A Scallop Optimizer and Its Application in 3-D UAV Path Planning
TL;DR
The proposed SPO provides an effective alternative optimization tool for complex constrained engineering optimization tasks such as 3D UAV path planning and significantly outperforms 14 mainstream metaheuristic algorithms, including PSO, DE, SHADE, and DBO, on most test functions.
Abstract
To address the inherent limitations of existing metaheuristic algorithms in solving complex three-dimensional (3D) unmanned aerial vehicle (UAV) path planning problems, such as premature convergence and weak adaptability to multi-dimensional flight constraints, this paper proposes a novel Scallop Optimizer (SPO). Inspired by the composite survival behaviors of scallops, including filter-feeding, zigzag predation evasion, group defense, and byssus memory, SPO integrates five coordinated search modules: Adaptive Energy State Switching Mechanism (AESSM), Filter-Feeding Mechanism (FFM), Leaping-Zigzag Evasion Mechanism (LM-ZM), Group Defense-Dispersion Mechanism (GDDM), and Byssus Historical Optimal Memory Mechanism (BHM). AESSM dynamically adjusts the energy state of each individual to switch between global exploration and local exploitation; FFM enhances fine local search accuracy; LM-ZM improves the capability to escape local optima; GDDM maintains population diversity in real time; and BHM avoids redundant repeated search. Comprehensive numerical experiments on the IEEE CEC 2017 (D = 30) and CEC 2022 (D = 10 and 20) benchmark suites demonstrate that SPO achieves minimum average rankings of 1.14, 1.58, and 1.50 across three test sets, and significantly outperforms 14 mainstream metaheuristic algorithms, including PSO, DE, SHADE, and DBO, on most test functions. Further ablation experiments verify that the GDDM module contributes the most to performance improvement, with an average ranking degradation of 6.00 upon its removal. When applied to multi-constraint 3D UAV path planning, SPO obtains a minimal total flight cost of 1298.32, reducing the comprehensive path cost by 40.41% compared to the worst-performing optimizer, and generates collision-free, smooth trajectories with zero terrain and threat penalty costs. Statistical Wilcoxon signed-rank and Cohen’s d tests further validate the significant statistical superiority of SPO. The proposed SPO provides an effective alternative optimization tool for complex constrained engineering optimization tasks such as 3D UAV path planning.