Inspired by the collective behaviors observed in bird flocks and fish schools, this paper proposes a novel Decentralized Model Predictive Flocking Control (DMPFC) framework to enable UAV swarms to autonomously track predefined reference trajectories while avoiding collisions and maintaining a stable quasi \(\alpha \) -lattice formation. Unlike traditional approaches that rely on switching between predefined swarm formations, this framework utilizes identical local interaction rules for each UAV, allowing them to dynamically adjust their control inputs based on the motion states of neighboring UAVs, external environmental factors, and the desired reference trajectory, thereby enabling the swarm to adapt its formation dynamically. Through iterative state updates, the UAVs achieve consensus, allowing the swarm to follow the reference trajectory while self-organizing into a cohesive and stable group structure. To enhance computational efficiency, the framework integrates a closed-form solution for the optimization process, enabling real-time implementation even on computationally constrained micro-quadrotors. Theoretical analysis demonstrates that the proposed method ensures swarm consensus, maintains desired inter-agent distances, and stabilizes the swarm formation. Extensive simulations and real-world experiments validate the approach’s effectiveness and practicality, demonstrating that the proposed method achieves velocity consensus within approximately 200 ms and forms a stable quasi \(\alpha \) -lattice structure nearly ten times faster than traditional models, with trajectory tracking errors on the order of millimeters. This underscores its potential for robust and efficient UAV swarm coordination in complex scenarios.
