A SLAM with simultaneous construction of 2D and 3D maps based on Rao-Blackwellized particle filters

This paper presents a SLAM (Simultaneous Localization and Mapping) method which builds 2D grid maps and generates the OctoMap based on Rao-Blackwellized particle filters. This work combines wheeled odometry and laser scan with particle filter algorithm to get the pose of the robot, and at the same time fuses the data of depth camera to generate OctoMap, OctoMap is an integrated open source framework based on octree, which is well known for its memory efficiency for the representation of 3D environments. The traditional 3D point cloud map cannot be applied in robot navigation. But OctoMap is a 3D occupancy grid mapping, which can be applied to 3D path planning of flying robots and other robots that are equipped with manipulators. In short, the experimental results demonstrate that the proposed methods can make a robot to synchronize building 2D and 3D maps very efficiently.