Subject:
|
Re: Robot intelligence
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Mon, 15 Mar 1999 21:44:18 GMT
|
Original-From:
|
Mark C. Langston <SKRITCH@HOME.antispamCOM>
|
Viewed:
|
1003 times
|
| |
|
|
> I'm trying to design a robot pet (using the Cybermaster and neural brain AI
> software in VB). The problem I have encountered is that apart from the
> basics (the desire for 'food') it is difficult to find suitable targets for
> it's behaviour. How for example do I get it to explore it's world, how do
> you create a desire to mate etc, etc. Is this described in the books you're
> referring to? And are these books on the internet somewhere?
It's all a matter of framing it in the correct reward/punishment context.
E.g., if you want it to learn its environment, punish it for not running
into things, reward it for actively avoiding obstacles (which opens the
question: How're you mapping the environment, but...)
Mating: Err...unless you're going to use some IR transfer of code between
RCXen and some form of evolutionary routine or genetic algorithm on the
code, I don't see the benefit of this (hmm...there's an idea...)
Anyway, Braitenberg's "Vehicles" is good for this sort of thing, as
is the history of robots built at the MIT robotics lab by Brooks' team
and earlier autonomous agent work by Pattie Maes (cf the MIT media lab
and robotics lab web pages for both professors and their papers). Also
grab early copies of the proceedings (both paper and [more fun] video)
of the A-Life conferences, the emergent computation conferences, and
search for items like "autonomous agent" and "swarm behavior" or
"hive behavior" in various search engines. Also cf. things like
"Boids", Brooks' subsumption architecture, Tom Ray's TIERRA project,
etc.
And the two TAB books that Mario mentioned are, AFAIK, not online.
Both were good in their day (I think I bought them around 15 years or
so ago), and are good "hands-on" intros to the topic. However,
if you're serious about trying some of this, you may want to get
your feet a bit more wet before your phylogeny recapitulates 20+
years of ontogeny. :)
Oh, and I forgot Fred Martin, also at MIT. Shame on me. Go look at
everything he's done, and look up the BEAM robotics competition, etc.
All wonderful resources.
(Sorry, I'm at work right now, and most everything above can be
referenced directly from MITs web page, or quickly via search engines,
otherwise I'd supply relevant URLs.)
--Mark
--
Did you check the web site first?: http://www.crynwr.com/lego-robotics
|
|
Message has 2 Replies:
 | | Re: Robot intelligence
|
| (...) Yes, those books are surely more than 10 years old. But I've been out of robotics in the last years and those are the most recent I had at home :) Mark, many thanks for your suggestions and resources. Mario (26 years ago, 15-Mar-99, to lugnet.robotics)
| | | Re: Robot intelligence
|
| (...) I just got my mindstorms kit a week ago, and I was thinking of doing this sort of thing using a stack or some other data type in c++... Does the rcx have enough memory to make this possible? Dan -- Did you check the web site first?: (URL) (26 years ago, 16-Mar-99, to lugnet.robotics)
|
Message is in Reply To:
| | Re: Robot intelligence
|
| Hi Mario, I'm trying to design a robot pet (using the Cybermaster and neural brain AI software in VB). The problem I have encountered is that apart from the basics (the desire for 'food') it is difficult to find suitable targets for it's behaviour. (...) (26 years ago, 15-Mar-99, to lugnet.robotics)
|
5 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
