Multi-objective evolutionary design of fuzzy logic controllers for car parking problem

This paper proposes an automatic method to design a fuzzy logic controller for the automated car parking problem. To tackle the problem, design of a fuzzy logic controller is solved under the multi-objective evolutionary optimization framework, which requires three factors: an encoding scheme, design of multi-objective evaluation criteria, and design of proper evolutionary operations. Along with the parameters of antecedent membership functions, consequent parameter vectors are defined in a chromosome so that a fuzzy logic controller can modify its antecedent/consequent parameters and rule structure simultaneously during evolution. Three optimization criteria are proposed to find a set of good fuzzy logic controllers using Pareto optimality. The proposed algorithm is applied to the well-known car parking problem and produces a set of good fuzzy logic controllers that can control the motion of a vehicle for automated parking. Each fuzzy logic controller in the set has unique characteristics and can be selected according to users´ preferences, which is one of the major advantages of using the multi-objective evolutionary optimization