Particle Swarm Optimization for Self-Organizing Obstacle Avoidance of UAV Swarms

Organizing a swarm of unmanned aerial vehicles (UAVs) to achieve the desired collective behavior is a huge challenge due to the multitude of tunable parameters in the UAV self-organizing model, especially in the environment with complex obstacles. This paper proposes an optimization framework based on the willingness control method (WCM) and particle swarm optimization (PSO). Specifically, we introduce relevant indicators for UAVs motion from five perspectives: consistency, connectivity, collision avoidance, obstacle avoidance, and spatial distribution. We then treat this optimization task as a single-objective optimization problem and design a corresponding fitness function based on the proposed indicators. We conducted simulation experiments on the optimized and unoptimized models in scenarios with narrow channels and dense obstacle fields. The performance of the optimized model showed significant improvement, validating the effectiveness of the proposed indicators. The results confirm that the proposed motion indicators can effectively evaluate the UAVs models, leading to better performance of the PSO algorithm in adjusting the parameters of the UAVs models.