The author proposes a novel semantic SLAM approach, SD-SLAM, to address the challenges of dynamic scenes using LiDAR point clouds.