StephenRKass

May 6th, 2016 at 1:10 PM ^

I'm still pretty doubtful. Don't see how a car can foresee all the possible factors necessary. Only way I think it could work would be if this was used on "tracks," which would be pretty limited. However, that could allow for you to drive to a bay where you get on a "track" (say on an interstate) and then are driverless/autonomous until you get off the interstate.

JLo

May 6th, 2016 at 1:32 PM ^

When a user hails a car through Lyft, the user tells you a) where they are, and b) where they're going.  Lyft/GM can make sure that they only send out the autonomous vehicles when the route between the start and finish points is well-mapped and navigable by our new robot overlords.  

As techological capabilities and mapping data improve, they can gradually increase the variety of routes to which they send out the autonomous vehicles.

ghostofhoke

May 6th, 2016 at 1:13 PM ^

I'm sure we're about to be deluged with comments from the car industry crowd that will tell us the future is not a real thing and it will never happen. Cue Ford employees....now



Sent from MGoBlog HD for iPhone & iPad

PopeLando

May 6th, 2016 at 1:26 PM ^

A car can probably self drive under controlled conditions. I'm interested in what kind of shape the roads need to be in for this to work; my guess is high quality with little deviation. I'm really looking forward to proliferation of self driving vehicles, because it implies that our road system has been kept to standard. A friend of mine studied drivers as part of his data management systems analysis for his grad degree. Basically, his conclusion is that we as drivers are not characterized as much by our adherence to rules as much as our ability to conceive of alternative actions: this is why a power outage at a stop light is an inconvenience rather than a disaster, or how we can deal with a funeral procession. It would be impressive if a machine could automatically process everything we do when driving. Sorry for the essay. This is fascinating, but I need to know more than what the article tells us.

Gallagher

May 6th, 2016 at 2:20 PM ^

Most of what I've seen is that these cars need roads with quality lane markers. In faded lane markers on the road, they're confused as to what to do.

There's pothole avoidance tech out there, so that wouldn't be an issue. And in the future, autonomous cars become mainstream, infrastructure costs would go down too, so it wouldn't be a burden on the taxpayer.



Sent from MGoBlog HD for iPhone & iPad

Codeman

May 6th, 2016 at 1:34 PM ^

Care to elaborate on that?  Self driving cars will become more common once the data shows they're significantly safer than humans.  Why wouldn't that be a good thing?  1.3 million people die in car crashes/year.  It's insane that we're OK with that, and wouldn't welcome changes that will reduce that number.

Njia

May 6th, 2016 at 1:46 PM ^

There are significant issues with respect to liability and regulatory laws, for starters. When current laws went on the books, no one even remotely contemplated them. This article in February's Observer (http://observer.com/2016/02/why-driverless-cars-will-screech-to-a-halt/) raises some interesting questions that don't have easy answers. It boils down to the fact that while computers make decisions, they don't use judgment. Human judgment is flawed, sure; but that "gut check" or "intuition" (whatever you want to call it) is far more complicated that simply following a decision tree. I disagree with the article's author that autonomous vehicles are "doomed," but I think there's going to be a huge reality check that will cause some re-evaluating. Right now, we are still coming up the hype curve.

Codeman

May 6th, 2016 at 1:55 PM ^

This is a great point.  I'm pretty heavily invested in the area so I certainly have some bias on the subject.  Obviously time will tell, but my educated guess is that the data collected will be compelling enough to get the regulators moving quickly.  If driverless vehicles are as successful as I think they will be, the market forces pushing it forward will be enormous.  Lowering property damage, medical bills, costs of shipping, and of course lost lives would have a massive economic impact.  But of course, I could be naive and completely wrong.

Greg McMurtry

May 6th, 2016 at 1:41 PM ^

Everybody else can drive them but not me. My drive in the morning consists of 100 cars in the left lane, zero in the right lane at an intersection (red light). I go in the right lane, light turns green, I zip ahead of the first car as I am also the first car. This saves me 10 minutes. Can a computer do that? Will it?



Sent from MGoBlog HD for iPhone & iPad

Codeman

May 6th, 2016 at 1:36 PM ^

The technology is getting close.  Advances in deep learning and improved sensors has allowed a huge amount of progress to occur over the last few years.  

As for the regulators, it may take more time.  What will happen is that the cars will be running simulations while humans drive.  This data can then be used to show the percentange of accidents that would have been preventable if an AI was driven (and the number that would have occurred if an AI was driving).  Once that data is available and shows a significant improvement in safety, it will be far easier to argue for fully autonomous vehicles.  

I'd expect to start seeing autonomous vehicles on the road around the end of the decade.  Autonomous semi's will likely quickly follow.

FYI: I've been studying AI/machine learning for the past 6 months with particular interest in autonomous vehicles.

PopeLando

May 6th, 2016 at 1:41 PM ^

Autonomous tractor trailers should really be the first application. Imagine if all a semi driver had to do was to meet his truck at the correct offramp, then drive it to the in-town destination or customer, then drop it off at the correct on-ramp to send it to its next destination. Rinse and repeat.

PopeLando

May 6th, 2016 at 1:45 PM ^

Long ago, the US Army almost banned its use of the sewing machine to make uniforms. Why? Because it would put too many seamstresses out of work. Progress means change. Any new technology will eliminate some positions and create others. Might not be immediately, but it's inevitable. The alternative, of course, is to stop innovating. Is that what you're suggesting?

PopeLando

May 6th, 2016 at 4:10 PM ^

There's a certain segment of the internet commenter population which can pull politics into any discussion anywhere. It's actually quite impressive. The mental leap required to turn a conversation about driverless cars into a diss of a presidential candidate is...completely off the wall.I would love to see a flowchart of that thought process.