Optimization of the Fuzzy Logic Controller for an Autonomous UAV

In this paper, we describe the optimization of membership functions in an application employing a hierarchical Fuzzy Logic Controller. The size of the rule base is made manageable by using a unique formulation, known as Combs method, to help control the problem of ‘exponential rule expansion’. The optimization is performed using a steady state genetic algorithm with a dynamic fitness function. The controller being developed is designed to fly a small, autonomous parafoil, suitable for short-range reconnaissance and land survey applications. The optimization process is performed in the Matlab/Simulink software environment and incorporates fuzzy logic modules developed in the Matlab Fuzzy Logic Toolbox. Hardware limitations in terms of memory, computational speed and cost were critical factors driving the need for this simple yet robust control algorithm.