Vision-Based Estimation and Tracking on Fixed-Wing Multi-UAV Systems

Accurate tracking of multiple fixed-wing UAVs is a crucial challenge in autonomous aerial systems, particularly in GPS-denied environments where traditional sensors struggle. The bearing-angle estimator extracts target position and velocity purely from onboard visual data, eliminating the need for external positioning systems. This article presents a vision-based estimation and tracking framework for multi-UAV systems that integrates a YOLOv8 and ByteTrack-based tracker based on the bearing-angle state estimation method. ROS2 and Gazebo-based simulation validates the approach, demonstrating successful onboard real-time tracking of three UAVs in formation flight, making it a solution for vision-based multi-UAV coordination.