Yes, that's a real (sort of) word. We've started with some crisp rules, which, in the context of fuzzy logic, means clear-cut, hard-defined data, which we then fuzzified (again, a sort of real word) by assigning membership functions to sets. The last step of the process is to defuzzify the data and make a decision. For this, we use simple Boolean operations, such as the following:
IF health IS critical THEN cast healing spell
Now, at this point, you may be saying, "Hold on a second. That looks an awful lot like a binary controller," and you'd be correct. So why go through all the trouble? Remember what we said earlier about ambiguous information? Without a fuzzy controller, how does our agent understand what it means ...
