Target Tracking Research Based on UAV Localized Upper Prototype
Xiao Kang, Huimin Chen · 2025
In response to the problems of slow inference speed, large volume, inability to carry small unmanned aerial vehicles, and high cost in the field of target tracking in current unmanned aerial vehicles, the mainstream target tracking algorithm was studied, and the kernel correlation filtering algorithm was ultimately selected for improvement. This algorithm can quickly infer and track the tracked target, and it was deployed on the domestically produced small upper prototype RK3566. Therefore, a unmanned aerial vehicle target tracking system based on the domestically produced upper prototype RK3566 was designed. The experimental results show that the deviation between the tracking trajectory of the drone and the trajectory command issued by the upper prototype RK3566 is less than 0.05m, and the mean position error of the drone on each axis in the three-dimensional coordinate axis is less than 0.15m. After deploying the nuclear correlation filtering algorithm, the domestically produced small-scale upper level prototype RK3566 can control unmanned aerial vehicles to quickly and efficiently track targets.