Abstract
In this paper, we propose a fuzzy decision making system for a car racing game. We use two approaches to generate training data for constructing the fuzzy system and two types of sensor informations as input data to the system. In computational experiments, we compare the performance of four types of fuzzy systems.