Cooperative tracking control of UAV and UGV based on YOLOv7 and DeepSORT

The emergence of cross-domain joint technology has spurred extensive research on Unmanned Aerial Vehicle-Unmanned Ground Vehicle collaboration. However, most current studies neglect the UAV’s flight tracking capability, which shows great potential for automated battlefield attacks and criminal pursuits. This paper focuses on UAV flight tracking and proposes a method using YOLOv7 and DeepSORT, equipped with an algorithm for automatic obstacle avoidance during tracking. The algorithm decomposes the motion of the UAV into x, y, z velocities and yaw angular velocity, and relevant velocity models are built to adapt to the complex environment. The proposed method is demonstrated in both simulation and flight experiments.