Vision-Based Formation Control for an Outdoor UAV Swarm with Hierarchical Architecture

Formation control of a UAV swarm is challenging in outdoor GNSS-denied environments due to the difficulties accomplishing relative positioning among UAVs. This study proposes vision-based formation strategy that could be implemented absence an external system. The hierarchical architecture has been constructed for using modified leader-follower strategy. leader derives and broadcasts locations follower UAVs, while UAVs calculate their inputs achieve desired formation. localization accomplished state-of-the-art deep learning algorithms like YOLOv7 DeepSORT. RflySim-based simulation conducted verify feasibility conceptualization, validation made with real flight test comprised five quadrotors. Results show robustness framework error within 0.3 m. Moreover, without GNSS achieved monocular camera entry-level AI platforms onboard, which promoted applications broader scenarios.