A Fuzzy Logic Vision and Control System Embedded with Human Knowledge for Autonomous Vehicle Navigation

A method for visually navigating an autonomous vehicle in an unstructured environment with fuzzy logic is summarized in this paper. Fuzzy logic is used with a human embedding method to process the raw vision data and control the autonomous vehicle. Specifically, the ability to classify pixels in the image space is discussed as a means of locating a safe path. Human embedding is used as a method to train the fuzzy logic system. The trained system is shown able to follow and recover to paths in the unstructured environment. Additionally, a hierarchical system that fuses the visual information with other sensors is discussed. The addition of an optical sensor provides the controller with critical information that helps make intelligent decisions in the presence of sensor imprecision and uncertainty.