In order to solve the parameter sensitivity and local convergence problems in the three-dimensional path planning of UAVs, this paper proposes three improvement strategies. First, involves strengthening the algorithm’s foundation through the construction of a three-dimensional obstacle model and the formulation of a fitness evaluation function. Second, interval-constrained Logistic chaotic mapping is used to optimize the initial population distribution to enhance diversity. A nonlinear iterative framework incorporating Cauchy mutation is proposed to dynamically optimize exploration-exploitation trade-offs. Empirical evaluations indicate the improved algorithm’s significant advantages over state-of-the-art methods in global optimization and stability metrics.