Anti-Collision Formation Control for Drone Swarms Tracking Aerial Targets Under Wind Disturbance

Drone swarms are increasingly used for complex aerial missions requiring coordination, adaptability, and wide-area observation. This paper presents a robust decentralized formation control strategy for a group of quadrotor drones assigned to track and monitor a moving aerial target in dynamic environments. The main challenges addressed include maintaining stable coordination under wind disturbances and preventing inter-drone collisions, without relying on centralized supervision or global shape feedback. A backstepping control optimized by Flower Pollination Algorithm is proposed, where wind effects are directly integrated into the dynamics and compensated through feedforward terms. In contrast to other formation strategies, the proposed approach enables formation-level coordination using only local relative position data, facilitating decentralized implementation. Each follower drone maintains a regular pentagon formation around the target to ensure full viewing angle. While ideal sensing is assumed for the simulation, the framework is compatible with practical sensor constraints. To enhance performance, the control gains are optimized using the metaheuristic flower pollination algorithm, selected for its balance between convergence speed and global search capability. Extensive simulations demonstrate that the proposed system achieves accurate tracking, stable formation maintenance during target movement in different scenarios, effective collision avoidance, and robustness against realistic wind disturbances. Results show overshoot and a settling time ranging from 1 to 2 seconds, depending on the initial and final positions of each drone in the pentagonal formation, which confirms stability and precision. This contribution bridges robust nonlinear control and fully decentralized swarm coordination, offering a scalable and resilient solution for multi-agent aerial operations.