Particle filter based simultaneous localization and mapping using landmarks with RPLidar

Abstract

Simultaneous localization and mapping (SLAM) is one active research area in robotics. SLAM using only landmarks is an efficient method without relying on dead reckoning (DR) or inertial navigation system (INS), hence informations such as position provided by inertial devices will simply abandoned. To optimize the use of available information, one novel approach of SLAM for indoor positioning with only RPLidar, a low cost laser lidar, is proposed in this paper. First, one improved structure of SLAM using landmarks with particle matching algorithm is introduced. Second, a novel landmark selection method is presented, which takes the quality of observation into consideration too besides the angles between the landmarks. Third, the number of the landmarks needed in the triangulation approach in localization is decreased by utilizing the range information provided by the RPLidar. Experimental results show that the new approach for SLAM with only RPLidar works well, which demonstrates that the low cost low precision laser lidar can also play significant role in robotics with the aid of particle matching and landmark selection algorithms.