Virtual bubble filtering based on heading angle and velocity for unmanned surface vehicle (USV)

The present paper discusses the basic theory of an environment recognition algorithm used in unmanned surface vessels for autonomous navigation. It aims to acquire environment data as (x, y, z) with three-dimensional LIDAR independently and remove noise data occurring during data acquisition. In particular, a large number of noise sources are generated in LIDAR sensors, which are operated based on optics, due to the refraction of light and large number of bubbles. Accordingly, in the present paper, research was conducted on algorithms based on the velocity and heading angle of unmanned surface vessels to delete water waves at the rear side, a basic characteristic of unmanned surface vehicles (USVs). A virtual region was set at the rear side considering the dynamic characteristics of USVs, and then a rotation radius was calculated by applying the Ackerman-Jantoud type algorithm to the virtual region created using the velocity and heading angle, thereby determining the size of the region according to velocity. According l y, noise data could be removed depending on the behavior of USVs.