Road recognition for navigation of an autonomous vehicle by fuzzy reasoning

Abstract

An autonomous vehicle that operates in out-door environments requires fast image processing, noise protection, and robustness to environmental changes. This paper presents a method for its vision navigation based on fuzzy reasoning. Some researchers use fuzzy inference to detect edges by classifying the pixel in an image into a border region or a uniform region according to the luminance differences between this pixel and its neighboring ones. On the basis of the above principle, this paper presents a method which can recognize white lines on the street or edges on the road by integrating some special knowledge into a fuzzy rule base. The method was implemented on the THMR-III mobile robot and some experiments were performed to demonstrate its effectiveness and robustness.