Let's Make Robots!

Thinking about entering Trinity College Fire-Fighting Contest

     The Trinity College Fire-Fighting Contest is the closest competition (to me) I know about, has anyone else considered entering?   In about a month registration opens and I was thinking of entering this year.  (definitely will enter next year)

There home page is:  http://www.trincoll.edu/events/robot/

And the rules are here : http://www.trincoll.edu/events/robot/Rules/default.asp


The things I need to figure out are:

  • Candle detection - When I went as a spectator a few years ago the big thing was a glass encapsulated UV detector with the CMU cam 1 the up and coming contender.  I cant find anything on the UV detector and i have a CMU cam.  But I want the extra points for building it from scratch.

  • Sound start - It's a little tricky in that it has to respond to 3.8khz but not 2.5khz.  I'm not worried i have lots of analog experience.  This is where I’ll start.

  • Navigation - This is the hard bit.  The rules require the uneven floors (ramps) and variable door positions.  with extra points for returning to the start position and only searching rooms once.  But can be as simple as following the wall (without touching it) until you stumble on the candle.  I have some ideas for using a SD card and building a map as i go.  This part will get a lot of commuting time meditation.


     In the rules they worn of a harsh lighting environment.  They understate how bad it is.  the contest is in a gym with florescent and mercury vapor lights, windows and a hanger sized door that is sometimes open reviling white concrete in direct sunlight.  Idon’t think the sun was directly on a arena but I am not sure.

So that just leves me to ask

  1. Is anyone else thinking of going?
  2. Anyone have suggestions for a flame detector?
  3. Does anyone know a good Arduino friendly navigation or mapping algorithm?
  4. If i go will ignoblegnome loan me the LMR banner for my pit area ;-?

Not going to happen this year,

     Too much life got in the way,

The audio pressure detect is done but i need to tweek it to allow me to measure the frequency.

The first experiments in the filter for the flame detector went well.

I decided i am going to have to make a custom board for it.  Something like to the spider controller but built around a AT-Megga-8 and using a boost power supply so i can run it off a 3.6V battery pack.





Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
I am thinking about maybe giving it a shot too.

I am wondering if a Wii camera would detect a fire.

Yes it can. I already built a robot that finds the flame of a candle and drives to it to put it off. I just need to install the fan to blow it, then I'll show it.

so how big a fan do you think it would take ? just asking no reson

I got a ducted brushed motor fan from hobby king some time ago, I need to make a driver for it and it's good to go. But I don't know how powerful it is...

Don't forget you get extra points if you don't use a fan.  And testing your camera in a bright sunlit room will save you a lot of headaches.

Go for it dude!!!! :)


Here's a flame sensor that may or may not work. http://www.robotshop.com/ca/dfrobot-flame-sensor.html

Depending on how windy it is indoors, you may also want to put a CO2 detector on there too.


I've used A* on an Arduino before, but not sure if it will fit your needs for this contest. There's lots of ways to make it more optimal for microcontrollers!


Good luck! ^_^

That sensor looks like the standard IR detector .It will detect the flame but the real problem is the FLAMING BALL OF DEATH (aka the sun) that will invariably be directly behind the candle.  I was thinking along the lines of a mechanical filter made of stacked corrugated cardboard so it cant see anything above the walls.

A* looks like the answer for the return trip but I will still have to build the map as I go.

Thanks for the suggestion and encouragement.


For the return you might use the line maze algorithm, but follow the walls instead. The idea is to go in following the walls, check the rooms for the candle, if the candle is not there, mark it a dead end on the intersection node, so at return you drive straight instead of going back inside the room. To detect the rooms, you have a wite strip at the entrance of the room.