Optimize
1.	Hardware
1.1.1   Robot Base
Square bases get hung up on corners.  Round bases are better - 
they slide off anything they accidentally bump into.  

1.1.2   Drive System  
The modified servo drive system is cheap, but has limited 
resolution and limited repeatability for controlled motion.  
Next time I would use stepper motors or precision DC motors 
with integral shaft encoders and an H-bridge. 

1.1.3   Wheels 
Fat or thick wheels have more drag than skinny or thin wheels. 
Next time I'd use thinner wheels for less drag in the turns. 
Casters - the rear caster we used was terrible: it would lock 
unpredictably and throw our robot off course while recovering 
from turns or backing up.  I poked fun at Parallax's Pingpong 
ball on a wire rear caster, but I now have second thoughts: 
it's superior because it doesn't turn or rotate - it has 
constant friction, but the keyword here is CONSTANT. 

1.1.4   Microcontroller 
"Willitrun" used two contollers - a Basic Stamp I just to run 
the servos and a Basic Stamp II to handle all the sensors 
plus navigation.  Problem was: no Interrupt pin on the Stamps.  
All the sensors had to be polled in a loop, a very inefficient 
approach.  As it turned out, the Stamp II wasn't fast enough 
to poll the wheel encoders continuously, only for small maze 
segments. Next time I'd replace the Stamp II with a 16F84 or 
similar processor with a hardware Interrupt, and design an 
Interrupt priority encoder to handle all the sensors.  

1.2.1   IR Object Proximity Sensors 
The day before the contest we moved our left and right object 
sensors to a point just forward of the front of the robot.  
This is so they would "see" any openings to rooms in advance 
to allow more time to initiate turning into them.   
That was OK but sometimes resulted in the robot making the 
turn too quickly and getting hung up on the opening's edge. 
Next time I might mount 2 sets of sensors on each side for both 
an "opening advance warning" and a "center of opening" detection. 
   
1.2.2   Flame Detectors 
The home-rolled flame detectors worked so well, I'd use them 
again.  Perhaps some added circuitry for programmable stepping 
of their gain.  

1.2.3   Electronic Compass  
Our compass got whacked the day before the contest, so it 
never got used.  Next time I'll order a spare. 

1.2.4   Fan 
Next year I'd like to use a balloon.  It won't blow until you're 
right at the candle!


This project was a joint effort by Ward College of Technology 
students Richard Acosta, Clifton Belcher, Fortenary Boyce, 
Shaun Butler, Jamison Cormier, Ron Klimas, Stanley Kusmider, 
and Brian Rekrut under the direction of Prof. Michael Horn.  
It was our first experience with robotics, and at robot 
design competition.  The entire project was completed from 
concepts to working robot in 8 weeks.  The Trinity Fire-Fighting 
Robot Contest was an exhausting challenge for such short notice. 
Next time I'd want to start a year ahead of time!  
Because of the short time, this is the best I could do with 
documenting the project for now, but I will update this 
webpage with additional drawings and details as time permits. 
-regards, and best of luck in robotics, Ron Klimas 
-email: klimas@mail.hartford.edu

Back to Robot Main Page