A Bayesian algorithm for vision based navigation of autonomous surface vehicles

This paper presents a visual perception system for an autonomous unmanned surface vehicle (USV). The vision system used is a combination of two primary sensors - a video camera, and a gimbaled planar LID AR system. Fusion between these sensors is used to identify targets and obstacles by both depth and color. A probabilistic color analysis method developed from Bayes Theorem is developed for color based object classification, even in uncertain lighting conditions.