A study on constructing fuzzy systems for high-level decision making in a car racing game

In this paper, we examine the performance of fuzzy rule-based systems in a car racing domain. Fuzzy rule-based systems are used for high-level decision making of a car agent. We examine two methods that generate a set of training patterns for constructing fuzzy rule-based systems. We also examine the effect of sensory information on the high-level decision making. The performance of four types of fuzzy rule-based systems is compared in a series of computational experiments. The analysis of using different types of sensory information and different methods for generating training patterns is also performed.