Reinforcement Self-Organizing Fuzzy Control Using Ant Colony Optimization

This paper proposes a Reinforcement Self-Organizing Fuzzy Control method using Ant Colony Optimization (RSOFCACO). Only reinforcement signals are required when using the RSOFC-ACO for fuzzy controller design. There are no fuzzy rules initially in RSOFC-ACO. An online fuzzy clustering method is used to generate fuzzy rules automatically during control process. The fuzzy clustering method flexibly partitions the input space and requires a smaller number of rules than a grid-type partition. The consequent part of each fuzzy rule is designed using Ant Colony Optimization (ACO). All candidate consequent control actions of a fuzzy rule are listed in advance. The tour of an ant is regarded as a combination of consequent actions selected from every rule. The used ACO approach aims to optimally select the consequent part from a set of candidate actions according to ant pheromone trails. The RSOFC-ACO method is applied to truck backing control problem and its performance is compared with other reinforcement fuzzy control methods to verify its efficiency and effectiveness.