Subject:
|
Re: Design
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 6 Dec 2005 14:03:11 GMT
|
Original-From:
|
PeterBalch <peterbalch@compuserve.[IHateSpam]com>
|
Viewed:
|
1287 times
|
| |
|
|
steve wrote:
> But there are many elegant algorithms that have come from the games
> industry (A-star navigation, Potentially Visible Sets, Adaptive social
> behaviors, etc). These are things that robotics will eventually
> benefit from once all of the low level issues are solved.
Most of what you say is very reasonable but that particular argument is one
I have difficullty with. It's a seductive idea. It was an AI attitude for
thirty years and still is in some quarters.
The idea is that you can write elegant algorithms and deduce your way to
intelligence. That a robot can look around and deduce a model of its
environment then deduce what to do next. It sounds reasonable and I've met
several "planners" who wholeheartedly believe in it. Once the people
working on the dull low-level drudge-work have clawed their way up into the
lofty realms of real AI, there will be a whole library full of wonderful
techniques ready to pull off the shelf.
The subsumption people would argue that there is only low-level stuff. Just
layer after layer of it. I believe them. We won't know who's right until
the two ends meet (which always seems to be scheduled a few decades in the
future).
I don't believe that any human or other animal ever deduced their way out
of a real-world problem. Deduction is for when you're playing chess or for
post-hoc explanations. Everything that philosophers have said about
high-level functions in the brain has turned out to be nonsense. Every new
brain-research result underlines the fact that there are no elegant
algorithms. All we ever find is another layer of low-level bodges.
Evolution produces wonderful "robot intelligence" but it never produces
elegance: only bodges.
Peter
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
