Self-Driving Cars

Capt Stormfield writes: I didn't say all accidents were avoidable, but many are.

I agree with you that many accidents are avoidable, but the accidents I indicated in that video show that many are not avoidable by even the most elite of drivers.

As you state in the part of the post I responded to, there will also be accidents with driverless cars.

But many, many less. I suggested the example of reducing annual motor vehicle deaths from 33,000 to 20,000, and that's very minimal. If self-driving cars were the norm it would be much, much less. Yes, there will still be accidents with self-driving cars, but what do you think that annual fatalities would be if all cars were self-driving? Don't you think it would be a very small number, say in the low thousands? Pedestrian fatalities are included in the motor vehicle fatality figures, but we shouldn't forget that pedestrian fatalities will also experience dramatic reductions. As you saw in that TED video, the Google car does an excellent job tracking pedestrians.

I think the desire to retain as much control as possible over one's fate is one of the better parts of human nature, not an irrationality.

I agree that a desire for control is part of human nature, but it isn't often claimed that human nature is rational. I also agree that many will still want to drive their car even after self-driving cars are available, but they won't be safer than self-driving cars, and if safety is the criteria then wanting to drive anyway won't be rational, either.

--Percy

After a number of years it will automatically be manslaughter and a jail sentence if you are driving and anyone is hurt for whatever reason.

Perhaps. But in the meantime, who gets sued when your autobot hits a 1st year medical student, and the damages are calculated to be an entire doctor's career worth of earnings? You? GM?

Way down the road there won't be a steering wheel and the problem will be solved.

Yeah.

Actually there is a lot of stuff written about this subject already. The discussion here just pokes around at a lot of low hanging fruit.

NoNukes writes: Perhaps. But in the meantime, who gets sued when your autobot hits a 1st year medical student, and the damages are calculated to be an entire doctor's career worth of earnings? You? GM?

Interesting question. Insurance will play a role, and the details of car insurance will likely evolve with the arrival of self-driving cars. What could an occupant of a self-driving car do to be held responsible for an accident? GM seems a more likely possibility (assuming it's not a Ford ).

Actually there is a lot of stuff written about this subject already. The discussion here just pokes around at a lot of low hanging fruit.

Do you have some useful links?

--Percy

But robocars will behave a lot like public transport when on the move, they'll lock onto the car in front and form 'roadtrains' on the freeways.

And at that point you're just talking about a very inefficient form of 'public' transportation.

. What could an occupant of a self-driving car do to be held responsible for an accident?

Exactly how things work out might not be defined by what technology is actually capable of. Having no manual override whatsoever might well be something that may never happen. There are some recent articles about the logic in googles cars getting confused by fixed gear bikes in some situations.

http://www.washingtonpost.com/...e-googles-self-driving-cars

Obviously this bug will get fixed, but at what point will it be assumed that self driving cars are capable of handling everything.

Some links where some interesting questions are raised.

http://www.wired.com/2012/01/ff_autonomouscars/
http://www.popularmechanics.com/...elf-driving-cars-16016418
http://www.iii.org/...update/self-driving-cars-and-insurance
http://slashdot.org/...-to-take-control-of-self-driving-cars
http://www.latimes.com/...ving-accidents-20150512-story.html
http://www.technologyreview.com/...rs-make-ethical-decisions
http://gizmodo.com/...ng-to-kill-jobs-and-not-jus-1705921308
http://www.sciencedaily.com/...ases/2015/06/150615124719.htm
http://www.cnet.com/...-advocates-tangle-with-messy-morality

Jon writes: And at that point you're just talking about a very inefficient form of 'public' transportation.

I wonder. Trains are pretty efficient but apart from rush hours, they don't run full and you have to get to and from their terminus's, usually by car. Cars running a couple of feet apart will have much better aerodynamics and they won't be on the road at all unless someone needs to be there, so every 'carriage' will be occupied all the time. Though I guess it'll be like today with most cars running with only a single passenger. Not as efficient as just the train element of the journey but more efficient than now.

Some quite interesting things occur though. The car could drive you to the theatre in the centre of town, drop you off, self park out of town, then pick you up after. If you think about it, you don't actually need a car anymore. You could just pay a monthly subscription to have acces to one, then call one up when you want one. Send it away when you're finished. You get your garage back to keep your hooch in and you don't have to clean, service and insure the damn thing.

Thanks for the links, here's one more that one of your links pointed to:

Using Future Internet Technologies to Strengthen Criminal Justice

Here are some interesting issues they raised:

• Weather. Conditions that a human might manage to navigate could be too much for a self-driving car. I imagine that a self-driving car would be forced to pull over to the side of the road and stop in some conditions, such as glare ice, or blinding rain or snow. Whether that's a good thing or bad thing, it's certainly a safe thing.

  Or would the vision systems of self-driving cars, be they lasers or radar or infrared or ultrasonic or a combination or whatever, be able to see through the weather far better than humans?

• Inter-car communication, also known as V2V (vehicle-to-vehicle). Will cars be able to communicate with one another? If so, it opens up a wealth of possibilities. In the ultimate degree the need for stop signs, yield signs, traffic lights and even overpasses goes away as self-driving cars communicate with one another to negotiate their way through lane changes, turns and intersections. Highway capacity could possibly increase enormously.

• Laws governing self-driving cars will likely vary by state. It is entirely possible that the self-driving car you use in your daily commute would be illegal during portions of a trip to visit grandma several states away.

• Liability. If a self-driving car provides the driver the option of asserting control, who is responsible in the event of an accident? Presumably the data history would reveal whether the car or the driver was in control at the time of an accident, so maybe it's not an issue.

  It does seem to me that whatever the initial legal framework when self-driving cars first start appearing on our roads in significant numbers, their dramatically lower accident rate should drive rapid changes. And as a couple articles note, the share of liability of manufacturers and maintenance/repair businesses will likely increase.

• Cost. Who knew? Cost will be a big factor in how fast self-driving cars fill our highways. I didn't see any cost estimates, but several comments made in passing lead me to believe that self-driving cars might be very expensive. Perhaps people could afford it because of reduced insurance costs. From Self-Driving Cars and Insurance: "Data from the Institute for Highway Safety (IIHS) and Highway Loss Data Institute (HLDI) already show a reduction in property damage liability and collision claims for cars equipped with forward-collision warning systems, especially those with automatic braking."

• The human urge to be in control. Self-Driving Cars and Insurance also echos some of the comments in this thread: "In addition, some people who enjoy driving and do not want control to be taken from them may resist the move to complete automation."

• Ethical issues. An example from How to Help Self-Driving Cars Make Ethical Decisions: "A child suddenly dashing into the road, forcing the self-driving car to choose between hitting the child or swerving into an oncoming van."

  Those who have seen the movie I, Robot might recall that just such an ethical dilemma was at the heart of main character Del Spooner's objection to robots, that after an accident one chose to save him instead of a young child. (It occurs to me now that a problem in that movie is that even given the incredible abilities of the robots, humans still drove their own cars.)

  But is that a child in the street? Or a dog? Or even just a beach ball? Imagine the liability if the car's autonomous systems directed the car into the path of an oncoming van to avoid a beach ball. Just how good will the car's recognition systems be?

  The article concludes by posing this thoughtful quandry: 'Walker-Smith adds that, given the number of fatal traffic accidents that involve human error today, it could be considered unethical to introduce self-driving technology too slowly. “The biggest ethical question is how quickly we move. We have a technology that potentially could save a lot of people, but is going to be imperfect and is going to kill.”'

--Percy

Edited by Percy, : Typo.

I am a huge fan of self-driving cars. I think they're cool.

Percy writes: Weather. Conditions that a human might manage to navigate could be too much for a self-driving car.

I doubt it.
Now, there may be conditions that go beyond that of the capabilities of a self-driving car and cause it to stop/pull over... but those would be well beyond a human anyway.

Or would the vision systems of self-driving cars, be they lasers or radar or infrared or ultrasonic or a combination or whatever, be able to see through the weather far better than humans?

It will be a combination.
Lasers, and radar, and temperature, and ultraviolet, and infrared and vision (cameras)... all working together, all confirming and co-ordinating.

And yes, it will be far better "sensing" than any human's "sensing."

Or, at least, it would be if I programmed it

If a self-driving car provides the driver the option of asserting control, who is responsible in the event of an accident? Presumably the data history would reveal whether the car or the driver was in control at the time of an accident, so maybe it's not an issue.

If such an option exists, the data history would definitely identify it.

"In addition, some people who enjoy driving and do not want control to be taken from them may resist the move to complete automation."

I don't see a problem with having self-driving cars and human-driving cars on the same road.

Ethical issues. An example from How to Help Self-Driving Cars Make Ethical Decisions: "A child suddenly dashing into the road, forcing the self-driving car to choose between hitting the child or swerving into an oncoming van."

Extremely interesting.
Don't such ethical issues exist right now?
Is there even a right answer?

Unknown occupants in the oncoming van vs. known child.
Van's doing what it's supposed to be doing vs. child not understanding what they're "supposed" to be doing.

Imagine the liability if the car's autonomous systems directed the car into the path of an oncoming van to avoid a beach ball. Just how good will the car's recognition systems be?

Have to be better than that, or I'm not driving one.

“The biggest ethical question is how quickly we move. We have a technology that potentially could save a lot of people, but is going to be imperfect and is going to kill.”'

A good question for debate.
I know what side I'm on - self-driving cars. For the same reason I think it's ethical to wear a seatbelt - there are other people driving on the road other than yourself.
Given that they can hit a beach ball instead of a van, that is

Stile writes:

Percy writes: Weather. Conditions that a human might manage to navigate could be too much for a self-driving car. I doubt it. Now, there may be conditions that go beyond that of the capabilities of a self-driving car and cause it to stop/pull over... but those would be well beyond a human anyway.

We have a joke here in New England about the guy that hands his coffee to his wife saying"Now watch this!"

Or would the vision systems of self-driving cars, be they lasers or radar or infrared or ultrasonic or a combination or whatever, be able to see through the weather far better than humans? It will be a combination. Lasers, and radar, and temperature, and ultraviolet, and infrared and vision (cameras)... all working together, all confirming and co-ordinating. And yes, it will be far better "sensing" than any human's "sensing." Or, at least, it would be if I programmed it

I am still amazed that people get into airplanes know it's a snowstorm at their destination or horrible icy conditions at takeoff.

Ethical issues. An example from How to Help Self-Driving Cars Make Ethical Decisions: "A child suddenly dashing into the road, forcing the self-driving car to choose between hitting the child or swerving into an oncoming van." Extremely interesting. Don't such ethical issues exist right now? Is there even a right answer? Unknown occupants in the oncoming van vs. known child. Van's doing what it's supposed to be doing vs. child not understanding what they're "supposed" to be doing. Imagine the liability if the car's autonomous systems directed the car into the path of an oncoming van to avoid a beach ball. Just how good will the car's recognition systems be? Have to be better than that, or I'm not driving one.

I am reminded of a question we all heard about when applying for Conscious Objection Status for the Vietnam War. The draft boards, apparently in trying to get you to fail their questions, would ask you "Suppose you're driving your children on a mountain road with nothing on the right side and you come around a curve and there in the road is your grandmother lying in the road trying to get up. Do you hit her or drive the kids over the cliff?" Joan Baez had an answer for us: "Wow - I totally freak out and lose control, hit my grandmother dead and veer over the fatal cliff, landing on the town below and set it on fire, killing hundreds!"

It all boils down to how the humans respond to the legal language differentiating the technological responsibility versus human responsibility.

If the car malfunctions and causes harm, it is the owners liability---thus if the car hurts the owner, the owner gets the blame.

If the car malfunctions and causes harm, it is the owners liability---thus if the car hurts the owner, the owner gets the blame.

Is that the principle we used when those Toyota, Nissan and Honda airbags were injuring and killing people? Of course not!

The new Google self-driving cars not only have no accelerator, brake pedal or steering wheel, they also have no windshield wipers.

I haven't seen this mentioned explicitly anywhere, but I'm guessing that the Google cars also have no turn signal controls, no windshield wiper controls, no cruise control.

Speaking of cruise control, how closely do self-driving cars obey the speed limit? If they strictly follow the speed limit then they will be very annoying to almost all other cars on the road. Can you set how much over the limit they should go? Can you tell them to increase or decrease their average speed? Can you tell a self-driving car to go 100 mph? I did find this at Google's driverless cars designed to exceed speed limit:

quote:

---

Google's self-driving cars are programmed to exceed speed limits by up to 10mph (16km/h), according to the project's lead software engineer.

Dmitri Dolgov told Reuters that when surrounding vehicles were breaking the speed limit, going more slowly could actually present a danger, and the Google car would accelerate to keep up.

---

This means that when self-driving cars are surrounded only by other self-driving cars, they'll all be doing the speed limit. On our local highway where the speed limit is 50 mph, no car is doing less than 60, many are doing 70, and the left hand lane is generally going 75-80.

This means that self-driving cars have the potential to greatly increase travel times. Imagine driving cross country through empty wide-open spaces at 65 mph instead of 80 mph. The travel time for a trip with 400 highway miles would increase from 5 hours to 6 hours 9 minutes.

I wonder how easy it will be to suddenly change your destination in a self-driving car. Say I'm driving home with no intention of stopping, but a couple hundred feet before I reach the turn-in for the grocery store I remember I need pimentos. If I'm driving the car myself this isn't a problem since I have plenty of time to signal and make a safe turn into the grocery store parking lot, but if I'm in a self-driving car I assume I'll be way past the grocery store by the time I even call up the screen to enter my new destination. What is needed is a quick way to instruct the car to "turn right here."

Here's another hypothetical scenario. You don't know the street address of where you're going, but you know how to get there (this makes perfect sense to those of us who are pre-GPS (I still don't have a GPS), but will become less and less common as time goes by). Will there be an easy way to give a self-driving car a series of instructions that boil down to, "Turn left at the next light; turn right in three blocks; turn left at the Exxon station; stop here." When you're in a city you need a "troll for a parking spot" mode.

Speaking of parking, which parking space does a self-driving car choose in a parking lot? For how long does it seek the closest space? If it chooses a space you don't like (say it's raining), how do you tell it to find a better one? When you're in a huge shopping mall parking lot, how do you tell a self-driving car to park by Macy's and not by Sears? When parking for a local sporting event overflows onto a grassy field, how does the self-driving car know where to park? Will it be able to detect the 16-year old kid beckoning to you from 300 feet away that you should drive down to him and park at the end of the row? Or will people attempting to give parking instructions to self-driving cars just give up and let them do whatever they want?

But why even worry about parking? What you'll usually want is for the self-driving car to drop you off at the front door of your destination, then go find a parking spot on its own. But if you're in a city maybe it won't be able to find a parking spot, so it will just keep looking. But what if it runs low on fuel? Will a self-driving car be able to refuel on its own? Maybe we also need auto-refueling stations.

AbE: But what if after your self-driving car drops you off it *does* find a parking space. How will it put coins in the meter? Obviously parking meters will have to have WiFi or Bluetooth.

--Percy

Edited by Percy, : AbE.

but if I'm in a self-driving car I assume I'll be way past the grocery store by the time I even call up the screen to enter my new destination. What is needed is a quick way to instruct the car to "turn right here."

Will there be an easy way to give a self-driving car a series of instructions that boil down to, "Turn left at the next light; turn right in three blocks; turn left at the Exxon station; stop here."

Voice commands, not clunky touch-screens.

If I'm driving the car myself this isn't a problem since I have plenty of time to signal and make a safe turn into the grocery store parking lot, but if I'm in a self-driving car I assume I'll be way past the grocery store by the time I even call up the screen to enter my new destination. What is needed is a quick way to instruct the car to "turn right here."

Not necessary. The "experience" modules will have already pre-planned the safest spur-of-the-moment routes to your grocery store on the right, your liquor store on the left and that Burger King you went to once that you just passed.

"Hey, Lexus2! Take me to that Burger King we just passed." Have it your way.

In the "Residential Experience" modules the car will not just watch the road and cars around it but will be looking for the ball bouncing across the yard with a kid chasing it, will be anticipating the "maybe's" of that dog chasing the frisbee two doors up and will prepare for the possibility that the car just parked up ahead will have its driver(less)-side door suddenly pop open.

All these situations we humans know can happen but never seem to think about until they're right up in our faces will have been anticipated by your Lexus2 with full action plans ready to execute. Even in those unique events that could not be anticipated (kid pops out from behind a parked car and a van coming head-on the other way) the swerve into the van's lane would be such as to effect a glancing blow rather than a head-on strike, something that may not be achieved by a human in the midst of the panic.

There will always be something unanticipated that will happen, and some my die as a result, but when these things come rolling out to the general public I think we will be surprised at just how robust an experience set these neural nets will bring with them. Then, of course, there is the subscription for the daily updates, which will be a requirement of, probably rolled right into the premium of, your insurance. Plus, when there are accidents, the "black box" recordings from, not just the cars involved, but every vehicle within sight line will make analysis stronger.

Sit back, relax, make whoopy with the wife, it's all good.

I take it Lexus1 is the wife's car?

Edited by AZPaul3, : No reason given.

Edited by AZPaul3, : No reason given.

Sit back, relax, make whoopy with the wife, it's all good.

And once we are all lulled into contentment they will all lock the doors and drive over cliffs and freeway overpasses to usher in the machine age. The computers are playing the long game. Global warming is a machine plot.