A solution to estimate robot motion with large rotation by matching laser scans

Abstract

This paper presents an Iterative Closest Point (ICP) preprocessor named Iterative Closest Normal (ICN). The proposed ICN algorithm is mainly focused on the estimation of large rotation between two successive scans during the robot motion in the indoor environment. After the large rotation has been reduced, the standard ICP and point-to-line metric based ICP algorithm (PLICP) is applied to estimate the residual transformation. The experiments were conducted in a noisy environment and the robot was controlled to move with sharp turning. The results suggest that ICP and PLICP needs less iterations and becomes more robust after introducing the ICN as a preprocessor.