Robot cars finish DARPA Challenge



 Science > Physics > Robot cars finish DARPA Challenge

LINK TO THIS PAGE  


rating :  0   |  0


  Page 1 of 1

1

 
Topic: Science > Physics
User: "Randy Poe"
Date: 11 Oct 2005 03:33:23 PM
Object: Robot cars finish DARPA Challenge
http://www.betterhumans.com/News/4711/Default.aspx
I stumbled on this while looking for something else at
DARPA. For the first time, a robot (five of them) has
completed the DARPA Challenge, an autonomous vehicle
cross-country race.
- Randy
.

User: "Sbharris[atsign]ix.netcom.com"

Title: Re: Robot cars finish DARPA Challenge 11 Oct 2005 04:06:53 PM
Randy Poe wrote:

http://www.betterhumans.com/News/4711/Default.aspx

I stumbled on this while looking for something else at
DARPA. For the first time, a robot (five of them) has
completed the DARPA Challenge, an autonomous vehicle
cross-country race.

- Randy

COMMENT:
For those of you who don't know, DARPA is the Defense Advanced Research
Projects Agency, the central R&D organization of the DoD,
which in this case was trying (by offering a $2 million prize) to
enlist university help to meet congressional mandate to have 30% of
military vehicles autonomous by 2015. DoD has been feeling the pressure
on its long and poorly-protected land-supply lines in Iraq, and is
still smarting from the political fallout of what we might call the
"Jessica Lynch Syndrome." This robot vehicle thing is the intended
"fix" for that, but as with all technology, it will eventually be
coming to a civilian use near you.
Technological progress *is* accelerating. This weekend I actually drove
to Primm, just across the Nevada/California state line (East of Whiskey
Pete's, behind the Buffalo Bill's casino), and watched the robot
automobiles navigate the Nevada desert without aid, following an
internal map given them 2 hours before, using only their own image
systems and GPS sensors. (human control was limited only to several
kinds of a stop switches). The robots went though tunnels under the
main I-15 highway (having to give up, and then re--acquire, GPS lock),
over hills, train tracks, through dust clouds on a dry lake bed, and
finally down a winding dirt road in a mountain canyon. They even passed
each other.
Last year, they all crashed ridiculously within a few miles. This year,
5 of them made it all the way through the 130-odd mile course, which
was in many places not marked, most of it out of site from the
start-line, and nearly all of it on dirt. Though they didn't announce
it officially, we in the crowd knew the Stanford car had probably won,
inasmuch as it was launched second (between the two top seed
Carnegie-Mellon 'bots), but despite several minutes head-start by the
leader, managed to pass the leader and still come in first. The first
three robots made it to the finish line after 7 hours of racing, all
within 20 minutes, or so. They probably would have been faster, if they
hadn't had an absolute speed limit for the course.
Carnegie-Mellon greatly outspent Stanford and expected to win if
anybody did. The Stanford team looked inordinately pleased with
itself, and seemed to have drawn on a larger portion of student body.
Half the bleachers were shouting "STAN-LEE!", "STAN-LEE!" and not
because they were fans of spiderman. "Stanley" was the converted Taureg
used by the Stanford team, and the VW people looked pleased with
themselves, too. Their Taureg logo "Drivers Wanted" had been changed to
"Drivers Not Required."
A friend of mine in the big DARPA tent next to the bleachers where we
saw the 'bots leave and return, saw an international news article
appear minutes after the finish, by googling the web with his cell
phone(!). Actually it was a TREO-- a complete palmtop computer and
camera as well (his photos go by Bluetooth over the cell link to
internet to his friends or home computer, where they get tagged and
added to his "Life BLOG"--- what, you don't do that, also?) About 12
million other people queried the DARPA website from around the world,
over the previous couple of hours. All this was pretty much science
fiction, even 5 years ago. As was my $30 microwave oven. Every time I
look around me these days, I see a new piece of technology sprout up
like a mushroom, or an old piece suddenly become ridiculously
inexpensive.
As Ray Kurzweil reminds us, in terms of knowledge and progress, we'll
repeat the entire gain of the 20th century in the first 14 years of the
21st. Then do it again in the next 7 years after that. Then again in
3.5 years. Then 1.75 years... Somewhere about that time (2030, give or
take), either magical things will happen, or else Skynet will wake up
and terminate us. In any case, be there or be square. (After seeing the
robot cars take off down the desert road at over 30 mph, with laser
scanners and computers in place of human drivers, Skynet just got a
little more real for ME.)
SBH
.
User: "Ron Baker, Pluralitas!"

Title: Re: Robot cars finish DARPA Challenge 11 Oct 2005 08:05:46 PM
"Sbharris[atsign]ix.netcom.com" <sbharris@ix.netcom.com> wrote in message
news:1129064813.681229.326640@g43g2000cwa.googlegroups.com...


Randy Poe wrote:

http://www.betterhumans.com/News/4711/Default.aspx

I stumbled on this while looking for something else at
DARPA. For the first time, a robot (five of them) has
completed the DARPA Challenge, an autonomous vehicle
cross-country race.

- Randy



COMMENT:

For those of you who don't know, DARPA is the Defense Advanced Research
Projects Agency, the central R&D organization of the DoD,

<snip to save bandwidth>

robot cars take off down the desert road at over 30 mph, with laser
scanners and computers in place of human drivers, Skynet just got a
little more real for ME.)

Interesting. Nice post.
--
rb
.

User: "Traveler"

Title: Re: Robot cars finish DARPA Challenge 11 Oct 2005 06:52:41 PM
On 11 Oct 2005 14:06:53 -0700, "Sbharris[atsign]ix.netcom.com"
<sbharris@ix.netcom.com> wrote:


Randy Poe wrote:

http://www.betterhumans.com/News/4711/Default.aspx

I stumbled on this while looking for something else at
DARPA. For the first time, a robot (five of them) has
completed the DARPA Challenge, an autonomous vehicle
cross-country race.

- Randy



COMMENT:

For those of you who don't know, DARPA is the Defense Advanced Research
Projects Agency, the central R&D organization of the DoD,
which in this case was trying (by offering a $2 million prize) to
enlist university help to meet congressional mandate to have 30% of
military vehicles autonomous by 2015. DoD has been feeling the pressure
on its long and poorly-protected land-supply lines in Iraq, and is
still smarting from the political fallout of what we might call the
"Jessica Lynch Syndrome." This robot vehicle thing is the intended
"fix" for that, but as with all technology, it will eventually be
coming to a civilian use near you.

Technological progress *is* accelerating. This weekend I actually drove
to Primm, just across the Nevada/California state line (East of Whiskey
Pete's, behind the Buffalo Bill's casino), and watched the robot
automobiles navigate the Nevada desert without aid, following an
internal map given them 2 hours before, using only their own image
systems and GPS sensors. (human control was limited only to several
kinds of a stop switches). The robots went though tunnels under the
main I-15 highway (having to give up, and then re--acquire, GPS lock),
over hills, train tracks, through dust clouds on a dry lake bed, and
finally down a winding dirt road in a mountain canyon. They even passed
each other.

Last year, they all crashed ridiculously within a few miles. This year,
5 of them made it all the way through the 130-odd mile course, which
was in many places not marked, most of it out of site from the
start-line, and nearly all of it on dirt. Though they didn't announce
it officially, we in the crowd knew the Stanford car had probably won,
inasmuch as it was launched second (between the two top seed
Carnegie-Mellon 'bots), but despite several minutes head-start by the
leader, managed to pass the leader and still come in first. The first
three robots made it to the finish line after 7 hours of racing, all
within 20 minutes, or so. They probably would have been faster, if they
hadn't had an absolute speed limit for the course.

Carnegie-Mellon greatly outspent Stanford and expected to win if
anybody did. The Stanford team looked inordinately pleased with
itself, and seemed to have drawn on a larger portion of student body.
Half the bleachers were shouting "STAN-LEE!", "STAN-LEE!" and not
because they were fans of spiderman. "Stanley" was the converted Taureg
used by the Stanford team, and the VW people looked pleased with
themselves, too. Their Taureg logo "Drivers Wanted" had been changed to
"Drivers Not Required."

A friend of mine in the big DARPA tent next to the bleachers where we
saw the 'bots leave and return, saw an international news article
appear minutes after the finish, by googling the web with his cell
phone(!). Actually it was a TREO-- a complete palmtop computer and
camera as well (his photos go by Bluetooth over the cell link to
internet to his friends or home computer, where they get tagged and
added to his "Life BLOG"--- what, you don't do that, also?) About 12
million other people queried the DARPA website from around the world,
over the previous couple of hours. All this was pretty much science
fiction, even 5 years ago. As was my $30 microwave oven. Every time I
look around me these days, I see a new piece of technology sprout up
like a mushroom, or an old piece suddenly become ridiculously
inexpensive.

As Ray Kurzweil reminds us, in terms of knowledge and progress, we'll
repeat the entire gain of the 20th century in the first 14 years of the
21st. Then do it again in the next 7 years after that. Then again in
3.5 years. Then 1.75 years... Somewhere about that time (2030, give or
take), either magical things will happen, or else Skynet will wake up
and terminate us. In any case, be there or be square. (After seeing the
robot cars take off down the desert road at over 30 mph, with laser
scanners and computers in place of human drivers, Skynet just got a
little more real for ME.)

Thanks for the report. This is all very impressive as far as the
engineering prowess of human beings is concerned. However, I don't
think Darpa's Grand Challenge has gotten us any closer to a truly
general adaptive artificial intelligence being than we were fifty
years ago. IMO, what those engineers accomplished is just GOFAI (good
old-fashioned AI). We will need a major breakthrough before AI appears
on the scene. That's the hard part. The engineering aspect of it will
be trivial once we figure it out.
Louis Savain
Why Software Is Bad and What We Can Do to Fix It:
http://www.rebelscience.org/Cosas/Reliability.htm
.
User: "hanson"

Title: Science, Technology, the DARPA Challenge 12 Oct 2005 02:08:40 AM
"Traveler" <traveler@nospam.net> wrote in message
news:nhjok1h8qjgmdcd2o029h5g22oqds0see6@4ax.com...
[Randy Poe ] wrote:

http://www.betterhumans.com/News/4711/Default.aspx


[Sbharris] <sbharris@ix.netcom.com> wrote:

DARPA is the Defense Advanced Research Projects Agency,
the central R&D organization of the DoD,...... [snip]....
I see a new piece of technology sprout up like a mushroom,
or an old piece suddenly become ridiculously inexpensive.

.... in terms of knowledge and progress, we'll
repeat the entire gain of the 20th century in the first 14 years of the
21st. Then do it again in the next 7 years after that. Then again in
3.5 years. Then 1.75 years... Somewhere about that time (2030, give or
take), either magical things will happen, or else Skynet will wake up
and terminate us. In any case, be there or be square. (After seeing the
robot cars take off down the desert road at over 30 mph, with laser
scanners and computers in place of human drivers, Skynet just got a
little more real for ME.)


[Louis]

Thanks for the report. This is all very impressive as far as the
engineering prowess of human beings is concerned. However, I don't
think Darpa's Grand Challenge has gotten us any closer to a truly
general adaptive artificial intelligence being than we were fifty
years ago. IMO, what those engineers accomplished is just GOFAI (good
old-fashioned AI). We will need a major breakthrough before AI appears
on the scene. That's the hard part. The engineering aspect of it will
be trivial once we figure it out.
Louis Savain

[hanson]
You are not alone in that assessment.
Not only with AI (not the uncle) but also with physics, as....
== Professor Carver A. Mead of Caltech, expressed his sentiments
with: "It is my firm belief that the last seven decades of the twentieth
century will be characterized in history as the dark ages of physics."
....and in some minds the problem is even more wide spread as:
== F.A Hayek, Nobel Price for Economy says: "In the future, Humanity
will see in our Epoch an Era of superstition, essentially associated
with the names of Marx, Freud and Einstein!"
But trumping even these 2 luminaries was none other but poster
== James Driscoll aka Spaceman who did shine with t/is remark:
"A truly new thought appears very seldom"
And it is precisely that "new thought" we are still waiting for...
ahahaha... ahahanson
.

User: "Sbharris[atsign]ix.netcom.com"

Title: Re: Robot cars finish DARPA Challenge 11 Oct 2005 08:05:27 PM
Traveler wrote:

On 11 Oct 2005 14:06:53 -0700, "Sbharris[atsign]ix.netcom.com"
<sbharris@ix.netcom.com> wrote:
Thanks for the report. This is all very impressive as far as the
engineering prowess of human beings is concerned. However, I don't
think Darpa's Grand Challenge has gotten us any closer to a truly
general adaptive artificial intelligence being than we were fifty
years ago. IMO, what those engineers accomplished is just GOFAI (good
old-fashioned AI). We will need a major breakthrough before AI appears
on the scene. That's the hard part. The engineering aspect of it will
be trivial once we figure it out.

Louis Savain

I don't agree. Estimated power of the human brain is around 100
teraflops, give or take (massively parallel of course). We're not
within 1% of that now, on the biggest computers ever built, let alone
anything you can put on a robot.
You can't program true strong AI on an insect-sized brain. Which is all
the hardware we have right now. You CAN do insect-like stuff. Which we
are doing. This is one example.
SBH
.
User: "Traveler"

Title: Re: Robot cars finish DARPA Challenge 12 Oct 2005 09:14:49 AM
On 11 Oct 2005 18:05:27 -0700, "Sbharris[atsign]ix.netcom.com"
<sbharris@ix.netcom.com> wrote:


Traveler wrote:

On 11 Oct 2005 14:06:53 -0700, "Sbharris[atsign]ix.netcom.com"
<sbharris@ix.netcom.com> wrote:


Thanks for the report. This is all very impressive as far as the
engineering prowess of human beings is concerned. However, I don't
think Darpa's Grand Challenge has gotten us any closer to a truly
general adaptive artificial intelligence being than we were fifty
years ago. IMO, what those engineers accomplished is just GOFAI (good
old-fashioned AI). We will need a major breakthrough before AI appears
on the scene. That's the hard part. The engineering aspect of it will
be trivial once we figure it out.


I don't agree. Estimated power of the human brain is around 100
teraflops, give or take (massively parallel of course). We're not
within 1% of that now, on the biggest computers ever built, let alone
anything you can put on a robot.

This is Hans Moravec's (CMU) hopelessly flawed argument. The truth is
that, at least 1/4 of the brain's 100 billion neurons are used for
managing the body's various functions. Another 1/4 is used for such
things as procreation, thirst, hunger, olfactory and gustatory
processing, etc... None of these things are necessary in a synthetic
robot. In addition, only a very small fraction of the brain's neurons
are active at any one time. We have this uncanny ability to focus on
one thing at a time while ignoring all others. This has a tremendous
effect in lowering the computing power requirements. But that's not
all. The brain's neurons, regardless of what the self-appointed
pundits insist on claim, are just discrete signal (pulse) processors.
We don't need complex floating point-like operations to process
discrete signals. In my estimation, we can lower the 100 teraflop
estimate to just one teraflop.

You can't program true strong AI on an insect-sized brain. Which is all
the hardware we have right now. You CAN do insect-like stuff. Which we
are doing. This is one example.

A honeybee has an an amazingly sophisticated behavioral repertoire.
Yet, it has only about a million neurons, a good percentage of which
is used for purposes having little to do with its behavior. We could
certainly emulate the complexity of a honeybee's behavior on a modern
desktop machine if we knew how. If we only understood the principles
of intelligence, we would accomplish amazing things with the machines
we have now. We don't need the power of the human brain to have a
self-driving machine that could learn to drive, not only in the
desert, but also in the streets and highways of NYC or London during
rush hour.
One more thing. It is complete nonsense that we need to wait twenty
years or so before we can have enough computing power to simulate the
brain. We can do it with the technology we have now, using distributed
clusters, if only we understood the principles. IMO, the computing
power argument is put out by failed GOFAI scientists like Moravec and
others in order to cover up their failures and cluelessness.
Louis Savain
Why Software Is Bad and What We Can Do to Fix It:
http://www.rebelscience.org/Cosas/Reliability.htm
.





  Page 1 of 1

1

 


Related Articles
 

NEWER

pg.1612     pg.1232     pg.940     pg.716     pg.544     pg.412     pg.311     pg.234     pg.175     pg.130     pg.96     pg.70     pg.50     pg.35     pg.24     pg.16     pg.10     pg.6     pg.3     pg.1

OLDER