Fire Fighting Robot 2001

Take a look at the Competition Pictures

• Processor: 16F877 Microchip Pic running at 20Mhz
• Sensors:
  • 4x home made IR Ranging sensors for distance measurements
  • Omron reflective fiber optic sensor for the ground
  • 3 Phototransistor array for candle seeking
  • UV Sensor for candle detection
  • Vector 2x Compass Module for heading measurements (did not work in the crowded gym at Trinity)
  • Microphone and tone decoder for starting signal
• Drive System: Gearhead motors from Jameco directly coupled to wheels with home made H-Bridge driver
• Extinguishing System: High power box fan
• Power: 12v 4ah Sealed lead acid battery

Remarks
This robot was our first attempt at the Trinity College Fire Fighting Robot Challenge. Problems we had:
Drive
Our biggest problem was the drive system. The gearhead motors were very weak and difficult to control. The driver was of open loop type so the logic was guessing whether the motors were actually doing what they were supposed to but rarely did. Because the system was open loop we would try to program the robot to move consistently, but because the motors were weak sometimes they didnt move at all. The problem came with either going to fast or not moving at all. Because of this we would gradually stray into the walls even with correction from our distance sensors and stall, the program would be unaware of this and we would loose.

Because of the base configuration the robot couldn't climb ramps to prove that it was a non-dead reckoning robot hurting us with points.

Distance Measurement
Although the homemade IR Rangers were fun to build and we learned alot they were pretty unstable. Often they would randomly give extreme measurements that would confuse the program. They Also had a large error that got bigger as distance increased.

Candle Detection
In our last trial at competition the robot entered the second room, which had the candle in it, and activated its UV sensor to see if it was present. If the candle was detected the robot would begin searching for the candle and eventually extinguish it. However if not detected the robot would back out of the room and try another room. Our sensor goofed. Because of last minute changes and inadequate testing the UV Sensor bulb was mounted on its side with the least sensitive side pointed towards the candle. The Sensor didnt pick up the UV Light and we left the room.

What was good:
Extinguishing System:
The high power box fan mounted on the front of the robot worked fine. Fans are easy to control with a simple relay or transistor. They dont need to be recharged and you dont need to clean up after them.

Candle Seeking
The Last system the group made for the robot was an array of 3 phototransistors connected to the ADC port of the pic. The 3 phototransitors were mounted at slight angles from each other with the center one pointing straight forward. When needed the pic would get reading from the 3 sensors and would be able to tell if there was a bright light source (candle) straight ahead, to the left, to the right, or not present. The robot would continuously sample this sensor and follow its reading until the line around the candle was reached and would then turn on the fan.

This sensor was the easiest to build and worked the best. Construction was very quick and programming very simple. Initial attempts of locating the candle such as scanning and remembering the position of the brightest source and using the UV Sensor for seeking worked horribly. We are even thinking of making a larger version of this sensor for next year with 5 or 7 phototransistors for greater resolution.

What's next:
Next year we hope to change many things including a complete redesign of the drive systems with more powerful motors with encoder feedback and a configuration capable of climbing ramps. We are also going to use more of, and better rangers. The one controller in the robot was very overwhelmed with the amount of work required to run the sensors and the motors. Next year we will have separate sensors to control each system and communicate to the main controller via an I2C bus.