A Laser SLAM Method for Dynamic Environments

Simultaneous localization and mapping technology is an essential part of intelligent mobile platforms for self-exploration to realize autonomous mobility in complex scenarios. Most current laser SLAM algorithms are designed based on static environments, however, there are often many dynamic objects in the real environment that are prone to decrease the accuracy of laser SLAM algorithms. To improve the localization accuracy of intelligent mobile platforms, this paper proposes a dynamic environment-oriented laser SLAM method. Firstly, the 3D laser point cloud is accumulated to construct submap and region of interest The grid is divided according to the angle and radius, and the grid occupancy descriptor is calculated by the height distribution of the points in the grid. Then the potential dynamic region can be determined by the difference of occupancy descriptor to fit a ground plane to extract the dynamic points. Finally, the convergence of the points is determined it converges the refined points are fed into the lidar odometry to obtain the optimal position, otherwise, the number of the grid is adjusted adaptively to remove the dynamic points again. The experimental results demonstrate that the proposed method can effectively remove the dynamic points in dynamic environments, and improve the accuracy of point cloud map and the positioning accuracy of intelligent mobile platforms.