Registration of Lines in 2D LIDAR Scans via Functions of Angles

Registration of LIDAR scans is an important problem in many applications that need to align two LIDAR scans optimally. This paper presents a novel algorithm for registration of LIDAR scans. The algorithm combines two already known but different concepts. The first one is the registration of lines and the second one is the registration in polar coordinates. The main idea is a new representation of the surrounding area that makes the mentioned combination possible. The algorithm works with lines, specifically with dynamic line segments in polar coordinates that are expressed as functions of angles. This representation significantly decreases the computation time and enhances the resulting alignment. The presented algorithm was designed for indoor environments that are rich in objects that can be represented by line segments. The algorithm has been tested with many indoor LIDAR scans and is able to align them with high accuracy and in real-time.