ABSTRACT
Most of visual synchronous localization and mapping (VSLAM) algorithms are designed based on static scenes and the influence of moving objects in the scene can not be ignored. Due to the presence of moving objects in real scenes, the feature points of the visual odometer is usually easy to be mismatched, which affects the positioning and mapping accuracy of the SLAM system, resulting in the low robustness in practical applications. AVSLAM algorithm based on the ORB-SLAM2 based on deep learning is proposed in this paper, whichoutperforms on identifying and removing dynamic feature points in the tracking procedure of the SLAM system operated in dynamic environments. The local optical flow method is combined to determine the dynamic objects in the environment. In addition, an improved model based on DeepLabv3+ is introduced, which can simultaneously perform object detection and semantic segmentation for accurately removing dynamic feature points. Finally, the accuracy and robustness of the semantic segmentation model and improved algorithms are evaluated after the Cityscapes and TUM datasets are adopted. The experimental results on the TUM datasets show that in high dynamic scenarios, proposed algorithm improves the accuracy of trajectory estimation by 60.69% and 84.91%, respectively compared to ORB-SLAM2 method. The groups of experiments show that the SLAM algorithm based on the improved DeepLabv3+Semantic network model significantly improves the robustness and accuracy of the VSLAM system in the dynamic environment.
- Davison, Andrew J., “ MonoSLAM: Real-time single camera SLAM.” IEEE transactions on pattern analysis and machine intelligence 29.6 (2007): 1052-1067.Google Scholar
- Mur-Artal, Raúl, J. M. M. Montiel, and J. D. Tardós.“ ORB-SLAM: A Versatile and Accurate Monocular SLAM System.” IEEE Transactions on Robotics 31.5(2015):1147-1163.Google ScholarDigital Library
- Mur-Artal, Raul, and Juan D. Tardós. “ Orb-slam2: An open-source slam system for monocular, stereo, and rgb-d cameras.” IEEE Transactions on Robotics 33.5 (2017): 1255-1262.Google ScholarDigital Library
- Chen L C, Papandreou G, Kokkinos I, “ Semantic image segmentation with deep convolutional nets and fully connected CRFs [EB/OL] .” (2016-06-07) [2020-10-15] . https://arxiv.org/abs/1412.7062.Google Scholar
- Chen L C, Papandreou G, Kokkinos I, “ DeepLab semantic image segmentation with deep convolutional nets atrous convolution and fully connected CRFs [J]. ” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40 (4): 834-848.Google ScholarCross Ref
- Chen L C, Papandreou G, Schroff F, “ Rethinking atrous convolution for semantic image segmentation [EB/OL] .” (2017-12-05) [2020-10-15]. https://arxiv.org/abs/1706.05587.Google Scholar
- Chen L C, Zhu Y K, Papandreou G, Schroff F, “ Encoder-decoder with atrous separable convolution for semantic image segmentation[M] ” // Ferrari V, Hebert M, Sminchisescu C, Computer science. Cham : Springer, 2018, 11211 : 833-851.Google Scholar
- Wu C,Sun J,Wang J, “ Encoding-decoding Network With Pyramid Self-attention Module for Retinal Vessel Segmentation[J].” International Journal of Automation and Computing, 2021, 18(06): 973-980.Google ScholarDigital Library
- Le Cun Y, Bengio Y. “Convolutional networks for images, speech, and time series [J] . ” The Handbook of Brain Theory and Neural Networks, 1995, 3361 (10) : 255-258.Google Scholar
- Yu F, Koltun V. “ Multi-scale context aggregation by dilated convoluyions [EB/OL] . ” (2016-04-30) [2020-10-15] . https://arxiv.org/abs/1511.07122.Google Scholar
- Cheng X Y, Zhao L Z, Hu Q, “ Real-time semantic segmentation based on dilated convolution smoothing and lightweight up-sampling [J] . ” Laser & Optoelectronics Progress, 2020, 57 (2) : 021017.Google ScholarCross Ref
- He K M, Zhang X Y, Ren S Q, “ Deep residual learning for image recognition [C]” // 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 27-30, 2016, Las Vegas, NV, USA. New York: IEEE Press, 2016 : 770-778.Google ScholarCross Ref
- Lin M, Chen Q, Yan S. “ Network in network [EB / OL]. ” (2014-03-04) [2020-10-15] . https://arxiv.org/abs/1312.4400.Google Scholar
- Lin T Y, Dollar P, Girshick R, “ Feature pyramid networks for object detection [C] ” // 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , July 21-26, 2017, Honolulu, HI, USA. New York : IEEE Press, 2017: 936-944.Google ScholarCross Ref
- Cai Y, Huang X G, Zhang Z A, “ Real-time semantic segmentation algorithm based on feature fusion technology [J] .” Laser & Optoelectronics Progress, 2020, 57 (2) : 021011.Atul Adya, Paramvir Bahl, Jitendra Padhye, Alec Wolman, and Lidong Zhou. 2004. A multi-radio unification protocol for IEEE 802.11 wireless networks. In Proceedings of the IEEE 1st International Conference on Broadnets Networks (BroadNets’04) . IEEE, Los Alamitos, CA, 210–217. https://doi.org/10.1109/BROADNETS.2004.8Google ScholarDigital Library
Index Terms
- SLAM system based on improved DeepLabv3+ semantic network model
Recommendations
Dynamic Scene Vision SLAM Based on Target Detection in RGB-D Images
ICCPR '23: Proceedings of the 2023 12th International Conference on Computing and Pattern RecognitionVisual SLAM is easily interfered by movable objects in dynamic scenes, which reduces the localization accuracy and robustness due to existence of inaccurate key points on movable objects. To address this problem, this paper proposes a visual SLAM ...
Monocular SLAM System in Dynamic Scenes Based on Semantic Segmentation
Image and GraphicsAbstractThe traditional feature-based visual SLAM algorithm is based on the static environment assumption when recovering scene information and camera motion. The dynamic objects in the scene will affect the positioning accuracy. In this paper, we propose ...
Visual Odometer Algorithm Based on Dynamic Region Culling
Artificial IntelligenceAbstractThere are many moving objects in the dynamic scenes. Due to excessive changes in feature points on dynamic objects, large positioning errors will be caused, which will have a great impact on mapping. Existing algorithms have well operating ...
Comments