| |
On 5/24/05, Brian Davis <brdavis@iusb.edu> wrote:
> In lugnet.events.brickfest, Suzanne Rich Green wrote:
>
> > Have RCX owners considered pooling their units to
> > run an event-long simulation?
>
> I've never thought of doing it at a mass event (OK, possibly because I've
> not
> been to any), but I've thought of doing it on my own. I suspect the biggest
> drawback is having your RCX's already built into a MOC to bring to BrickFest.
>
> > Each RCX unit is configured in exactly the
> > same way for this group project.
>
> If that's the case, the first thing is to make these 'bots as minimalist as
> possible, to encourage the maximum participation. In the same vein, make them
> modular (so folks can use their RCX in a differne MOC if the simulation isn't
> running at the time).
>
> > Example projects:
> >
> > - Watch the Game of Life, Live.
>
> That takes a *LOT* of computing units, for minimal return... although it's
> an
> interesting experiment in communication.
>
> > Proof of Emergence. units are given "termite
> > woodchip piling" rules [1].
>
> I like that one. Can it be done with a two-motor 'bot? Adding a third motor
> for an "arm" would be good, but limiting the parts to, say, a single RIS set
> might be better.
> "Woodchip elements"... cubes (2x2 up to 4x4)? Two colors give a lot more
> possibilities for interesting behaviors.
> Another possibility is to have two very different behaviors in the
> "termites", collectors & dispersers. The individual robots could switch
> between
> behaviors based on any number of things (number of collection trips taken,
> number of times an object is hit, etc.).
>
> Two other options. (1) "Ants" that run around line-following and
> line-drawing. (2) "Worms": these are based on an old Scientific American
> article
> (70's?) about simulating ocean-bottom worms. The worms live on an isometric
> grid
> (equilateral triangle graph paper). Warms crawl along the lines, and there are
> set rules for how to turn at each junction. Nothing eaten at the junction?
> Turn
> 120° to the right. One segment eaten on the left? Continue straight. The
> patterns were amazing. With real robots on a real (melamine) floor, it
> wouldn't
> be exact, or even repeatable... but it would be cool to watch.
> If you have a system of simple robots following such "hard-wired" rules,
> make
> it a genetic evolutionary system. After a set period of time, all the robots
> stop, and one (chosen at random) becomes the master, while the rest are
> numbered
> slaves. The master compares scores, and based on them the most successful rule
> sets are copied into the "offspring", and they run it again.
>
> Drat. I think I need more RCXs... "Um, honey? What's my spending limit on
> LEGO again?..." ;-)
>
> --
> Brian Davis
I like the ants idea. What you could have is a "home base" or
somewhere the RCX's can return to (some simple sensor based thing),
and then a feeding area. In between you have any number of obstacles.
The RCX's draw, or follow lines left by others, and wander in an
otherwise random path if they are not following a line. Use a
probability (random number) to determine if they are line following
(and for how long) or if they are going to go and wander.
With some kind of "scoring" based upon how long they go between feeds
and base, you could get the group to optimize a route from base to
feeding grounds. However - it would be wise to use some kind of fading
ink so you dont get a crowded mess with whats drawn - somethign that
fades after around 20-30 minutes (or less).
The base could be a stationary RCX transmitting one message. The
feeding ground(s) could be other RCX's. For even more fun- add RCX
predators - which use a message to tell the ant RCX's to decrease the
food value.
--
http://orionrobots.co.uk - Build Robots
Online Castle Building RPG -
http://www.darkthrone.com/recruit.dt?uid=V30311Q30294A30328K30226H30311D30345
|
|
Message is in Reply To:
10 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
