Visual SLAM Algorithm based on Dynamic Feature Point Filtering and Motion Probability Propagation

Wenbo Bai; Haiyun Gan; Jun Li; Jialin Wang

doi:10.6919/ICJE.202502_11(2).0013

Visual SLAM Algorithm based on Dynamic Feature Point Filtering and Motion Probability Propagation

Wenbo Bai, Haiyun Gan, Jun Li, Jialin Wang

International Core Journal of Engineering, (2025), Vol.11, No.2, pp.108-121

Published: January 17, 2025

DOI: 10.6919/ICJE.202502_11(2).0013

PDF: Download Full Text PDF

Abstract

To effectively filter out dynamic feature points in dynamic scenes and thus improve the localization accuracy of visual SLAM algorithms, a visual SLAM algorithm based on dynamic feature point filtering and motion probability propagation was proposed. This algorithm built upon ORB-SLAM3 by adding a new detection thread. This detection thread employed the Detectron2 algorithm for object detection and instance segmentation on keyframes. Subsequently, in the tracking thread, the motion probability of feature points was updated through feature matching and matched point propagation. After propagating the motion probability for each point, all dynamic points were removed. Experimental results showed that on the TUM dynamic scene fr3/walking_xyz dataset sequence, compared with ORB-SLAM3, the proposed algorithm reduced the average total time consumption in the local mapping thread by 13.39%, reduced the average time consumption in local BA optimization by 23.52%, and reduced the root mean square error (RMSE) of absolute trajectory error by 94.7%. The proposed algorithm ensures real-time performance while effectively filtering out dynamic feature points, thus enhancing the localization accuracy and precision of the SLAM algorithm.

Keywords: Dynamic Scene; ORB-SLAM3; Keyframe; Detectron2.

APA Citation: Wenbo Bai, Haiyun Gan, Jun Li, Jialin Wang (2025). Visual SLAM Algorithm based on Dynamic Feature Point Filtering and Motion Probability Propagation. International Core Journal of Engineering, 11(2), 108-121. https://doi.org/10.6919/ICJE.202502_11(2).0013

References

Cadena C, Carlone L, Carrillo H, et al. Past, present, and future of simultaneous localization and mapping: Toward the robust-perception age[J]. IEEE Transactions on Robotics, 2016, 32(6): 1309-1332.
Taketomi T, Uchiyama H, Ikeda S. Visual SLAM algorithms: A survey from 2010 to 2016[J]. IPSJ Transactions on Computer Vision and Applications, 2017, 9. DOI: 10.1186/s41074017-0027-2.
Saputra M R U, Markham A, Trigoni N. Visual SLAM and structure from motion in dynamic environments: A survey [J]. ACM Computing Surveys, 2018, 51(2). DOI: 10.1145/3177853.
Liu Q, Duan F H, Sang Y, et al. A survey of loop-closure detection method of visual SLAM in complex environments[J]. Robot, 2019, 41(1): 112-123,136.
Merrill N, Huang G Q. CALC2.0: Combining appearance, semantic and geometric information for robust and efficient visual loop closure[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, USA: IEEE, 2019: 4554-4561.
Rünz M, Agapito L. Co-Fusion: Real-time segmentation, tracking and fusion of multiple objects[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA: IEEE, 2017: 4471-4478.
An Lifeng, Zhang Xinyu, Gao Hongbo, et al. Semantic segmentation-aided visual odometry for urban autonomous driving[J]. International Journal of Advanced Robotic Systems, 2017, 14(5): 1-11.
Yu Chao, Liu Zuxin, Iu Xinjun, et al. DS-SLAM: A semantic visual SLAM towards dynamic environments[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscata -way, USA: IEEE, 2018: 1168-1174.
Bescos B, Fácil J M, Civera J, et al. DynaSLAM: Tracking, mapping, and inpainting in dynamic scenes[J]. IEEE Robotics and Automation Letters, 2018, 3(4): 4076-4083.
Xiao L H, Wang J G, Qiu X S, et al. Dynamic-SLAM: Semantic monocular visual localization and mapping based on deep learning in dynamic environment[J]. Robotics and Autonomous Systems, 2019, 117: 1-16.
Wang Z M, Zhang Q, Li J S, et al. A computationally efficient semantic SLAM solution for dynamic scenes[J]. Remote Sensing, 2019, 11(11). DOI: 10.3390/rs11111363.
Redmon J, Farhadi A. YOLOv3: An incremental improvement [DB/OL]. (2018-04-08) [2019-06-03]. https://arxiv.org/abs/1804.02767v1.
Zhang J, Henein M, Mahony R, et al. VDO-SLAM: A visual dynamic object-aware SLAM system[DB/OL]. (2020-05-22) [2020-08-20]. https://arxiv.org/abs/2005.11052.
Chen Zhihuan, Wang Zuao, Li Xiangcheng. VSLAM Method Based on Depth Camera in Indoor Dynamic Scenes[J]. Journal of Inertial Technology, 2023, 31(04): 390-400.
Cui An, Zhang Xinying, Ma Yaohui. AGV Visual SLAM Algorithm Based on Adaptive Epipolar Constraint in Complex Environments[J]. Journal of Inertial Technology, 2024, 32(03): 234-241.
Liu Fengyu, Cheng Xianghong, Cao Yi. Indoor Dynamic SLAM Method Based on Deep Learning and Feature Point Velocity Constraint[J]. Journal of Inertial Technology, 2023, 31(05): 438-443.
Campos C, Elvira R, Rodríguez G J J, et al. Orb-slam3: An accurate open-source library for visual, visual–inertial, and multimap slam[J]. IEEE Transactions on Robotics, 2021, 37(6): 1874-1890.
Cheng Xiaoliu. Visual-Inertial SLAM: Theory and Source Code Analysis[M]. Beijing: Electronic Industry Press, 2023: 153-155.
Gao Xiang, Zhang Tao, et al. Fourteen Lectures on Visual SLAM: From Theory to Practice[M]. Beijing: Electronic Industry Press, 2019: 165-169.
Zhong F W, Wang S, Zhang Z Q, et al. Detect-SLAM: Making object detection and SLAM mutually beneficial[C]//IEEE Winter Conference on Applications of Computer Vision. Piscat-away, USA: IEEE, 2018: 1001-1010.
E. Rublee, V. Rabaud, K. Konolige, and G. Bradski. ORB: An efficient alternative to sift or surf. In Computer Vision (ICCV), 2011 IEEE international conference on, pages 2564–2571. IEEE, 2011.