Rough Tracking Method for Satellite Laser Communication Based on YOLOv5s(Invited)

Peng Yin; Xiaolong Ni; Chunyi Chen; Xin Yu

doi:10.3788/LOP240883

Laser & Optoelectronics Progress, Volume. 61, Issue 7, 0706012(2024)

Rough Tracking Method for Satellite Laser Communication Based on YOLOv5s(Invited)

Peng Yin¹, Xiaolong Ni^2,4,5、*, Chunyi Chen^3、**, and Xin Yu⁴

¹School of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun 130022, Jilin, China

²Institute of Space Ophotoelectronics Technology, Changchun University of Science and Technology, Changchun 130022, Jilin, China

³School of Computer Science and Technology, Changchun University of Science and Technology, Changchun 130022, Jilin, China

⁴School of Optoelectronic Engineering, Changchun University of Science and Technology, Changchun 130022, Jilin, China

⁵Changchun Lighteek Photonics Inc., Changchun 130022, Jilin, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(16)

Fig. 1. Overall structure diagram of YOLOv5s improvement

Download full size

Fig. 2. Schematic diagram of SIoU loss calculation

Download full size

Fig. 3. Structure of the CARAFE module

Download full size

Fig. 4. CBAM attention mechanism structure

Download full size

Fig. 5. Coordconv and traditional Conv structure diagram

Download full size

Fig. 6. Relationship between the number of anchor boxes and average IoU in K-means and K-means++

Download full size

Fig. 7. Training results of improved neural network on self-made spot dataset

Download full size

Fig. 8. Wireless optical communication turntable

Download full size

Fig. 9. Recognition effect of different coarse tracking capture algorithms. (a) Circle fitting algorithm; (b) centroid algorithm; (c) improved YOLOv5s algorithm

Download full size

Fig. 10. Static tracking experimental structure

Download full size

Fig. 11. Deviation of the horizontal axis and the pitch axis during tracking process between the spot position and the center point. (a) Deviation of the horizontal axis; (b) deviation of the pitch axis

Download full size

Table 1. Sample set composition
View table
Table 1. Sample set composition
Data set Positive Negative Sum Number of beacon spots
Beacon spot dataset 10000 1000 11000 10000

Table 2. Statistical table of test set detection results
View table
Table 2. Statistical table of test set detection results
Number of samples Category True positive False positive False negative True negative
1100 Beacon spot 982 3 7 108

Table 3. Jetson Orin NX operating environment
View table
Table 3. Jetson Orin NX operating environment
Project Parameter
Operating system Ubuntu 20.4
Python 3.8.10
PyTorch 1.8.0
CUDA 11.4
TensorRT 5.1.1

Table 4. Comparison between PyTorch framework and TensorRT framework
View table
Table 4. Comparison between PyTorch framework and TensorRT framework
Model Inference framework Data type Inference time /ms Frame rate /（frame/s）
YOLOv5s（changed） PyTorch FP32 15.6 64.1
TensorRT FP16 9.1 109.8

Table 5. Static tracking error
View table
Table 5. Static tracking error
a₁ b₁ a₂ b₂ Actual a₂ Actual b₂
45 0 18.43495 0 18.43513 0.00073
30 30 10.89339 10.89339 10.89304 10.89421
0 45 0 18.43495 -0.00119 18.43390
-30 -30 -10.89339 -10.89339 -10.89273 -10.89405
-45 0 -18.43495 0 -18.43421 0.00127

Tools

Get Citation

Copy Citation Text

Peng Yin, Xiaolong Ni, Chunyi Chen, Xin Yu. Rough Tracking Method for Satellite Laser Communication Based on YOLOv5s(Invited)[J]. Laser & Optoelectronics Progress, 2024, 61(7): 0706012

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites