Mobile robots operating in outdoor environments frequently encounter the issue of undesired traces left by dynamic objects and manifested as obstacles on the map, impeding the robot's ability to achieve accurate localization and navigation performance. To address this problem, we present a novel online map construction framework called RH-Map. Our framework leverages a newly proposed 3D region-wise hash map data structure for efficiently removing dynamic objects in real-time. It comprises a real-time dynamic object removal front-end module S2M-R and a lightweight back-end module for further removal. We conducted extensive experiments on the SemanticKITTI dataset, and the results demonstrate that our proposed method performs favorably compared to state-of-the-art approaches, and we further validated the proposed framework in real-world environment. The source code is released and available for the community.

翻译：在户外环境中运行的移动机器人经常遇到由动态物体留下的痕迹在地图上表现为障碍物的问题，这阻碍了机器人实现精确定位和导航性能。为解决该问题，我们提出了一种新颖的在线地图构建框架，称为RH-Map。该框架利用新提出的三维区域哈希地图数据结构，实时高效地去除动态物体。它包含实时动态物体去除前端模块S2M-R和用于进一步去除的轻量级后端模块。我们在SemanticKITTI数据集上进行了广泛实验，结果表明，与当前最先进方法相比，我们的方法表现更优，并在真实环境中进一步验证了所提框架。源代码已发布并可供社区使用。