Laser & Optoelectronics Progress, Volume. 61, Issue 18, 1812002(2024)
Multimodal Fusion Odometer Based on Deep Learning and Kalman Filter
Figures & Tables(11)
Fig. 7. Comparison of effects of each module of the odometer. (a) Trajectory diagram of the odometer on sequence 09; (b) trajectory diagram of the odometer on sequence 07
Fig. 8. Trajectory comparison and point cloud map. (a) Comparison of estimated trajectories by algorithms; (b) point cloud map
Fig. 9. Comparison chart of trajectories predicted by algorithms in the x, y, and z axes
Table 1. Comparison of prediction errors of different modules
View tableTable 1. Comparison of prediction errors of different modules
Sequence CP CP+IMU CP+IMU+ESKF trel /% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ Mean 0.80 0.47 0.68 0.42 0.66 0.41 00 0.72 0.40 0.76 0.45 0.75 0.43 01 0.92 0.30 0.34 0.21 0.33 0.20 02 0.96 0.50 0.82 0.34 0.80 0.34 03 0.70 1.29 0.54 1.01 0.53 0.99 04 0.42 0.29 0.33 0.35 0.30 0.35 05 0.68 0.35 0.67 0.44 0.66 0.43 06 0.54 0.26 0.53 0.23 0.51 0.22 07 0.68 0.42 0.50 0.37 0.47 0.35 08 1.26 0.54 1.19 0.54 1.18 0.52 09 0.83 0.32 0.85 0.31 0.83 0.31 10 1.16 0.51 0.96 0.41 0.94 0.40
Table 2. Comparison of prediction errors of different odometers
View tableTable 2. Comparison of prediction errors of different odometers
Sequence LO-Net PWCLO-Net MVLO DEMO DVL-SLAM LMVO Proposed trel /% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ trel/% rrel /［（º）/（100 m）］ trel /% rrel /［（º）/（100 m）］ trel/% rrel/［（º）/（100 m）］ Mean 1.33 0.69 1.41 0.67 3.50 1.44 1.16 – 0.98 – 0.94 – 0.66 0.41 00 1.47 0.72 3.87 0.65 2.77 1.78 1.05 – 0.93 – 0.99 – 0.75 0.43 01 1.36 0.47 0.72 0.37 3.76 0.80 1.87 – 1.47 – 1.87 – 0.33 0.20 02 1.52 0.71 1.26 0.62 4.82 2.26 0.93 – 1.11 – 1.38 – 0.80 0.34 03 1.03 0.66 1.09 0.84 2.75 1.39 0.99 – 0.92 – 0.65 – 0.53 0.99 04 0.51 0.65 0.61 0.35 1.81 1.48 1.23 – 0.67 – 0.42 – 0.30 0.35 05 1.04 0.69 1.03 0.50 3.81 1.43 1.04 – 0.82 – 0.72 – 0.66 0.43 06 0.71 0.50 1.20 0.50 4.03 1.24 0.96 – 0.92 – 0.61 – 0.51 0.22 07 1.70 0.89 0.94 0.75 3.61 1.41 1.16 – 1.26 – 0.56 – 0.47 0.35 08 2.12 0.77 1.72 0.77 2.75 1.61 1.24 – 1.32 – 1.27 – 1.18 0.52 09 1.37 0.58 0.83 0.35 3.76 1.92 1.17 – 0.66 – 1.06 – 0.83 0.31 10 1.80 0.93 2.21 1.70 4.65 0.51 1.14 – 0.70 – 0.83 – 0.94 0.40
Tools
Get Citation
Copy Citation Text
Long Li, Yi An, Lirong Xie, Zhuo Sun, Hongxiang Dong. Multimodal Fusion Odometer Based on Deep Learning and Kalman Filter[J]. Laser & Optoelectronics Progress, 2024, 61(18): 1812002
Paper Information
Category: Instrumentation, Measurement and Metrology
Received: Nov. 24, 2023
Accepted: Jan. 26, 2024
Published Online: Sep. 14, 2024
The Author Email: Yi An (anyi@dlut.edu.cn)
CSTR:32186.14.LOP232559
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20