Simple Implementation of Fuzzy Controller for Low Cost Microcontroller

Fuzzy logic inference system requires a long enough computational process, starting from fuzzification, matching rule base, drawing conclusions and defuzzification. By using C programming, fuzzy logic systems can be carried out by a low capability microcontrollers, for example PIC16F877A that has a flash memory capacity of 8KB and the maximum clock speed is 20 MHz. But if it will be applied to a microcontroller that has a smaller size of flash memory, there is a possibility that it cannot be realized, especially if it uses the gaussian membership function and each membership function inputs was devided in to more than three memberships. To overcome this problem we need a method to accelerate the fuzzy logic computation process. One of that method is using a linear interpolation from the table of input-output fuzzy system. In this paper, the linear interpolation was illustrated by using bilinear fuzzy interpolation. Based on the GNU Octave simulation, it was obtained data that the time being needed to complete the simulation for fuzzy bilinear Look Up Table has 7 times faster than the fuzzy system computation, whereas from the real implementation, it was obtained that fuzzy bilinear Look Up Table gives 2 times faster and can save program memory space around 25% compared to ordinary fuzzy system.