Design and Implementation of a Point Cloud Registration Algorithm for CERN Robotics Framework

This paper will explore the design and implementation of the new point cloud registration algorithm for CERN Robotics Framework (CRF) in C++. This project is initiated for the reason that, the old point cloud registration algorithm has a large codebase, and has numerous parameters to be tuned. These lead to an integration problem with the whole framework. Therefore, this implementation is designed based on ease of integration, and ease of maintenance. The proposed point cloud registration algorithm use edge features of each point cloud to construct a nonlinear least square cost function. Edges are obtained by using the eigenanalysis technique. Lastly, the implementation is compared with the Point Cloud Library's (PCL) Iterative Closest Point (ICP) algorithm.