|
[DPRG] Can-Can Rules
Subject: [DPRG] Can-Can Rules
From: David P. Anderson
dpa at io.isem.smu.edu
Date: Tue Sep 11 18:21:05 CDT 2001
Hi all,
I think I am the only robot builder that has ever won the Can-Can
contest by satisfying all it's requirements. As I understood them.
Here's what SR04 does.
1. The robot lights a red LED when it "identifies" a possible target.
In that mode it attempts to home in on and grasp the can. If the red
LED goes off and the green LED comes on, it means it has lost the target
or determined that the target was not a can.
2. The robot flashes a blue LED when it actually graps a target. It
flashes the LED once for the first can, twice for the second can,
three times for the third can, and so forth. In this way it indicates
both that it has RECOGNIZED and CAPTURED a can, and also that it is
COUNTING the number of cans that it has captured.
The "can count" is not actually incremented until the can is returned
to the starting area A and deposited. That way if the can is dropped,
for example, on the way back, the robot simply resumes searching, and
will not double-count that particular can when it finds it again.
3. The robot, in addition to collecting the cans, must visit areas
A, B, and C, "all the way over the line." When the sixth and final
can has been located and captured the robot checks to see if all three
areas have been visited. If not, it navigates to those areas on it's
way back to starting area A with the final can.
I believe this matches the intent of the original rules. Can collecting
was not so much the intent itself, as the development of grippers and methods
of navigation and object recognition. This leaves the judges quite a bit
of leeway to interpret the sophistication of the underlying technology.
For example, I toyed with the idea of mounting a leaf blower on a robot.
All I have to do is drive into the "C" goal and blow any cans out of the
T and into the main A-B corridor. Then drive down to the "B" goal and blow
everything back down to the "A" area. Then drive to the starting place
and declare I've collected 6 cans.
In the real world this might actually be a BETTER solution for cleaning up
empty soda cans. But cleaning up empty soda cans is not really our goal.
Developing nifty-keen robots is the real goal. Can collecting was envisioned
as a task that might help get us there.
</mount_soapbox>
Most folks who become involved in hobby robotics start off with visions
of R2D2 and C3PO, perhaps fetching a beer from the fridge or answering
the doorbell. In any case, operating and interacting in the same living
and working spaces as the local human population. Robot contests are
offered as an intermediate step to get us there.
But in practice robot contests, in my only-moderately-humble opinion,
almost always work against the development of real-world practical robots.
Robot contests tend to produce specialized clever solutions for the given
task at hand. They do no produce general purpose robots that can survive
in the real world, outside the pristine environment of the contest course
and it's artificial requirements.
They also put us in the entertainment business. I've seen this at work
both in the DPRG and with the Seattle Robotics folks. Certain events
which might be quite useful to the robot builders are scratched from the
contest because "the audience became bored." So we put ourselves in the
business of keeping a lay audience entertained. I have found that often
to be at cross-purposes with the development of functional robotics.
The nuts and bolts of science and engineering are usually boring for non-
scientists and non-engineers.
I'd like to see us concentrate less on contest rules and contests in
general, and more on general purpose robots that can successfully navigate
through the bedroom of an average 16-year old.
<dismount_soapbox\>
my $.03 worth,
dpa
More information about the DPRG mailing list
|
