The Application of Semantic-enhanced 3D Laser SLAM in Mobile Robots

Ruizhi Feng

doi:10.54097/dyrcj170

Authors

Ruizhi Feng

DOI:

https://doi.org/10.54097/dyrcj170

Keywords:

Semantic augmented SLAM, 3D LiDAR; mobile service robot; dynamic scene.

Abstract

For indoor environments with low Global Navigation Satellite System (GNSS) and dynamic changes, such as hospitals and shopping malls, traditional 3D LiDAR Simultaneous Localization and Mapping (SLAM) is prone to mismatch, false loops, and map drift under repeated topology, occlusion, and dynamic interference. Introducing semantic priors into the front-end, back-end, and loop closure detection of SLAM can significantly improve the robustness of localization and the interpretability of the task. This article outlines the workflow of semantic augmented SLAM, starting with ordinary 3D laser SLAM. It summarizes the architecture of front-end odometry, back-end optimization, loop closure, and map, and concludes the front-end dynamic point culling and class weighting, back-end semantic consistency, loop closure, and instance anchoring. It also introduces commonly used semantic augmentation methods, such as SuMa++ and dynamic-static object recognition, as well as their roles in motion segmentation and mapping. Based on service robot scenarios such as hospital corridors, supermarket restocking, and warehouse handling, this article summarizes common evaluation and practical experience related to localization and semantics, providing a reference for the application of semantic information in practical SLAM systems.

Downloads

Download data is not yet available.

References

[1] International Federation of Robotics. Service robots see global growth boom. Press Release, 2025, Oct 7. Available: https://ifr.org/ifr-press-releases/news/service-robots-see-global-growth-boom (accessed: Oct 24, 2025).

[2] Chen X, Milioto A, Palazzolo E, Giguère P, Behley J, Stachniss C. SuMa++: Efficient LiDAR-based semantic SLAM. Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019: 4530–4537.

[3] Chen X, Li S, Mersch B, Wiesmann L, Gall J, Behley J, Stachniss C. Moving object segmentation in 3D LiDAR data: A learning-based approach exploiting sequential data. IEEE Robotics and Automation Letters, 2021, 6(4): 6529–6536.

[4] Li F, Fu C, Sun D, Li J, Wang J. SD-SLAM: A semantic SLAM approach for dynamic scenes based on LiDAR point clouds. Unpublished Manuscript, College of Mechanical and Vehicle Engineering, Chongqing University; State Key Laboratory of Intelligent Vehicle Safety Technology, Chongqing Changan Automobile Co., Ltd.

[5] Ester M, Kriegel HP, Sander J, Xu X. A density-based algorithm for discovering clusters in large spatial databases with noise. Proceedings of KDD, 1996: 226–231.

[6] Geiger A, Lenz P, Urtasun R. Are we ready for autonomous driving? The KITTI vision benchmark suite. Proceedings of CVPR, 2012.

[7] Helmberger M, Morin K, Berner B, Kumar N, Cioffi G, Scaramuzza D. The Hilti SLAM challenge dataset. IEEE Robotics and Automation Letters, 2022.

[8] Ramezani M, Wang Y, Camurri M, Wisth D, Mattamala M, Fallon M. The Newer College dataset: Handheld LiDAR inertial and vision with ground truth. Proceedings of IROS, 2020.

[9] Hornung A, Wurm KM, Bennewitz M, Stachniss C, Burgard W. OctoMap: An efficient probabilistic 3D mapping framework based on octrees. Autonomous Robots, 2013, 34(3): 189–206.

[10] Nav2 Team. Costmap 2D. Nav2 Documentation, accessed 2025. [Online]. Available: https://navigation.ros.org/configuration/packages/costmap2d

[11] Rosinol A, Abate M, Chang Y, Carlone L. Kimera: An open-source library for real-time metric-semantic localization and mapping. Proceedings of ICRA, 2020: 5791–5798.

[12] Peng H, Zhao Z, Wang L. A review of dynamic object filtering in SLAM based on 3D LiDAR. Sensors, 2024, 24(2): 645.

[13] Wang W, You X, Zhang X, Chen L, Zhang L, Liu X. LiDAR-based SLAM under semantic constraints in dynamic environments. Remote Sensing, 2021, 13(18): 3651.

[14] Jaritz M, Vu TH, de Charette R, Wirbel É, Pérez P. xMUDA: Cross-modal unsupervised domain adaptation for 3D semantic segmentation. Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 12605–12614.

[15] Chen X, Milioto A, Palazzolo E, Giguère P, Behley J, Stachniss C. SuMa++: Efficient LiDAR-based semantic SLAM. Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Macau, China, 2019: 4530–4537.

[16] Li F, Fu C, Sun D, Li J, Wang J. SD-SLAM: A semantic SLAM approach for dynamic scenes based on LiDAR point clouds. State Key Laboratory of Intelligent Vehicle Safety Technology, 2025.