Belinda,
Years ago when I did a lab like this with real lego robots, I used a
stop watch and timed all the robots from several different starting
points.
Lately, I haven't been so careful about trying to compare their
performance in a quantitative way. Instead, the students typically
agree which behavior is best and are actually more interested in
talking about how the various approaches work.
Lisa
On Tue, 13 Mar 2007, belinda thom wrote:
Lisa,
Thanks for the pointer...its been useful to see how you created a multi-robot
world (much easier than what I'd been doing).
I now have a question about how you compared robots for your subsumption
robots finding lights in various flavors of room lab.
One approach would be to put students robot brains into the same room and see
which one finds the light first. But that adds a new complexity to the
problem that the students wouldn't have explicitly planned for. As an
alternative, I tried to run two pyrobot simulators at the same time, one
running each robot. But I then ran into socket address complaints. My current
thinking is to increment a counter for each update call, and to compare based
on the total number of steps each robot has.
Any other ideas?
--b
On Mar 9, 2007, at 7:58 AM, Lisa Meeden wrote:
Belinda,
I am teaching an introductory cognitive science class this semester
and came up with a fun Braitenberg world. It is based on an example
from the "Understanding Intelligence" book, where what they call
"helping behavior" emerges. It doesn't quite work the same way in
simulation, but still leads to some interesting interactions.
There are three robots all attracted to light, with light sensors on
their fronts. Two of the robots have light emitters on their backs.
Plus there is also a light in the environment.
You can see the lab here:
http://web.cs.swarthmore.edu/~meeden/cogs1/s07/lab04.html
Near the bottom of this page there are some links to the code you'd
need to run this example.
Lisa
---
Lisa Meeden Computer Science Department
Associate Professor Swarthmore College
Chair Cognitive Science Program 500 College Ave.
[EMAIL PROTECTED] Swarthmore, PA 19081
www.cs.swarthmore.edu/~meeden 610-328-8565
On Thu, 8 Mar 2007, belinda thom wrote:
Am also forwarding this one due to error in email address.
Hi,
I'm wondering if any of you have written some flashy looking Braitenberg
simulations via Pyro.
I've tried out individual Braitenberg brains in various worlds, but the
behaviour isn't so interesting when there's not several robots of
different types interacting. For a peek at collective behavior, I found:
http://people.cs.uchicago.edu/~wiseman/vehicles/
but the code is not available. The Understanding Intelligence book also
provides a simulator at:
http://www.ifi.unizh.ch/groups/ailab/people/lambri/mitbook/braitenberg/braitenberg.html
but the images look clunky and there's some glaring bugs.
Pyro has such a nice feel to it. I'd love to create some worlds of
similar quality to the ones shown in the wiseman links, but figured I'd
poll you all to see if anyone has already done this.
Also a question: is there a rule of thumb for how many robots can I put
into a world before the CPU bogs down?
Thanks,
--b
_______________________________________________
Pyro-users mailing list
[email protected]
http://emergent.brynmawr.edu/mailman/listinfo/pyro-users
---
Lisa Meeden Computer Science Department
Associate Professor Swarthmore College
Chair Cognitive Science Program 500 College Ave.
[EMAIL PROTECTED] Swarthmore, PA 19081
www.cs.swarthmore.edu/~meeden 610-328-8565
_______________________________________________
Pyro-users mailing list
[email protected]
http://emergent.brynmawr.edu/mailman/listinfo/pyro-users
