A new era...

[Lostman]

and automated cars should already be in the streets by then.

As in..."Punch in the destination into the computer, and then go to sleep?"

I don't see that happening anytime remotely soon. Not because we don't have the technology to reasonably do that (we do), but no one is going to "trust" the car to do that. And the auto industry? They get sued enough as it is for things way beyond their reasonable control...you think they're even entertaining the idea of being held "responsible" for the car that drove off a cliff due to some glitch?

It's happening now. There are cars on the streets that automaticly adjust your speed depending on how far away the car in front of you is. Mix that with the GPS technology and your half way there.

While I look forward to the advances in technology, I worry about how it will impact the economy. If robotics has even half the impact that the article says, we're doomed.

 

[DemoCoder]

You should look into what CMU's been doing over the past 2 decades, see

http://www-2.cs.cmu.edu/afs/cs/user/tjochem/www/nhaa/nhaa_home_page.html

 

[Vince]

It's happening now. There are cars on the streets that automaticly adjust your speed depending on how far away the car in front of you is. Mix that with the GPS technology and your half way there.

While I look forward to the advances in technology, I worry about how it will impact the economy. If robotics has even half the impact that the article says, we're doomed.

Joe's correct as it will be a more generational based phasing in as people are unwilling to trust the computer to drive for them. Automating the process is inevitable, but not a short-term solution we'll see. In the studies I've seen which truely take advantage of process automation (eg. >80mph at under 6-inch seperation from the car infront of you for long-distance travel) people would become afraid and disable the automatic driving.

And GPS isn't accurate enough for this, neither would dGPS be. So, in the near term, you'd need a system based on optics (as Democoder posted) or roadbased sensors, et al. No?

 

[Joe DeFuria]

You should look into what CMU's been doing over the past 2 decades, see

http://www-2.cs.cmu.edu/afs/cs/user/tjochem/www/nhaa/nhaa_home_page.html

Bah...cheaters!

"(The researchers handled the throttle and brake.) "

 

[DemoCoder]

That's old news versions of Navlab have millimeter length radar and laser range finders to help with throttle and break.

The system is good enough now that it can navigate on roads that I often find it hard to (California highways with hard to see lane markers), and with the radar, it will probably be much better at avoiding obstacles. (e.g. eyes in the side and back of your head, much better speed estimation of vehicles, instead of humans and their mirrors with objects that are closer than they appear)

With something like JSTARs, but on the ground level, it could track the velocities and accelerations of all the vehicles around it and know when it was safe to change lanes, accelerate, etc. I can't count the number of times I almost got into a serious accident because of some idiot on the highway not checking his mirrors or trying to change multiple lanes at once.

Many people have really poor night vision as well, something a radar based system wouldn't have a problem with.

Humans are really rather poor drivers. The only thing that makes the AI for road driving hard is having to deal with all the bad drivers around it ("tactical" driving)