Name
Abstract | ||
|---|---|---|
Simultaneous Localization and Mapping is one of the hotspots in the field of mobile intelligent robot research. Over the past decades, many excellent SLAM systems with good performance have been developed. However, many of the systems make the assumption that the environment is static. In this paper, we propose a key segmentation frame based semantic SLAM (KSF-SLAM) method to deal with autonomous navigation in dynamic environments, which can reduce computational complexity. First, a key segmentation frame selection strategy is designed, so that it is unnecessary to perform segmentation in all the image frames. When a key segmentation frame arrives, the semantic segmentation is performed by SegNet, and the dynamic key points in the frame can be stored at the same time. Moreover, an efficient semantic image generation method is proposed when dealing with non-key segmentation frames. Optical flow tracking of the dynamic key points is performed between key segmentation frame and current frame before the next key segmentation frame arrives to generate semantic images for dynamic key points removal in non-key segmentation frames. By this way, an efficient semantic tracking module is added to the SLAM system to remove dynamic objects in dynamic environments. Experiments on TUM RGB-D datasets, KITTI datasets and in real-world environments are conducted to verify the effect of the proposed method. When compared with ORB-SLAM2 and DS-SLAM, the method proposed in this paper can significantly improve the real-time performance of the SLAM system while the positioning accuracy are equivalent. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1007/s10846-022-01613-4 | Journal of Intelligent & Robotic Systems |
Keywords | DocType | Volume |
Dynamic SLAM, Semantic segmentation, Key segmentation frame, SegNet, Optical flow tracking | Journal | 105 |
Issue | ISSN | Citations |
1 | 0921-0296 | 0 |
PageRank | References | Authors |
0.34 | 12 | 6 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Yao Zhao | 1 | 1926 | 219.11 |
| Zhi Xiong | 2 | 24 | 11.65 |
| Shuailin Zhou | 3 | 0 | 0.34 |
| Zheng Peng | 4 | 0 | 0.34 |
| Pascual Campoy | 5 | 436 | 46.75 |
| Ling Zhang | 6 | 0 | 0.34 |