Three-Dimensional Lidar Localization and Mapping with Loop-Closure Detection Based on Dense Depth Information

This paper presents a novel lidar SLAM system for localizing mobile robot to build map of the environment. To identify unknown transform matrix, we design new scan-matching approach, in which point cloud segmentation algorithm is additionally integrated. Different from traditional normal distribution registration, our newly proposed one incorporates ground remover and method. By employing divide space into different cells, can guarantee continuity convergence cost function. tackle recognition difficulties that camera-based loop-closure detection heavily depends on environment’s appearance, depth-completion introduced fuse sensor data ensure robustness algorithm. Moreover, bags binary words (DBoW) are adopted improve image-matching quality. Finally, experimental results presented illustrate effectiveness system.