A Multi-UAV Cooperative Ground Target Tracking System Based on a Two-Layer State Fusion Structure
Ruosi Kong, Shouwen Yao, Zeling Lan, Yu Wang, Heweiqi Gou · 2022
UAVs (Unmanned Aerial Vehicles) have been used in many missions, such as disaster management, traffic monitoring, target tracking, etc. However, an individual UAV has a limited field of view and is prone to lose target in a tracking task. In this paper, we propose a system framework for cooperative tracking of ground moving targets by multi-UAVs to achieve better performance in long time target tracking. Firstly, the UAV-based algorithm AutoTrack is adopted to track targets. Secondly, we propose a two-layer state fusion estimation architecture and use an extended Kalman filtering method to fuse the information and obtain accurate target localization. Finally, experiments on tracking targets from three videos taken under different situations are implemented. Results show that the proposed system performs better than a single UAV. The fps of the tracking algorithm is 25.1 which meets the need of realtime tracking. By applying the fusion estimation, the position error of x and y direction are reduced by 94.2% and 90%.