Optics and Precision Engineering, Volume. 33, Issue 8, 1314(2025)
Object-level dynamic simultaneous localization and mapping method fusing object detection and optical flow
Figures & Tables(12)
Table 1. Comparison of ATE on the Bonn Dataset
View tableView in ArticleTable 1. Comparison of ATE on the Bonn Dataset
Sequence O3［ 18 ］OA ［ 29 ］OS［ 22 ］IF［ 24 ］RS ［ 25 ］Ours 2021 2022 2023 2024 2024 / TOP-1 0 0 0 0 12 13 Balloon 0.042 0.254 0.036 0.028 0.025 0.031 Balloon2 0.072 0.219 0.035 0.029 0.024 0.031 Balloon Tracking 0.038 0.151 0.039 0.041 0.030 0.037 Balloon Tracking2 0.075 0.262 0.078 0.059 0.029 0.073 Crowd 0.386 0.144 0.020 0.027 0.017 0.016 Crowd2 0.704 0.080 0.024 0.028 0.023 0.021 Crowd3 0.339 0.156 0.026 0.038 0.031 0.023 Kidnapping 0.030 0.384 0.029 0.029 0.025 0.028 Kidnapping2 0.026 0.348 0.025 0.025 0.024 0.023 Moving No Box 0.469 0.288 0.024 0.025 0.020 0.025 Moving No Box2 0.041 0.367 0.033 0.037 0.033 0.041 Moving O Box 0.364 0.436 0.220 0.262 0.266 0.058 Moving O Box2 1.063 0.391 0.352 0.134 0.269 0.077 Person Tracking 0.422 0.519 0.051 0.041 0.037 0.044 Person Tracking2 0.834 0.636 0.055 0.056 0.037 0.061 Placing No Box 0.773 0.148 0.022 0.021 0.016 0.019 Placing No Box2 0.023 0.104 0.021 0.022 0.018 0.018 Placing No Box3 0.040 0.187 0.027 0.043 0.025 0.024 Placing O Box 0.320 0.278 0.246 0.177 0.248 0.051 Removing No Box 0.023 0.135 0.022 0.017 0.015 0.021 Removing No Box2 0.023 0.132 0.022 0.025 0.020 0.020 Removing O Box 0.384 0.147 0.357 0.197 0.288 0.057 Synchronous 1.072 0.084 0.012 0.012 0.010 0.006 Synchronous2 1.250 0.066 0.007 0.009 0.006 0.006
Table 2. Comparison of ATE on the TUM Dataset
View tableView in ArticleTable 2. Comparison of ATE on the TUM Dataset
Sequence DS［ 16 ］O3 ［ 18 ］BS［ 20 ］OS［ 22 ］RS ［ 25 ］Ours 2018 2021 2022 2023 2024 / TOP-1 0 0 0 0 3 8 fr3/s/static / 0.007 / 0.008 0.006 0.006 fr3/s/xyz 0.013 0.011 0.015 0.013 0.009 0.008 fr3/s/half 0.019 0.019 0.016 0.023 0.015 0.015 fr3/s/rpy / 0.026 / 0.041 0.019 0.019 fr3/w/static 0.007 0.014 0.010 0.006 0.006 0.005 fr3/w/xyz 0.016 0.115 0.015 0.017 0.015 0.013 fr3/w/half 0.030 0.158 0.026 0.025 0.023 0.019 fr3/w/rpy 0.035 0.535 0.036 0.033 0.033 0.029
Table 3. Comparison of RPE on the TUM Dataset
View tableView in ArticleTable 3. Comparison of RPE on the TUM Dataset
Sequence DS［ 16 ］O3 ［ 18 ］BS［ 20 ］OS［ 22 ］RS ［ 25 ］Ours 2018 2021 2022 2023 2024 / TOP-1 0 0 0 0 0 8 fr3/s/static / 0.008 / / 0.008 0.006 fr3/s/xyz 0.014 0.013 0.014 0.012 0.012 0.008 fr3/s/half 0.024 0.023 0.017 0.021 0.019 0.015 fr3/s/rpy / 0.032 / 0.062 0.024 0.022 fr3/w/static 0.009 0.020 0.013 0.008 0.008 0.005 fr3/w/xyz 0.022 0.070 0.020 0.018 0.019 0.013 fr3/w/half 0.028 0.092 0.025 0.024 0.026 0.019 fr3/w/rpy 0.045 0.206 0.047 0.045 0.042 0.040
Table 4. Precision improvement-ATE and RPE
View tableView in ArticleTable 4. Precision improvement-ATE and RPE
Method ATE RPE RSO-SLAM［ 26 ］1.662 0.158 Ours 0.925 0.128 Accuracy Enhancement 44.3% 19.0%
Table 5. Time comparison between proposed method and dynamic SLAM
View tableView in ArticleTable 5. Time comparison between proposed method and dynamic SLAM
Method Experimental
Platform
Average Duration Dyna-SLAM［ 17 ］CPU+TeslaM40 >350 OVD-SLAM［ 23 ］i7+RTX2060 27.8 RSO-SLAM［ 26 ］i7+RTX3070 45.6 Ours R7+RTX3060 26.4
Tools
Get Citation
Copy Citation Text
Shuguang LI, Qinmei CHEN, Jinlong SHI, Suqin BAI, Chenggen WANG, Xin ZHUO. Object-level dynamic simultaneous localization and mapping method fusing object detection and optical flow[J]. Optics and Precision Engineering, 2025, 33(8): 1314
Paper Information
Category:
Received: Nov. 11, 2024
Accepted: --
Published Online: Jul. 1, 2025
The Author Email: Jinlong SHI (shi_jinlong@163. com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20