Scalable Drone Swarm Formation Planning via Distributed Hierarchical Optimization

Multiple robust schemes have been proposed for multi-agent formation preservation in dense environments. However, these approaches exhibit prohibitive computational complexity when scaling to larger swarms. To address this limitation, this paper proposes an optimization-based hierarchical framework for efficient planning of large-scale drone swarm formation. In this paper, the optimal formation trajectory is proposed as the reference trajectory, disregarding dynamic constraints and obstacle avoidance while using fixed to-be-optimized parameters invariant to sampling density. We then formulate the reference trajectory into an optimization framework using a normalized trajectory error to quantify the deviation between the optimized trajectory and the reference trajectory. By aligning the optimized trajectory with the reference trajectory for formation planning, the approach decouples formation coordination from trajectory optimization. In simulations, we validate its reliability for formation cooperation and conduct experiments in scenarios with 5 to 12 robots.