Self-driving semis trucks approved for testing in Nevada

[roadfro]

From The Guardian (5/6/15): Nevada clears self-driving 18-wheeler for testing on public roads

Drone trucks could soon be plying US highways after Nevada authorities on Wednesday granted a license to test self-driving trucks on public roads.
<...>
Daimler's 18-wheel Inspiration has now been certified for use on public roads in the state, and yesterday the non-human (well, less-human) big rig rolled out across the Hoover Dam, negotiating some (but not all) turns and twists all by itself. For the tougher curves, it had some help from a driver inside the cab.

Google's own automated car is also licensed for use in Nevada; both the Daimler truck and the Google car are authorized for use under Nevada's autonomous driving law, which "requires that drivers obtain a special endorsement on their driver's licenses"  according to the National Highway Traffic Safety Administration.

[Sykotyk]

It is a bit ironic that the 'self-driving' vehicles are required to have drivers with special endorsements.

[Pete from Boston]

I think I've seen this one. 

http://youtu.be/5MtAMc4i8OA

[mcarling]

I own that film in iTunes.  I never imagined that the truck did not have a human driver, though he is never shown.

This will be a huge improvement for safety going forward.  Trucks will become more predictable and problems with sleepy, distracted, or ill drivers will become a thing of the past.  Trucks will stay in their lanes better, not follow too closely, keep a safe speed at all times, etc.

The technological problems have all been solved.  Now it's a matter of public acceptance and bring the law up to where the technology is.

[roadfro]

Maybe I don't drive around trucks too often... Sure there are issues with tired/ill truckers, but I just don't see some of those other safety issues with trucks. They are professionally licensed drivers, and typically operate a truck way better than the average driver operates a car. You can make a case for the speed differential, but it's a truck and to be expected.

[iBallasticwolf2]

I agree. Truck drivers are usually good drivers. In fact it should be more of a problem with bad car drivers. There are so many bad car drivers that passing on the right had to be legalized. The self-driving cars will cut down on @SSHOLE drivers.

[Pete from Boston]

Is this going to become a "Cars don't kill people–drivers kill people" public debate?

The likely internet arguments practically write themselves.

"A driver removed is a passenger that can't kill."

"First they photographed our license plates, then they put a VMT chip in our car, now they want to control where it goes!"

Can't wait.

[roadfro]

Is this going to become a "Cars don't kill people–drivers kill people" public debate?

Let us not allow it to become that...

[mcarling]

I fully agree that the average truck driver is a better driver than the average car driver.  However, I have certainly seen some aggressive truck drivers.  For example, several months ago, in rain so heavy that visibility was such that reading the signs was impossible, I was driving on the freeway at the speed limit and a truck was following me at a distance 10 to 15 feet, trying to intimidate me into driving faster.  I eventually exited and then re-entered the freeway to escape him.

Also, a collision with a truck is typically more severe than a collision with another car.

Simply for economic reasons, the trucking companies will be quick to replace drivers with automation, as fast as the law allows.  I expect to see a lot of unemployment among professional drivers within five years.

Anyway, it's irrelevant that most truck drivers are better drivers than most car drivers because no one is proposing that truck drivers be replaced by car drivers.  Automated, self-driving trucks are safer than truck drivers and that is relevant because truck drivers will be replaced by automated, self-driving trucks.

Of course, self-driving cars are coming too, but I expect that transition to be much slower.

[roadfro]

I doubt automated trucks will be widespread and displacing truck driver jobs in 5 years. The technology is not widespread (and thus not yet inexpensive enough to deploy to large fleets), and I don't think most states have even passed appropriate legislation to allow automation yet. As it is in Nevada, you still need a driver behind the wheel, one who has an endorsement to drive a vehicle with autonomous capabilities.

[Henry]

It would be an awesome thing to have! If they could make self-driving cars, why not the big rigs too?

[Sykotyk]

The problem with switching to autonomous driving while requiring a driver to be in the vehicle in case of necessity... is boredom. The same problem train and plane operators encounter. When the vehicle drives itself, it's hard to keep yourself alert at all times for an emergency situation. The actual act of driving keeps you alert.

The level of drivers falling asleep while their vehicle drives itself will probably be a much more troublesome statistic. And when the technology fails and the human is required to take control is when those problems will present itself.

If you want automation for trucks, simply put the freight on a railcar and ship it that way. Mile-for-mile it's so much cheaper. The problem is it tends to be major metro area to major metro area. And the number of cities serviced isn't near what train service used to be.

[Brandon]

If you want automation for trucks, simply put the freight on a railcar and ship it that way. Mile-for-mile it's so much cheaper. The problem is it tends to be major metro area to major metro area. And the number of cities serviced isn't near what train service used to be.

The problem with that is that trucks are used more and more often to get an intermodal container from the railyard to its final destination which may be nowhere near the railyard.

[mcarling]

The problem with switching to autonomous driving while requiring a driver to be in the vehicle in case of necessity... is boredom. The same problem train and plane operators encounter. When the vehicle drives itself, it's hard to keep yourself alert at all times for an emergency situation. The actual act of driving keeps you alert.

Agreed.

The level of drivers falling asleep while their vehicle drives itself will probably be a much more troublesome statistic.

I very much doubt that, but we'll see.  I expect to see the opposite statistic: that human drivers who took control in an emergency more often made things worse.  I expect that in 20 or so years that it will be illegal for a human driver to take control from the autonomous system in the few vehicles (mostly produced 2015-2020 or perhaps 2015-2025) in which it's even possible for a human to take control.

And when the technology fails and the human is required to take control is when those problems will present itself.

The requirement to have a human in the vehicle who can take control in an emergency is purely a legal requirement, intended only a temporary measure while autonomous driving systems are proving themselves.  There is no technical need for a human driver to be present.  The autonomous systems already have much lower failure rates than human drivers, even much lower than professional drivers.  The legislators are merely playing CYA.  That will change over the next few years, driven mainly by demand from the insurance companies.

[vdeane]

There is no technical need for a human driver to be present.

What about when driving outside the TINY MINORITY of roads that Google has mapped in the detail necessary for a self-driving car to operate at all?  What about on a snow covered road outside of sunny California?  What about work zones that Google doesn't know about (many of which last for less than a day, such as for a bridge inspection).  The current self-driving cars on the road are a smoke and mirrors trick designed to delude the media into thinking self driving cars are imminent.  They're not.  There is a LOT of work yet to be done to make them usable AT ALL outside of the Bay Area.

That will change over the next few years, driven mainly by demand from the insurance companies.

I doubt very much the insurance companies would be in favor of anything that might threaten the legal requirement of buying car insurance.  Take the human out of the equation and put control in the hands of computers that don't cause accidents, and the legal reasoning for making it a requirement (that someone doesn't get screwed if they're in an accident with someone who had no insurance and was at fault) goes away.  Car insurance would become like renter's insurance, nice to have if there's a disaster or something, but which impacts nobody but yourself if you don't have it.

[mcarling]

There is no technical need for a human driver to be present.

What about when driving outside the TINY MINORITY of roads that Google has mapped in the detail necessary for a self-driving car to operate at all?  What about on a snow covered road outside of sunny California?  What about work zones that Google doesn't know about (many of which last for less than a day, such as for a bridge inspection).

The map is not the terrain.  The map data only helps autonomous vehicles to find their destination.  All safety-related data used to drive autonomous vehicles is gathered in real-time from the actual terrain.  The worst-case scenario with bad map data is that an autonomous vehicle has to drive around a while to find its destination.  The regulators and the insurance companies and the voters don't care much if an autonomous vehicle gets a bit lost for a while.

There is a LOT of work yet to be done to make them usable AT ALL outside of the Bay Area. 

Autonomous vehicles are being used outside the Bay Area.

I doubt very much the insurance companies would be in favor of anything that might threaten the legal requirement of buying car insurance.  Take the human out of the equation and put control in the hands of computers that don't cause accidents, and the legal reasoning for making it a requirement (that someone doesn't get screwed if they're in an accident with someone who had no insurance and was at fault) goes away.  Car insurance would become like renter's insurance, nice to have if there's a disaster or something, but which impacts nobody but yourself if you don't have it.

Your conclusion doesn't follow from your premise and your premise is wrong anyway.  Autonomous vehicles do and will continue to have failures and cause accidents, just at a much lower rate than human drivers.  The absence of a human driver does not remove the legal reasoning for legal liability, though it might eventually lead to increased political pressure for no-fault auto insurance in some states, which the insurance companies love because it eliminates expensive and risky litigation.  The fight over traditional versus no-fault insurance is essentially a fight between the insurance companies and the trial lawyers.

[vdeane]

The map is not the terrain.  The map data only helps autonomous vehicles to find their destination.  All safety-related data used to drive autonomous vehicles is gathered in real-time from the actual terrain.  The worst-case scenario with bad map data is that an autonomous vehicle has to drive around a while to find its destination.  The regulators and the insurance companies and the voters don't care much if an autonomous vehicle gets a bit lost for a while.

Not with Google's cars.  Google literally sends out cars with sensors to profile the roads their self-driving cars will operate on down to the tiniest pothole.

Your conclusion doesn't follow from your premise and your premise is wrong anyway.  Autonomous vehicles do and will continue to have failures and cause accidents, just at a much lower rate than human drivers.  The absence of a human driver does not remove the legal reasoning for legal liability, though it might eventually lead to increased political pressure for no-fault auto insurance in some states, which the insurance companies love because it eliminates expensive and risky litigation.  The fight over traditional versus no-fault insurance is essentially a fight between the insurance companies and the trial lawyers.

And why should I have liability for something that I have no control over does?  If that were the case, I would refuse to so much as step inside a self-driving vehicle for any reason.  And again, if you go to no-fault where there is no chance that someone else not having insurance would cost you money, why require it?  If someone wants to risk the financial loss or has the money to cover the damages to their own car, let them.  They would be harming nobody in those circumstances.

Between my current car and my last one, I have essentially had a car for six years now.  In those six years, I estimate that I have personally paid the insurance company more money than they paid in the single claim we made on my Accord 3-4 years ago; and that is AFTER factoring in the fact that my parents were footing the bill for four of the six years I was driving (I was in college for those years and the car was still technically registered to my Mom during that time).  It's highway robbery.  But the requirement to have insurance makes sense right now ONLY because someone else might be depending on YOUR insurance in an accident.

If you think self-driving cars are coming soon, I invite you to read the article below.  It should clear up the misconceptions the media is feeding the public about the technology.
http://www.technologyreview.com/news/530276/hidden-obstacles-for-googles-self-driving-cars/

[DJStephens]

this discussion brings to mind "Maximum Overdrive", a film based on a Stephen King book.  Film starred Emilio Estevez, better known as "Sporto" in the Breakfast Club.   Believe he was either half or full brother to Charlie Sheen.   

[triplemultiplex]

Google's efforts for a self-driving vehicle are not necessarily how this technology will ultimately be implemented.  The meticulous mapping of all those routes is time consuming and prone to being outdated.  Better to have systems that don't rely on following an internal map of potholes and driveways and instead be able to respond to what it senses in front of it.  And these smart cars should have smart roads; where things like traffic signals should be able to communicate what phase they are in and when they will next cycle.  Roads A B and C can communicate their real-time traffic conditions and the car can make it's decision on what route to take based on that.

I won't dismiss the entire concept just because of Google's shortcomings or insurance company mumbo-jumbo.  This technology is going to revolutionize transportation within my lifetime.

[iBallasticwolf2]

Google's efforts for a self-driving vehicle are not necessarily how this technology will ultimately be implemented.  The meticulous mapping of all those routes is time consuming and prone to being outdated.  Better to have systems that don't rely on following an internal map of potholes and driveways and instead be able to respond to what it senses in front of it.  And these smart cars should have smart roads; where things like traffic signals should be able to communicate what phase they are in and when they will next cycle.  Roads A B and C can communicate their real-time traffic conditions and the car can make it's decision on what route to take based on that.

I won't dismiss the entire concept just because of Google's shortcomings or insurance company mumbo-jumbo.  This technology is going to revolutionize transportation within my lifetime.

I completely agree with you.

[vdeane]

Regarding the mapping... given the obvious limitations, one should ask why Google is bothering with it at all.  It seems to me that they're doing it because they can't do it another way (yet) and want the publicity that comes from being first.  I would put the arrival of the the technology as further out than what the media claims.  Remember, if technology moved at the pace the media claimed, we'd all have flying cars by now.

As for smart roads... good luck getting states and municipalities (or even the federal government) to pay for that.

And you can take my stick shift, manually driven car from my cold, dead hands.

[empirestate]

And you can take my stick shift, manually driven car from my cold, dead hands.

Well, again, that's really not what it's going to come down to. You can keep your car as long as you please, and if they're still making them after you get rid of the current one, you can get another. What will eventually happen is just that there will be no such thing being manufactured for you to clutch in said hands.

However, I do agree with you that the timeline of this happening won't be as rapid as many current reports suggest, so it's likely that by the time it's a reality, the coldness and deadness of those hands will be literally true.


iPhone