Let's Make Robots!

SplatBot MkIII - Fire fighting Wild Thumper!!

Locates fires and puts them out.

SplatBot MKIII is my first attempt at a fire fighting robot. The "Wild Thumper" 6WD chassis is the production version of my original 6WD chassis and the new flame tracking sensor is a new version of my IR compound eye. My first attempt at an IR compound eye was tested (and failed) on my original SplatBot.

My first challenge is to see if the new eye actually works. The outputs tested ok with a multimeter but getting it to work with the Wild Thumper robot controller to opperate a pan / tilt assembly and track an object is another problem.

The design is very similar to SplatBot MkII except much larger and the bottles are mounted higher so the suspension can flex without hitting the bottles.



After a few problems with electrical noise and a faulty IC socket on the flame tracker I finally got the robot to track movement of my soldering iron. The sensors are much slower to respond than the phototransistors on the IR compound eye so the eye can loose track of an object that moves too quick but for locating a fire this is not a big issue.

I am definitely going to have to add other sensors for navigation as this chassis has the high speed 34:1 ratio gearboxes. I will also have to add a remote kill switch. My biggest worry is that it will run into the fire, catch fire itself and then run around berserk possibly creating other fires. This might defeat the purpose although it might make a good Anti-mov entry.


I have attached a video of the flame sensor tracking the motion of a 60W soldering iron.



I have attached a video of the robots second attempt at putting out the fire (the first attempt ended with the robot sitting on the fire). Overall it's working but it needs a lot of fine tuning. Right now I am using RC to guide the robot to the fire and then the autonomous system detects the fire, aims the cannon and runs the pump.

Next step is to adjust the suspension to handle the full 4Kg load of water better and improve the cannon. The software needs to improve it's aim and to check the fire after it's out for hot spots. In the second attempt there was still some flame that the robot missed.



I have added more videos including the first failed attempt to fight a fire. This was due to the fact I was using a TV remote to control the robot and the flames interferred with the signal.

The Wild Thumper is a tough little bunny and despite sitting in the flames the only damage was one motor housing got singed. The aluminium chassis absorbed enough of the heat to protect all the wiring and electronics.

In the third attempt I have fixed a problem with the tracking system and added a scanning function. You can see about 1/3rd of the way through the video of the third attempt that a small flame is initially missed until a scan is done. The flame is then detected and extinguished.

At the moment I have not had time to adjust the suspension and all the waters weight is on the back wheels. This will be corrected in the next video.




Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

I remember making a volcano as a kid for a school project. The vinegar / baking soda mixed with some food colouring made an impresive lava flow.

I'll definitely give it a try as one of the many videos I want to do with this robot. First I need to get navigation sensors fitted.

I think this is a really good idea; the pressure generated by the mixture can be used as the propellant, and you could dispense with the power-hungry and heavy pumps...

it would be possible to either mix the components well beforehand and keep them pressurised, or on startup, or on demand.


With the current robot design the two liquids would mix prior to entering the pump. I think I will need a seperate pump for each liquid and then let them mix in the barrel of the cannon. The additional pressure caused by the chemical reaction would increase the range.

Looks fantastic, Oddbot!

I await real fire for the video not only a candle :D

Weight should be 3kg and not 3 grams, I guess ;-)

Yes it will have to be a real fire to be detected at any distance. Even though a candle flame is quite hot it radiates very little heat.

The robot would have to be so close to detect a candle flame that it would more likely put the candle out by running it over. This robot is designed to put out small fires before they become big fires. Water is being used for now like the original SplatBot but I will probably fit it with a small fire extinguisher later.

Nice robot!

A fire extinguisher will be an interesting addition. CO2? nothing to clean afterwards. Or will you go for something messier like foam or powder?

What flow rate can you get from the "water cannon"?

The flame from a candle will have a decent heat radiation per unit area but not much of it will hit the sensor. For a larger flame with the same heat radiation per unit area you will get more heat flux on the sensor. With small fires it will be important to tilt the sensor when searching for a fire to get maximum exposure to the fire.


I agree that CO² is a great "no mess" solution for indoors but Gareth also has some interesting ideas about foam.

I do not know the exact flow rate of this pump. That is one test I need to perform.

As it is the sensors on the PCB are all tilted out slightly so the sensor can lock onto the position of the fire and aim directly at the fire. The software aims the cannon by trying to ballance the left/right readings and the up/down readings.

Hey very nice,great looking robot.

Oddbot, you're like the Willy Wonka of robot stuff. It's about time you start giving out golden tickets :)

Well if you call free stuff to reveiw "Golden Tickets" then there will be some coming soon. Just waiting for the manuals to be printed.