Abstract： To study the autonomous driving behaviour of intelligent vehicles, a visual navigationbased microintelligent vehicle system was proposed and implemented in this paper. The frame of the proposed system was introduced from the aspects of the hardware/software architecture and the data communication scheme. The performance of the traditional lane detection method is very limited in the microroad traffic environment with illumination variations. In the proposed approach, a gray scale morphology tophat transformation and brightness profile scanbased lane detection method was introduced to conquer this problem. To analyze the autonomous driving behaviour of intelligent vehicles in the smallscale traffic environment, a fuzzybased steering control algorithm was designed for simulating the behaviour of drivers. The experimental results show that the proposed scheme is able to effectively simulate the autonomous driving behaviour of vehicles in the real environment, and provides a new approach for the research of intelligent transportation system.