计算机工程与应用2025，Vol.61Issue(22)：55-74,20.DOI:10.3778/j.issn.1002-8331.2503-0065
动态环境中具身智能机器人三维激光SLAM研究综述
Review of 3D LiDAR SLAM Research for Embodied Intelligent Robots in Dynamic Environments
摘要
Abstract
Simultaneous localization and mapping(SLAM)is a key technology for ambient interaction and autonomous decision-making in embodied intelligent robots,and most of the current SLAM algorithms based on 3D LiDAR are based on static environments,whereas the presence of dynamic objects will lead to a reduction in the localization and mapping accuracy of LiDAR SLAM algorithms.Based on this issue,the relevant research on dynamic LiDAR SLAM algorithms by scholars at home and abroad is elaborated in detail.According to the different principles of dynamic object detection,the methods of removing dynamic objects are classified into semantic segmentation-based,ray tracing-based,visibility-based,etc.,and the main ideas and related application algorithms of each method are analyzed.The objects with different degrees of dynamics are classified,and the processing strategies corresponding to the different categories of dynamic objects in the LiDAR SLAM framework are summarized,and the difficulties and mainstream algorithms of online real-time processing,offline post-processing,lifelong SLAM strategies are introduced.The main accuracy evaluation indexes and datasets of dynamic LiDAR SLAM algorithms are also outlined.The future development trend of dynamic LiDAR SLAM algorithms is outlooked.关键词
动态环境/具身智能机器人/激光SLAM/动态物体去除Key words
dynamic environment/embodied intelligent robots/LiDAR SLAM/dynamic object removal分类
计算机与自动化引用本文复制引用
李经丹,于军琪,冯春勇,王奔,王楷文,伍勇华..动态环境中具身智能机器人三维激光SLAM研究综述[J].计算机工程与应用,2025,61(22):55-74,20.基金项目
陕西省重点研发计划(2024GX-ZDCYL-02-04) （2024GX-ZDCYL-02-04）
西安建筑科大工程技术有限公司技术研发项目(XAJD-YF24N008) （XAJD-YF24N008）
西安建筑科技大学2024年度新型城镇化青年观察计划(2024GCJH21). （2024GCJH21）
