A Visual-Based Target Tracking Framework for UAV Using Model Predictive Contouring Control

Tracking moving targets is fundamental task in many applications of uncrewed aerial vehicles (UAVs). In practice, the visual information is hard to be processed in real time for detecting fast-moving and agile targets, and the conventional control method is difficult to track them. To address the challenge, a monocular vision based comprehensive framework for UAVs to track fast moving targets is proposed. In particular, the real-time visual target detection and measurement method with YOLOv8 is designed, and the model predictive contouring control is employed as the UAV control algorithm to optimize the control process, enabling the UAV to follow randomly moving targets. Finally, experimental results verify the effectiveness of the proposed framework under realistic scenarios.