• Acta Photonica Sinica
  • Vol. 54, Issue 1, 0106003 (2025)
Taifei ZHAO1,2,*, Jiahao GUO1, Yu XIN1, and Lu WANG1
Author Affiliations
  • 1Institute of Automation and Information Engineering,Xi'an University of technology,Xi'an 710048,China
  • 2Xian Key Laboratory of Wireless Optical Communication and Network Research,Xi'an 710048,China
  • show less
    DOI: 10.3788/gzxb20255401.0106003 Cite this Article
    Taifei ZHAO, Jiahao GUO, Yu XIN, Lu WANG. UAV UV Information Collection Method Based on Deep Reinforcement Learning[J]. Acta Photonica Sinica, 2025, 54(1): 0106003 Copy Citation Text show less
    UAV information collection diagram
    Fig. 1. UAV information collection diagram
    Geometric relationship between drone and ground sensor
    Fig. 2. Geometric relationship between drone and ground sensor
    Schematic diagram of drone movement
    Fig. 3. Schematic diagram of drone movement
    Improved DDQN algorithm framework
    Fig. 4. Improved DDQN algorithm framework
    Algorithm convergence analysis
    Fig. 5. Algorithm convergence analysis
    Trajectories of improved DDQN algorithm and DDQN algorithm with 20 ground sensors
    Fig. 6. Trajectories of improved DDQN algorithm and DDQN algorithm with 20 ground sensors
    Effects of different UV TX full beam angle on Information Collection time and energy consumption
    Fig. 7. Effects of different UV TX full beam angle on Information Collection time and energy consumption
    Effects of different UV RX FOV on information collection time and energy consumption
    Fig. 8. Effects of different UV RX FOV on information collection time and energy consumption
    Effect of different sensor quantities on information collection time and energy consumption
    Fig. 9. Effect of different sensor quantities on information collection time and energy consumption
    Effect of different sensor data volumes on information collection time and energy consumption
    Fig. 10. Effect of different sensor data volumes on information collection time and energy consumption
    Effect of different UAV flight altitudes on information collection time and energy consumption
    Fig. 11. Effect of different UAV flight altitudes on information collection time and energy consumption
    ParametersValue
    UV wavelength (λ)260 nm
    Transmit power (Pt)200 mW
    Atmospheric absorption coefficient (K)0.9×10-3 km-1
    Mie scattering coefficient (KsRay)0.24×10-3 km-1
    Rayleigh scattering coefficient (KsMie)0.25×10-3 km-1
    Filter transmittance (ηf)0.1
    Detector quantum efficiency (ηr)0.2
    Receiver aperture (Ar)0.25×10-4 m2
    Table 1. Ultraviolet light communication parameters
    ParametersValue
    Learning Rate (lr)0.000 1
    Discount Factor (γ)0.99
    Batch size (K)128
    Maximum number of training rounds (E)2 000
    Number of hidden layers in neural network3
    Number of hidden layer neurons512
    Initial greedy exploration probability0.8
    Termination of greedy exploration probability0.001
    Soft update factor τ0.01
    Experience buffer capacity50 000
    Table 2. Deep reinforcement learning training parameters
    Taifei ZHAO, Jiahao GUO, Yu XIN, Lu WANG. UAV UV Information Collection Method Based on Deep Reinforcement Learning[J]. Acta Photonica Sinica, 2025, 54(1): 0106003
    Download Citation