Testing
1.	Hardware

1.1   Fire-Fighting Robot Contest Maze 
A replica of the contest maze was designed and built by 
Stan Kusmider, our mechanical expert. 
  See document:  arena.gif

1.2	Robot Base
An issue came up right after the initial build of the robot. 
The original 3 bit motion control bus between the 2 controllers 
allow only 7 different motions.  Shaun Butler, our programmer 
wanted more combinations after looking over the maze.  
We expanded the motion control bus to 4 bits (16 combinations). 

1.2	Sensors
1.2.1   IR Object Detector
The sensors didn't seem to work as well in the hallway where our 
maze replica was located, compared to how they performed in the 
lab.  It was apparent the light was brighter since the suspended 
ceiling was much lower there.  This is where the value of R27 
(the IR LED current limiting resistor) got pushed down to 180 ohms 
to really pulse them as bright as possible to overcome the ambient 
light interference.  This helped quite a bit, but we had to 
equalize the gain of each IR Receiver module using small squares 
of the material reclaimed from a discarded PIR motion detector. 
It's properties were not any different for IR light than for 
visible light, but as we determined empirically, a single slice 
glued over a sensor yielded about 3 Db of attenuation at all 
wavelengths of light.  

1.2.2   Flame Detector 
The home-rolled flame detectors worked very well.  
The values of R4, R5, and R9 were adjusted so that the 
the left and right sensors could spot a flame 2 feet away 
but the front sensor would have less gain (for zeroing in 
on the candle) and would only toggle when within 12 inches. 


1.2.3   Wheel Optical Encoders 
Pots were used to empirically determine LED dropping 
resistor and phototransistor collector resistor 
optimum values of 470 ohms and 100K, respectively.  
Fixed resistors were mounted on the Sensor Board and 
wired in to the Basic Stamp II I/O pins P2 and P3.  
At our robot maximum speed, the outputs were an active low 
pulse train with 75 mS. pulsewidth repeating every 250 mS.  
or in other words, 4 pulses per second, which was about 
what we expected based upon a speed of 5 in/sec. 
We had calculated 1.129" per pulse * 4 pulses per Sec. 
equals aproximately 5 in/ second.

2.	Software
2.1	Robot Base
The new 4 bit bus will be used to direct cardinal motions: 
Hold, Forward, Back, Right, and Left and in addition: 
Gradual Right, Gradual Left, Back Right, Back Left, 
Back Gradual Right, and Back Gradual Left.  
  See document:  newcode.bas

2.2	Sensors and Navigation 
Our programmer Shaun Butler did all of the sensor and 
maze navigation software, the bulk of which was done in 
the one week prior to the Trinity Robot Contest.  
This is not his last revision, but is close: 
  See document:  carpetm.bs2

See the next section "Experiment" to see how we did at 
Trinity.

Back to Robot Main Page