Not logged in, Join Here! or Log In Below:  
 
News Articles Search    
 


Submitted by Iztok Bajec, posted on April 13, 2001




Image Description, by Iztok Bajec



In the above image you can see the FireStarter 2000, which was developed by the Computer Structures and Systems Laboratory at the Faculty of Computer and Information Science, Ljubljana, Slovenia, for the Slovenian Ministry of Defense. Its main objective is open fire spread prediction, obtained through fuzzy logic evaluation, and is intended to be used by firemen on the field. The prediction is based on 800 simple, readable conditional clauses, which take into account wind direction, wind speed, fuel, slope, state of neighboring cells, and current state of the evaluated cell. All these parameters are described linguistically (little, average, very...). It is interesting that with such simple conditional clauses we get nearly 80% fit.

In the image the represented region is a 4500m x 6000m DMR with resolution 20m x 20m and 0.1m height accuracy, which makes 67500 cells. The white outline represents the actual fire final extents. The regions coloured red through yellow represent the predicted fire, depending on the fire strength in the cell, whereas the regions coloured blue through green represent the flammability of the cell. The whole process runs in real time on a PII 400, 256MB ram, with a GeForce256 DDR.

My part in the development of the program was the visualization and the fuzzy logic evaluation engine. The visualization is done in both DirectX and OpenGL. Currently there is no thorough description of the program on my web page but I am still giving it, because who knows ... keep trying:
http://lily.fri.uni-lj.si/ibajec.

Iztok BAJEC


[prev]
Image of the Day Gallery
www.flipcode.com

[next]

 
Message Center / Reader Comments: ( To Participate in the Discussion, Join the Community )
 
Archive Notice: This thread is old and no longer active. It is here for reference purposes. This thread was created on an older version of the flipcode forums, before the site closed in 2005. Please keep that in mind as you view this thread, as many of the topics and opinions may be outdated.
 
Myrmecophagavir

April 13, 2001, 04:01 PM

Eh? Picture's changed already, only just looked at the last one.

That's an interesting twist on the usual terrain pics. Any plans to use the technology to create realistic fire conditions in a game? You could have your firemen running around the terrain trying to put it out. Or does it take too long to calculate things?

 
Mournblade

April 13, 2001, 04:47 PM

They appear to be contours - if you bothered to look closely.

 
MC BAXTON *UNCENSORED*

April 13, 2001, 04:49 PM

Actually as far as I know for weather prediction usually used supercomputers.
actually, direction of wind can change quickly, and significantly change the whole action. To predict this, computer needs much more information and processing speed, than 'normal' computer.
In fact, the more information computer has, the better it can predict fire. but you are talking about PII .. (pah ! I have PIII:)

Information about cells can outdate quickly... or at that moment when nobody expects that it's already outdated.

And how to prove that your prediction program works well ? You need at least some experiments to prove it.

 
Lion V

April 13, 2001, 04:55 PM

Hmph.
Those of us in the USA might have been able to use that last year.
Ah well.
Cool.
~V'lion

 
yq713

April 13, 2001, 05:39 PM

Hey we can still use it Florida.
Stupid fires.

 
Vorteks

April 13, 2001, 06:52 PM

That's interesting. It never occured to me that a program would be used in a situation like that.

 
Mournblade

April 14, 2001, 08:35 AM

At least you can marvel at the pretty graphics while your home goes up in smoke:-)

 
The Wolf

April 14, 2001, 08:57 AM

Very nice, how are you rendering the terrain?

oh yeah... were can i get these weapons? j/k :-)

 
iztok bajec

April 16, 2001, 07:41 AM

You are right, supercomputers are usually used for such calculations. But on the other hand the simplicity of FUZZY LOGIC makes it possible to calculate the next state of the cell being currently evaluated in near realtime even on a PC. And since each cell consists of information about wind speed, wind direction, fire, fuel etc. it is possible also to integrate the wind direction and wind speeds changes of which you are talking about with no substantial drop in calculation speed. Of course faster processors (or even the use of 2,4 etc of them) makes calculations faster and thus giving you more time to do other things (visualization).

As far as for the proving - the theoretical analysis of fuzzy cellular automata was conducted in the Ph.D. thesis of a member of the Laboratory of Computer Structures and Systems.

 
sgf

April 16, 2001, 10:40 AM


Actually as far as I know for weather prediction usually used supercomputers.
actually, direction of wind can change quickly, and significantly change the whole action. To predict this, computer needs much more information and processing speed, than 'normal' computer.
In fact, the more information computer has, the better it can predict fire. but you are talking about PII .. (pah ! I have PIII:)

I never realised you were a meteorologist and fireman as well as a graphics expert. There I was, thinking that these guys might know something about the spreading of fire, what with all the research they've put in, but now you've corrected me. What they really need is one of those Cray C90s, rather than, say, just using the output of someone's weather forecast as their input, and applying a bunch of simple rules to produce reasonably accurate results cheaply and quickly.

Information about cells can outdate quickly... or at that moment when nobody expects that it's already outdated.

Once again, it's great to see you're such an expert on fuzzy-logic based fire prediction.

And how to prove that your prediction program works well ? You need at least some experiments to prove it.

You're right. Despite the guy saying he had an '80% fit' on his results, he's obviously not fitted the results against anything. I think they need to go employ some arsonists to get them some 'test data'.

Thank you for your most insightful comments. In anyone else, I'd assume they were bullshitting about a topic they knew very little about, but in your case the amount of clue you have is pretty obvious.

 
This thread contains 10 messages.
 
 
Hosting by Solid Eight Studios, maker of PhotoTangler Collage Maker.