"The above makes me suspect a serious misunderstanding of fuzzy logic. Although fuzzy truth values are measured as numbers ranging from 0.0 to 1.0, fuzzy inference is typically performed on fuzzy sets (distributions of fuzzy truth values over some domain). The only way to "defuzzify" a single fuzzy truth value is to force that value to 0.0 or 1.0, but this term is normally used in reference to the conversion of a fuzzy set to a singleton (a single real number value- not necessarily in the 0.0 to 1.0 range).

I suggest reviewing any of the following:"

first of all, who said anything about defuzzification? i don't want to defuzzify my vars. i dont want a singleton. i want a fuzzy truth var.
i have 4 fuzzy truth vars: mNotStrategic, mStrategic, mVeryStrategic and mUnbeatable

i want to turn it into: mStrategic. where 0.0 is not strategic and 1.0 is un-beatable. so one fuzzy truth var.
so i can later on ask:
fuzzyIF( AND(mStrategic, mReachable))
{
goThere();
}


to alexjc:
"Keep it simple, get it working.

Write two separate functions, one for high shields and the other for low shields first. Get a feel for the problem. If you're stuck then you've started thinking too complex.

You can also use a lookup table to create a function and interpolate between values, that's one way to implement a response curve."

What do you mean keep it simple get it working? if one function doesn't work to get my job done, then how can to do what i want to achieve?
What do you mean write two seperate functions one for low shields, one for high shields? That would make this a Boolean situation. that would mean that when there are for example 6 shields and then a new one is added, suddenly the strategic value is completely different. i am using fuzzy logic because i need gradual change.

"Get a feel for the problem." Thanks for the info.
"If you're stuck then you've started thinking too complex." So i should give up? No. I'm not thinking too complex. i need a certain behaviour for my AI, and i need a system that get's that done. The only system i can think of is fuzzy logic.

I'm asking you, how can i do it without fuzzy logic, and the only answer i keep getting is "do it without fuzzy logic."
Please, i dont want to waste your time, so if you don't know it, please don't answer.

Quote: Original post by omniscient
I'm asking you, how can i do it without fuzzy logic, and the only answer i keep getting is "do it without fuzzy logic."

I gave you an answer, didn't you see my post? If you just want two options (do an action, or don't do it) then do something like:

if (very(or(mUnbeatable, mVeryStrategic))) {  // do action} else {  // don't do action}

If you happen to have a copy of "AI Game Programming Wisdom 4" hanging around, this is actually a good place to use my Multi-Axial, Dynamic Threshold, Fuzzy Decision Algorithm that I wrote about in there.

Otherwise, I can't contribute any more to this thread. Not without pasting about 60 pages from my book.

Quote: Original post by omniscient
What do you mean keep it simple get it working? if one function doesn't work to get my job done, then how can to do what i want to achieve?

[...]

So i should give up? No. I'm not thinking too complex. i need a certain behaviour for my AI, and i need a system that get's that done. The only system i can think of is fuzzy logic.

Listen, dumber people than you have solved this problem in a quarter of the time it took you to write all this in this thread -- and shipped their solutions in AAA games. :)


I'm telling you to keep it simple because you missed about 4 different solutions that people presented in *this* thread alone. Also included in this thread are continuous / fuzzy solutions to the problem (at least 2). So obviously the problem here is your approach to finding the solution. Keeping it simple will allow you to build up your problem solving skills and practical experience with fuzzy logic.


I'll repeat one solution just in case you didn't read: response curves, build them as a hand-crafted function using a lookup table, and create the custom non-linear function you need.

Alex

Quote: Original post by alexjc
I'll repeat one solution just in case you didn't read: response curves, build them as a hand-crafted function using a lookup table, and create the custom non-linear function you need.

Alex

Yep. That would nail it right there. I live and breathe on response curves with far more complicated decisions than this - and the simple ones as well.

ok. thanks. I still don't understand my initial problem, but i guess fuzzylogic is too complicated for what it returns.

I don't understand how I have to use response curves to do this, but i just read about it in game ai wisdom 1. I'm not gonna use it for this particular game, because i'm basically done with the ai. I guess I'll not use fuzzy logic for this kind of thing in the future though.
I think you guys are right that i completely misunderstand the point of fuzzy logic, for what it should be used.

What i was really looking for when i first read about fuzzy logic is a way to globally categorize vars, so that i can compare distance to health etc, in a readable way. i.e. the vars all range from 0 to 1, instead of distance ranging from 0 to 5000, and having an exponentially valued curve, ( meaning the value of sqrt(5000) is 0.5 and the value of 5000 is 1 for example), whereas health has a linnear value. I wanted to be able to readably compare those. I guess fuzy logic isn't the answer?

Anyway. I'll keep reading about response curves, and other techniques.

Or, if you want to use your initial idea of categorization, you could use the "binning" procedure I outlined earlier. Rather than a smooth transition, you would have crisp transitions between the levels of strategy, for example, but it would allow you to create that matrix of decision information that you desire.