Page 68 - Designing Autonomous Mobile Robots : Inside the Mindo f an Intellegent Machine

Thinking More Clearly Through Fuzzy Logic

insight into an issue can be cancelled out by a person who simply likes the charisma
of the opposing candidate, or who votes a straight party line.

But what if we used voting machines loaded with a thousand questions about current
issues, all of which were published ahead of time? Each voter could be presented
with ten of these questions, randomly selected from the thousand. Upon finishing
the quiz the voter would then vote that score to the candidates of his or her choice.

Such a concept is unlikely to be implemented in our society (largely because the pol-
iticians who make laws find charisma easier to feign than competence), but fuzzy
logic has that capability, and much more. The earlier example of a dual technology
motion sensor favored the passive infrared sensor because it is generally a better in-
dicator of a true alarm. Yet because the two sensors contributed proportionally to
their signal strength readings, the microwave could have more affect on the outcome
than the PIR. Thus fuzzy logic is itself a sort of democracy. The process can, however, be
taken to higher levels of processing.
In the coming chapters we will explore ways that the results of fuzzy calculations can
themselves be inputs to other layers of fuzzy logic. In the end, all of these concepts
are tools that the creative programmer can put together in endless combinations. The
power of this layering of simple concepts is hard to overestimate.

Adaptive fuzzy logic

The biggest advantage of software controls over their analog counterparts is that the
rules for signal processing can made be quite complex without adding to the system’s
cost. Since only a few parameters control the operation of a fuzzy logic “engine,” it
can be easily slaved to other controls to gain a degree of adaptivity.

Weighting trapezoids in response to other parameters

Returning to our example, as the ambient background temperature approaches the
temperature of a human body; the PIR sensor will become less sensitive because
there is no contrast between our intruder and the background. Any signal the PIR
does produce may need to be weighted more heavily to prevent our alarm system
from becoming insensitive. We could thus add a second axis to the fuzzy logic trans-
fer function that looked like Figure 4.4.

51