainly I wouldn't accept it. I would only be interested in a system that could be shown to have lower risk than human drivers."

Copied once again from your quote. There is only one way to read this.

NEOWatcher
2012-Apr-24, 07:44 PM
", certainly I wouldn't accept it. I would only be interested in a system that could be shown to have lower risk than human drivers.".
Arrg; slap me in the face. Yes; I guess the first sentence caught me off guard. I think I somehow interpreted the order of the sentence as to order of proof.
Sorry. I think we do agree.

swampyankee
2012-Apr-26, 06:42 PM
If safety is the only concern, why don't we just make the licensing exam very difficult? That will weed out a lot of traffic since public transportation becomes the only option for many. We'll also have the best of the best drivers on the road at all times.

The truth is, we depend immensely on our cars, as evidenced by the majority of cars containing a single occupant most times...

How many accidents are caused by alcohol and speeding? Why don't all cars have breathalyzers built-in to the ignition? The most obvious safety concerns seem overlooked when we say that automation is the key to safe roads.

The estimates I've read are that 50% of the fatalities involve impaired -- usually with ethanol or other, less legal, drugs -- drivers.

In most of the US, except urban areas (and even in some of them), public transport is so pathetically bad that making it significantly more difficult to get a license would result in large numbers of people driving without one. And even the best of drivers aren't at their best all the time, and recent studies have shown that driving while sleepy is just about as serious an impairment as a few shots of tequila.

LotusExcelle
2012-Apr-26, 07:06 PM
Driving while watching Top Gear is about 5 shots of tequila, in my informal survey.

swampyankee
2012-Apr-27, 12:57 AM
Driving while watching Top Gear is about 5 shots of tequila, in my informal survey.

Five shots of tequila while watching does make the host seem much more sensible.

Solfe
2012-Apr-27, 02:08 AM
Golden Earring's Radar Love (http://www.youtube.com/watch?v=JeRa3RtBiIU) is pretty much an automatic speeding ticket.

Of course, with a computer driven car that wouldn't happen any more.

danscope
2012-Apr-27, 02:20 AM
And just how will the local constabulary fill their coffers without speeding revenue? :)

swampyankee
2012-Apr-27, 03:01 AM
And just how will the local constabulary fill their coffers without speeding revenue? :)

They'll find something. Actually, in Connecticut, the towns get fairly little (iirc, it's 25%) of the money from a ticket. This is one of the reasons a lot of towns are lax (frequently far too lax) in traffic enforcement.

Solfe
2012-Apr-27, 03:56 AM
And just how will the local constabulary fill their coffers without speeding revenue? :)

With computer driven cars, that might not be a joke. I think the greatest hurdles for automated cars are all social.

The sheer (or is it shear?) number of things that computer controlled cars can impact is incredible. Computer controlled cars could reduce or remove the need for traffic stops, signals and other things that we take for granted. Yearly safety inspections might be redundant since the car would have to be able to do this themselves every day. More revenue gone, not to mention the repair shops that are going to have to do preventive maintenance instead of a series of more costly stop-gap measures. If cars could negotiate roadways without traffic jams, you may see faster highways speeds in places where the cars sensors can reach farther and much slower speeds than currently in use when they can't. That would mean fewer people being taken to hospitals either because the speed was way too high to survive or way to slow to cause injuries. Insurance as we know it may not be necessary.

Personally, I don't like the idea of a computer driven car. I don't even like automatic transmissions; that isn't driving, that's a ride a Disney World. I think the best way to go would be to use the technology in rapid transit. Currently in my city, the public transit is really bad. If you could make the public transit vehicles smaller, faster and more efficient, that would be a godsend. This sort of technology would do that. If I could just hop a bus in my neighborhood and it got me to where I needed to go, I wouldn't drive as much as I do.

Currently, I have removed two of my cars from the road for a variety of reasons, but primarily because my school tuition includes free public transportation*. I have to walk a mile and a half to get to a bus stop, but I get a pleasant ride to school after a leisurely walk. I save a ton on insurance, gas, and wear by not having these cars on the road. Plus I get a work out, have 30 minutes a day in the library and I go grocery shopping on my way home. This sort of transportation is more appealing to me than a car.

*I should mention that my "free" public transportation is actually covered by an additional non-negotiable $200.00 transportation fee and $60.00 parking pass (yes, I have to buy a parking pass AND bus pass. Don't ask, I can't figure it out either and I deal with these people on a daily basis) at my college. If I actually had (a choice) to pay for the service daily, it would actually be four times more expensive than driving a car. To break even, I would have to drive my 13 MPG conversion van and gas would have to be between 10 and 11 dollars a gallon.

Jens
2012-Apr-27, 04:08 AM
And just how will the local constabulary fill their coffers without speeding revenue? :)

Their revenues will likely drop, but so will their costs, because you won't need so many traffic cops. I don't know how it will balance out.

danscope
2012-Apr-27, 06:23 PM
That changes a little near the Conn/RI border. :)

danscope
2012-Apr-27, 06:37 PM
Yes, a lot of unexpected changes. And costs. Remember the Krell? You will have to build cars to that kind of standard. You know, even nuclear submarines have stuff that breaks. That's why we train crews.
I do think we could eliminate tail-gating and by default..... inter-swerving, you know, the mad dog slalom drivers who fancy themselves driving formula One, ducking into impossibly small holes to pass everything on the road because something inside makes them do it???? When they try to slalom, it would kill the engine or some such action to punish them. We certainly could limit speeders .
Oh, I can just hear the cries and protest, but probably necessary if we are to live with increasingly smaller and less powerfull cars. Gee, people would have to drive like
Beaver Cleaver's Mom. Maybe a good thing. And life expectancy would go up.
Maybe......... Nah!!!!!! Steve Martin as Orlich of York, Medieval Barber .
But really....

swampyankee
2012-Apr-28, 02:00 AM
I think there is a lot of automation that could be profitably added to cars. My first choice wouldn't be automation, it would be a "black box" data recorder that would record a few things, like speed, rpm, throttle position, speed, etc for use in accident reconstruction.

cjameshuff
2012-Apr-28, 02:40 AM
I think there is a lot of automation that could be profitably added to cars. My first choice wouldn't be automation, it would be a "black box" data recorder that would record a few things, like speed, rpm, throttle position, speed, etc for use in accident reconstruction.

Certainly, a progressive approach is most likely...collision avoidance sensors to keep you from, say, backing into something or someone at low speed (available in some forms, though rare), overrides to prevent loss of control due to panicked overreactions in braking or steering (already being implemented), tracking the local speed limit and displaying it and overspeed warnings, alerting the driver to things like intersections and turn-only lanes, warnings when drifting out of the correct lane...

danscope
2012-Apr-28, 03:46 AM
Never mind warnings. Just employ genuine governors. Simple. You can't exceed 70 MPH. Period.
And Stop the tail-gaiting. Easy enough to do. We will save lives, cars, gas and money just doing that.
You got technology? Use it.
Anyone opposed to that?

swampyankee
2012-Apr-28, 04:08 AM
Never mind warnings. Just employ genuine governors. Simple. You can't exceed 70 MPH. Period.
And Stop the tail-gaiting. Easy enough to do. We will save lives, cars, gas and money just doing that.
You got technology? Use it.
Anyone opposed to that?

There was a plan to require all cars sold in the US to have a governed top speed of (I think) 85 mph (iirc, it was during the Nixon administration). For all the howling, you'd be thinking that the law was as bad as requiring you to listen to William Shatner's singing.

danscope
2012-Apr-28, 05:18 AM
Dreadfull ! Isn't his record used to extract information....even the threat of it? :)

swampyankee
2012-Apr-28, 12:01 PM
Dreadfull ! Isn't his record used to extract information....even the threat of it? :)

Even Bybee thought that was going too far.

danscope
2012-Apr-28, 05:09 PM
And just for the record, we all love Bill Shatner, even if we sometimes have some fun at his expense.
Love the original series .... and the girls :)

neilzero
2012-Apr-28, 11:28 PM
"Stop the tailgating" No, but we can reduce it near term, to about one second per hour of driving time, for cars equiped with reasonable technology. When someone pulls in front of you, A capcitor charged to several hundred volts is connected to the break lights to give the vehicle behind earliest possible notice (probably not necesary for LED break lights) Eject some artifical smoke near the rear tires to indicate to the following driver that you are at extreme breaking. A high powered speaker emits the sound of screeching tires and breaks in case the driver behind has his eyes closed. Cut the fuel back to idle amount. Apply maximum breaking for about 10 milliseconds, while the computer analizes whether you will be rear ended if fast braking is likely to result in you being rear ended. With average conditions minimum safe following distance is achived in one or two seconds, but the guy behind you needs a nerve pill even if he was at mimum safe following distance. Neil