This paper presents an experimental evaluation and comparison of LiDAR SLAM algorithms for mobile robotics. We analyze the performance of four state-of-the-art methods for localization and mapping in terms of the capability in reconstructing a point cloud of the surveyed environment and of the required computational effort. More in detail, the cloud-to-cloud distance with respect to a ground-truth reference model and the density of the final point cloud are evaluated and compared. Experimental tests are conducted by performing repeated autonomous surveys on two different scenarios with a mobile robot, showing the advantages and disadvantages of the considered methods in reconstructing a 3D map.
