Search results

In recent decades, the field of robotic mapping has witnessed widespread research and development in light detection and ranging (LiDAR)-based simultaneous localization and mapping (SLAM) techniques. This paper aims to provide a significant reference for researchers and engineers in robotic mapping.

This paper focused on the research state of LiDAR-based SLAM for robotic mapping as well as a literature survey from the perspective of various LiDAR types and configurations.

This paper conducted a comprehensive literature review of the LiDAR-based SLAM system based on three distinct LiDAR forms and configurations. The authors concluded that multi-robot collaborative mapping and multi-source fusion SLAM systems based on 3D LiDAR with deep learning will be new trends in the future.

To the best of the authors’ knowledge, this is the first thorough survey of robotic mapping from the perspective of various LiDAR types and configurations. It can serve as a theoretical and practical guide for the advancement of academic and industrial robot mapping.

Over the decades, simultaneous localization and mapping (SLAM) techniques have been extensively researched and applied in robotic mapping. In complex environments, SLAM systems using a single sensor, such as a camera or light detection and ranging (LiDAR), often cannot meet the accuracy and map consistency requirements. This study aims to propose a tightly-coupled LiDAR-inertial SLAM system, which aims to achieve higher accuracy and map consistency for robotic mapping in complex environments.

This paper presents TC-Mapper, a tightly coupled LiDAR-inertial SLAM system based on LIO-SAM. The authors introduce the normal distribution-based loop closure detection method to the original one (i.e. the radius search-based method), which can enhance the accuracy and map consistency for robotic mapping. To further suppress map drift in complex environments, this paper incorporates a gravity factor into the original factor graph. In addition, TC-Mapper introduces incremental voxels (iVox) as the point cloud spatial data structure.

Extensive experiments in public and self-collected data sets demonstrate that TC-Mapper has high accuracy and map consistency.

TC-Mapper has two types of loop closure detections: the normal distribution-based method for correcting large drifts and the radius search-based method for fine-stitching, which can achieve higher accuracy and map consistency. The authors introduce iVox as the point cloud spatial data structure, which strives to attain a balance between precision and efficiency to the greatest extent feasible.