In recent years, unmanned aerial vehicle (UAV) has been widely used in target tracking tasks such as security, anti-terrorism, tracking and monitoring because of its wide and flexible perspective. At the same time, deep learning creates new conditions for UAV visual recognition and motion tracking because it has high recognition accuracy and fast recognition speed. This paper presents an UAV dynamic tracking algorithm based on deep learning. The algorithm mainly includes three parts: the PTZ control, relative position calculation and the UAV control. The main work includes: 1. The tracking target is identified through the deep learning network based on YOLOv3, and the PTZ angle is adjusted in real time by using the control method of combining the PID control with the time-delay control to reduce the impact of image transmission delay; 2. The relative position between the UAV and the target is calculated by combining the UAV flight altitude, UAV attitude, PTZ angle and the coordinate of the target in the image. At the same time, the relative position prediction is added to estimate the relative distance between the UAV and the target at the next frame time as a reference value of the UAV speed, so as to improve the dynamic tracking ability; 3. The relative position information is fed back to the flight control function, and the linear velocity of UAV is output by the PID controller. Through the simulation test in Airsim, it is proved that the proposed algorithm can effectively improve the stability and robustness of the UAV tracking system.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    UAV Dynamic Tracking Algorithm Based on Deep learning


    Contributors:
    Chen, Sifan (author) / Yang, Haoran (author) / Zhang, Ang (author) / Chen, Baihe (author) / Shu, Peng (author) / Xiang, Jin (author) / Lin, Chen (author)


    Publication date :

    2021-12-01


    Size :

    2339548 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    A Deep Reinforcement Learning Tracking Algorithm Based on Task Decomposition

    Yang, Kun / Shen, Ao / Xu, Nengwei et al. | IEEE | 2024



    Deep Learning Based Target Tracking and Diving Algorithm in Kamikaze UAVs

    Can, Erdem / Kirklar, Bedirye Ikbal / Namdar, Mustafa et al. | IEEE | 2024


    Deep Learning for Model-Based Multiobject Tracking

    Pinto, Juliano / Hess, Georg / Ljungbergh, William et al. | IEEE | 2023