Optimizing Swarm Size in UAV Path Planning: Balancing Fitness, Convergence, and Computational Efficiency

Unmanned aerial vehicle (UAV) path planning represents a pivotal component in autonomous navigation systems, with its primary objective being the determination of safe, efficient, and optimized flight trajectories within complex operational environments. Among various optimization methodologies, particle swarm optimization (PSO) has emerged as a prominent solution for path planning challenges, owing to its computational efficiency and straightforward implementation. Nevertheless, the efficacy of PSO is intrinsically linked to its parameter configuration, where different parameter selections can substantially influence the algorithm's search capacity and convergence dynamics. This research systematically examines the effects of four critical PSO parametersinertia weight, social weight, cognitive weight, and swarm sizeon both optimization quality and computational performance in UAV path planning scenarios. The investigation commences with the formulation of a mathematical framework for UAV path planning, incorporating path length and obstacle avoidance capability as primary evaluation metrics. Subsequently, through comprehensive simulation experiments, we meticulously analyze the influence of various parameter configurations on PSO's convergence characteristics, employing advanced data fitting techniques to establish precise mathematical relationships between parameter settings and algorithmic performance. The empirical results reveal that strategic parameter optimization substantially enhances PSO's global search capability, improves path quality, and expedites convergence rates. Based on rigorous analysis, the optimal parameter configuration is determined to be: inertia weight (w) = 0.659, social weight (c) = 1.349, cognitive weight (c) = 2.108, and swarm size (M) = 88. These findings not only provide theoretical foundations but also offer practical optimization strategies for parameter selection in UAV path planning applications, with potential extensions to other intelligent optimization domains.