Point-to-Surfel-Distance- (PSD-) Based 6D Localization Algorithm for Rough Terrain Exploration Using Laser Scanner in GPS-Denied Scenarios

Abstract

Mobile robots 6D outdoor localization algorithms using laser scanners in GPS-denied scenarios can rely on landmarks extraction or ICP-based scan matching. Both methods have significant disadvantages in rough terrain (lack of proper candidates for landmarks or problems with time consuming ICP-based scan matching), therefore, we proposed a new method based on robot's pose correction after every single laser scanner measurement, with the use of estimated distance between a scan point and the corresponding surfel on the reference 2.5D map known to mobile robot control system. The novelty of our method is that we do not have to make the point cloud registration into a common frame and we do not need extraction of landmarks from the point cloud as the landmark-based methods. Moreover, we do not require huge computational efforts in order to compare point clouds. We present the results of simulation tests using the data captured by FARO reference scanner and real terrain experiment with the use of our innovative laser scanner.