M2UD: A Multi-model, Multi-scenario, Uneven-terrain Dataset for Ground Robot with Localization and Mapping Evaluation

Yanpeng Jia, Shiyi Wang, Shiliang Shao*, Yue Wang, Fu Zhang，Ting Wang*

Project

• The dataset includes 58 sequences across 12 categories, collected from a diverse set of highly challenging environments, including urban areas, rural regions, open fields, long corridors, and mixed scenarios, etc, covering a total distance of over 50 km.
• We collect data under extreme weather conditions, including darkness, smoke, snow, sand, and dust, to support researchers in developing SLAM solutions for such extreme situations.

面向地面机器人的通用鲁棒激光SLAM技术

贾彦鹏, 王挺, 邵士亮*, 王少聪, 王世逸，曹风魁

Project

• 这是我们之前工作GR-LOAM和RGC-SLAM的进一步拓展，改进了之前的地面提取方案和特征提取方案，使得算法适用于各种型号的雷达。
• 另外，通过引入强度特征，以及几何相似性和强度相似性约束，使得算法取得更鲁棒的效果。

CAD-Mesher: A Convenient, Accurate, Dense Mesh-based Mapping Module in SLAM for Dynamic Environments

Yanpeng Jia, Fengkui Cao*, Ting Wang*, Yandong Tang, Shiliang Shao and Lianqing Liu

Project

• An accurate, consecutive, and dense meshing module adapting to dynamic environments, which is plug-and-play for conveniently integrating with various LiDAR odometry to further improve the pose estimation accuracy and filter dynamic objects.

TRLO: An Efficient LiDAR Odometry with 3D Dynamic Object Tracking and Removal

Yanpeng Jia, Ting Wang*, Xieyuanli Chen and Shiliang Shao

Project

• A dynamic LiDAR odometry that can filter dynamic objects through a 3D multi-object tracker and achieve more accurate pose estimation through bounding box posture consistency constraint.

A Compact LiDAR Odometry and Mapping with Dynamic Removal

Meifeng Zhang, Yanpeng Jia*, Shiliang Shao* and Shiyi Wang

Project

• A compact LiDAR odometry that can filter dynamic objects, extract ground points, and segmentation through range image and achieve more accurate pose estimation through ground normal and z-axis constraint.