evenyoubrutus

May 6th, 2016 at 1:00 PM ^

I've been saying for a couple years that my kidsaye never have to take driver's ed. Or if they do it will be much different than when I learned 15 years ago.

Codeman

May 6th, 2016 at 1:42 PM ^

It's a matter of probability.  Is the autonomous vehicle more likely to cause a crash than the guy texting while driving?  How about the teenager snapchatting?  We tend to overestimate our own abilities.  Humans aren't fantastic drivers.  Autonomous vehicles won't be perfect, but they don't need to be, they just need to be better than us (which from what I've researched, they already are or will be very soon depending on the situation).  

ScruffyTheJanitor

May 6th, 2016 at 2:26 PM ^

While auto-autos are likely to be safer, the burden of an accident is 100 per cent on the manufacturer. IN other words, unless there is some sort of public component (some sort of guiding apparatus like a wire or something), I have always assumed that manufactutres would be a little gun shy.

Of course, the fact that I am always uneasy in a vehicle that I can't control probably means I will always be suspicious despite all reason. 

Muttley

May 6th, 2016 at 5:23 PM ^

I haven't been in an accident in 25 years.

And I can recognize anomalies, like a policeman re-routing traffic.

Who decides which car goes first at a stop sign?

Until there is some handshaking/negotiation between cars, I think it's just too dangerous.

Yes, it's a matter of probability.  I doubt an automonous car could go 25 years of daily driving without an accident without any communication/protocols between cars.

Inuyesta

May 6th, 2016 at 6:53 PM ^

Why on earth would you assume that there won't be communication/protocols between cars?

Google's cars made it through 1.3 million miles of driving before they caused an accident; a quick search says that the average person drives 13,476 miles per year, so that's roughly equivalent to an average person going 96 years of driving without causing an accident.  So, Google bested your mark by 400%.  

For reference, human drivers average an accident every 165,000 miles.  And that accident Google caused? A fender-bender where a bus hit the car at 15 mph; hardly the stuff of nightmares.

 

 

MaizeAndBlueWahoo

May 6th, 2016 at 7:39 PM ^

I think what you mean is Google went that far without being technically at fault in an accident.  They have been involved in 17, rather higher than the usual rate.  It's not coincidence that they get rear-ended a lot more than other cars; their driving patterns aren't what human drivers expect.  Their cars cause accidents in the same way that one person slamming on the brake for no reason might cause a rear-ender three cars back.

Also, as a significant chunk of their miles come from driving the same streets at piddling slow speeds over and over ad infinitum, comparing their miles to the average human's miles is a little bit apples to oranges.

MaizeAndBlueWahoo

May 6th, 2016 at 6:56 PM ^

No....for self-driving cars to become widespread, they have to be better than us at things we're bad at, and not worse than us at things we're good at.  If they behave in ways people don't like, they won't be adopted.  Much like how certain collision avoidance technologies have been shown not to reduce crashes, because people turn them off for being annoying.  Or how GPS's still occasionally try to drive people into a lake.

Also, frankly, autonomous cars still operate on a brute-force principle, more or less.  They don't stop because they see a stop sign.  They stop because they have a map that tells them there's a stop sign at that intersection.  Google brags very heavily about its self driving cars, but they're very limited in where they can go because Google has mapped every square inch of the routes they drive, and not just in the Google Maps thingies that you and I see, but every detail a driver would need to know, such as how many lanes, what purpose (turn-only lanes, etc.), stop signs, traffic lights, speed limits, and so on.  Take a Google car outside its mapped comfort zone and it's not self-driving anymore.

Codeman

May 6th, 2016 at 8:54 PM ^

Agreed, it will not be an instantaneous shift. Autonomous vehicles will replace different market segments at a time (e.g. semi-trailers travelling on interstates will likely be one of the first).  

The concerns in your second paragraph are certainly warranted.  My only counter to your arguments is the need to incorporate projection of certain technologies going forward.  In this case, I believe that the inclusion of sensors into new cars will allow us to passively map a large segment of the US road systems over the next few years.  Also, the recent advances in deep learning are particularly promising for more intelligent driving by autonomous vehicles (computer vision has progressed leaps and bounds over the last 2 years in particular).  

I may be a bit optimistic, but I don't believe it's unwarranted.

OwenGoBlue

May 6th, 2016 at 2:11 PM ^

Whenever they debut they're going to in-effect be manned vehicles with an auto-pilot setting.

I think this is further away than has been reported; the technology is there and is so much safer than human drivers, but the liability question is going to keep them off the road. 

TheTruth41

May 6th, 2016 at 5:28 PM ^

You see this now with cars that assist in braking for you when it senses an object in the car's path.  The driver will feel in control but the auto part will make up for deficiencies in user error.  Over time they'll encompass more and more until eventually they're all self-driven.  I'd see it more as a process and evolution rather than a hard line where they start to go all automomous.  Maybe some cars but I don't know if all would be.  Hard to believe.

LSAClassOf2000

May 6th, 2016 at 2:18 PM ^

Maximum Overdrive ruined me for lawn mowers for about a year or so - that was some creepy shit to be hauled to see by a rather absent-minded sitter at the age of seven. Pretty sure the only reason I got in was that they were an employee at the Canton 6 at the time and nobody seemed to care about the age thing, or they weren't paid enough to care. Either way, I got to see a lot of films in the early to mid-1980s that way, some of them before I should have. It shows. 

ghostofhoke

May 6th, 2016 at 1:54 PM ^

I posit the creators of this technology will probably focus on all the solutions to the one offs like construction when it's appropriate and for the meantime focus on getting the 95% of the technology correct. That seems like a reasonable approach to the business and to say that the fact there are one-offs and that means the idea is doomed is absolutely moronic.




Sent from MGoBlog HD for iPhone & iPad

a different Jason

May 6th, 2016 at 1:35 PM ^

Now I wish I hadn't used that word. The thing about semis, they are at work so they have a different mindset. They often have someone in their ear saying they should have been there yesterday.

boliver46

May 6th, 2016 at 1:40 PM ^

but I commute 52 miles each way to work and back - of which 50 of them are currently under construction.  There are plenty of switchbacks along my route, where trucks are required and notified to use a particular lane (right or left depending on barriers and on-ramp traffic).

They either don't see, or DGAF about the signs as many do not make the appropriate switch...OR...they DGAF that you're currently occupying the lane they are supposed to be in and do not even signal and start coming over.

I wish I had a passenger or dash cam to record the nonsense I see on the daily.

kvnryn

May 6th, 2016 at 3:52 PM ^

This is basically the point Tesla is at right now (although a little more advanced than your description even). This is a video of a guy's entire commute using autopilot as much as possible. It's obviously a long watch, but he does a great job of explaining when and why autopilot is engaged/disengaged, and the situations that are difficult for autopilot to handle. In the end, it's a 45 minute commute and autopilot is in use 85% of the time.

 

 

Gameboy

May 6th, 2016 at 5:17 PM ^

This article is FUD. There is no system that is even close to run on its own next year, not even Google. This is most likely GM leasing out modified Bolt with auto driving capabilities to Lyft for testing. Auto taxi is still decades away.

mGrowOld

May 6th, 2016 at 1:05 PM ^

I cant even imagine how big the lawsuit will be the first time somebody is killed in an accident involving a driverless car.  It will be interesting to see if the industry survives that challenge.