|
[DPRG] Re: Allen Robots (Will's emotions)
Subject: [DPRG] Re: Allen Robots (Will's emotions)
From: Jeff Koenig
koenig.jeff at gmail.com
Date: Tue Mar 6 13:00:10 CST 2007
>Anybody else on this list writing behavior-based or subsumption code
at all? That's a good starting place.
If you change the word "writing" to "written", then yes, I can answer
affirmatively.
I wrote a good amount of subsumption code in Interactive C for the
Handyboard. It ran on K4 Smokey, the little robot that I let wander
in the DPRG warehouse a few years ago. The neatest thing it ever did
there was wander into a long piece of L-shaped ducting that was on the
ground. It made it to the end of the "L", turned around, and worked
its way back out. As far as I could tell, it never scraped the walls
(which is good, because it doesn't have a switched bumper). When any
of its rangefinders indicate something is close, it slows down. When
it has open field, it speeds up. When it was in the duct, it was just
inching along, since the side-mounted rangefinders reported very close
walls.
I didn't get to the point where I was successful in coding complex
tasks, although I did have behaviors for locating, approaching, and
"docking" either an object (a 12-oz can, no way to grab) or a
modulated IR beacon.
I also worked out several "wander" type behaviors that were successful
in navigating around my house, avoiding its stationary objects and
mobile residents. Since the robot didn't have encoders on the wheels,
I never did any odometry with it.
I never got to the point of setting flags, state-machine, or timer
type programming for real task completion, though. To the best of my
recollection, it was every behavior fighting for control of the
wheels.
Best Regards,
Jeff
More information about the DPRG mailing list
|
