Subject:
|
Re: Design
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Sun, 4 Dec 2005 17:07:21 GMT
|
Original-From:
|
PeterBalch <PeterBalch@[nomorespam]compuserve.com>
|
Viewed:
|
1358 times
|
| |
|
|
Mr S wrote:
> We need to build a cockroach or ant type robot
> that functions on its own, that can work in your
> living room and not get stuck.
> No matter how
> you present this, it won't be a very popular contest
> for clubs...
That's a very interesting point. Why won't it be popular?
There's just been a long thread on Seattle Robots then another on PARTS
about fine-tuning the rules for walking contests. There are similar threads
here about "can I use non-lego string?".
What is it about competition? Why must it be a competitive activity? I make
robots and program computers because it's a fascinating intellectual
challenge (and, fortunately, people pay me for it from time to time).
The prizes for the competitions are paltry. As someone said recently in one
of the threads: all you're competing for is Bragging Rights. Why get side
tracked argueing about whether robots should be handicapped by leg-length?
Why should competitions be about speed? Why not "don't get stuck behind the
sofa"?
> Soon you will be able
> to build hardware that previous robot builders could
> only dream about. We'll see the ability to create
> sophisticated hardware with much longer run times than
> 30 minutes. That's when you will be able to do more,
> be more autonomous.
I think that's a dangerous argument. For the last 40 years, people have
criticised AI researchers work with "it doesn't work in the real world".
And the researchers have always answered "it's just because I'm limited by
processor power. All I need is a bigger, faster computer". Over and over
again, they've been proved wrong.
> the trouble
> is that the cutting edge is not pushing in the right
> direction
The AI research paradigm fas always been to invent a toy world, write a
program that works in it then hope that the program will scale-up to the
real world when more processing is available. Animals have never done it
that way. Since before the pre-cambrian explosion, animals with legs and
eyes have had to function in the real world. As processing power increased
they were able to do more but they never inhabited a toy world.
Scale-up the brain's complexity, but always live with the real world's
complexity. We know that route works.
dan miller wrote:
> I can think of one field where autonomous robot-like, pseudo-AI is alive • &
> kicking, and making tons of money. One word: videogames.
> You click on an object, and the character picks it up.
> These are very sophisticated actions,
Toy world!
It's very easy to simulate emotions in game characters. The hard part is
for the characters to infer intention from the player and that's something
the game programmers try to avoid.
> As for the Darpa challenge
> DARPA dumbed down the race to
> make sure there were winners, etc.
Marvin Minsky was sceptical. His response was: "[they] owe us a sound,
clear and useful explanation of what, if anything, they have learned".
I can see both sides of the argument - it was the real world but
simplified. I was impressed.
> as soon as machines achieve something, we just move the bar up
Has that applied to other technologies? Not really. With AI and robots, the
bar is way out of reach. The only people who crow about their acheivments
are AI researchers. The rest of the world has seen on TV what real robots
will be able to do Real Soon Now.
Does completing the Darpa challenge mean that NASA will soom send an
autonomous robot to Mars? No.
I'd be absolutely astounded if an autonomous martian robot lasted a couple
of days. It would mistake a pit for a shadow, fall in and get stuck. If you
sent a cokroach, I'd be equally astounded if it ever got stuck. Go back in
two years and the planet would be crawling with them. The bar is not so
much out of reach as completely out of sight.
I would love to put cheap autonomous toy robots for into the hands of
children (and adults) who might come up with the wacky ideas that will
actually work. The intellectual attack seems to be getting us nowhere;
maybe evolution will work - it has done before.
Peter
|
|
Message has 2 Replies:
 | | New contest: Was Design
|
| Peter, Everyone interested. The thread (fractured as it was) was getting a bit long to quote everyone, so I thought I'd start this new one. Yes, Peter, I agree with 99% of all that you have stated. I think the 'boys' toy model works, even if only (...) (19 years ago, 4-Dec-05, to lugnet.robotics)
| | | Re: Design
|
| (...) Hmm, yes in a sense. Why the negativity about constrained or simulated worlds? We are not Nature; we are not constrained to work in a design space that never changes. You can mean different things by calling something a 'toy world'. I would (...) (19 years ago, 4-Dec-05, to lugnet.robotics)
|
7 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
