DPRG
DPRG List  



[DPRG] New research maybe has some ideas for us

Subject: [DPRG] New research maybe has some ideas for us
From: David Anderson davida at smu.edu
Date: Thu Apr 19 21:04:33 CDT 2012

Howdy,

I like the idea of penalizing longer connections, interesting that the 
brain does that already.  Something that Brad wrote got me thinking 
about some recent experiments we did at the DPRG:

"Maybe the local one uses geometry only to determine current heading, 
but gets stuck in canyons  etc."

When Ed Paradis was here over Christmas (we miss you Ed!) we built an 
impromptu "waypoint singularity trap"  i.e., canyon, to test out 
LegoBot's ability to handle such situations.   It does so by switching 
from navigation mode to perimeter following mode when it finds that the 
target waypoint it is seeking to acquire is directly behind the robot, 
and switching back to navigation mode when the target is directly in 
front of it.  This seems to work pretty well.  Here's the video we made:


<http://geology.heroy.smu.edu/dpa-www/robo/lego/20111220_robo2.mpg>


I did a longer version with a more verbose (too verbose?) explanation at 
home in my shop:


<http://geology.heroy.smu.edu/dpa-www/robo/lego/20111220_robo.mpg>


jBot has a similar behavior that is documented on it's web page, but 
it's not quite so obvious what's going on.   In particular, no new 
waypoints are added (or needed) for the robot to handle these 
situations.  I alluded to this in the DPRG talk I did a few months ago, 
but hopefully these videos make it more understandable.


happy roboting,
dpa




More information about the DPRG mailing list