|
[DPRG] DPRG ORG 08 Photos
Subject: [DPRG] DPRG ORG 08 Photos
From: Randy M. Dumse
rmd at newmicros.com
Date: Wed May 14 03:39:34 CDT 2008
dpa said: Tuesday, May 13, 2008 4:37 PM
> I think it might be really useful to all involved if
> each one can document what worked and what didn't, and
> what changes you have made or need to make to improve
> your robot's performance.
Our university course was only a 1 hour lab course, and starting
from scratch to competition RoboMagellan robot learning the
motion control concepts along the way was quite a lot to ask of
the students. I spent far more time on PID than perhaps I could
have, but I cautioned Shawn, even if he didn't compete well, I
wanted him to take away from the course the experience of doing
a careful PID tune. And so he did.
We also had the challenge of modern political correctness.
Students are now so protected from their own mistakes, we could
not gain access to a machine shop.
Here's something we learned first hand as a consequence:
Hot glue on chassis on a hot day on a hot motor is just asking
for trouble.
Continuing, here are other things we learned:
Omniwheels, even large ones, don't do that well outdoors.
Odometry on turns on strange surfaces with big knobby tires
never gives the same wheel base number twice. (We experienced
numbers between 15300 to 19400 ticks for a 90 degree turn.)
Setting an expected speed in profiling for a full battery, and
then having to run a weak battery, can cause really wild and
unexpected losses of control. Fortunately, they aren't wild and
fast unexpected results. Just wild. (Battery was too low for
anything wild.)
The differentials which come with the Emaxx trucks are really
sloppy. Sloppy drivetrains lead to great uncertainty in the
direction the body is actually pointing when first placed on the
course.
The gear down on the differentials and transmissions of the
Emaxx trucks are too low to work with 6:1 motors, and too high
to work with 161:1 motors.
We would have done better with two differential motors driving
directly to the wheels.
We need bigger motors, or direct drive.
We need to use the right charger for our batteries, or they will
be very low capacity when we most need them.
It is very hard to see a laptop in bright daylight.
It is very hard to build tools for debugging when the laptop
display is hard to see.
I think we came very close. If we'd overcome our battery problem
we could have entered events #1 and #2. If we'd had another day
before the contest to work, we could have had the sonars
integrated and done event #3. Another week of working on it
would have made it quite good, probably having our IMU on line.
Shoulda, coulda, woulda. We're just glad we were able to find
the problem with the battery not keeping up with the profiling
so we could get a run at event #1. Great contest, and
congratulations to all our fellow competitors, most of whom did
better than we did. Watch out for us next time. We might be
getting the hang of this. (And Shawn is going to graduate
school, so he will likely be back.)
Randy
More information about the DPRG mailing list
|
