Feasibility Evaluation and Stagnation-Strategy Analysis of a Standardized Artificial-Lemming Implementation for Three-Dimensional UAV Path Planning

This study evaluates a standardized artificial-lemming algorithm and three modifications for constrained three-dimensional unmanned aerial-vehicle path planning. A common segment-wise geometric evaluator supports 3600 runs across 12 synthetic scenes. The combined strategy and standardized baseline produce 296/360 and 298/360 feasible outcomes, respectively. The initialization mapping preserves the ideal uniform distribution. Feasibility certificates, uncertainty intervals, and candidate replay distinguish route existence from search reliability and explain perturbation rejection and collision transfer. An adaptive differential evolution comparator produces 335/360 feasible outcomes and significantly exceeds the combined strategy in one narrow-passage scene after multiplicity correction. A rapidly exploring random tree provides a separate geometric reference under a different query budget. Additional evidence comprises 300 runs on 12 newly generated maps, 225 waypoint-count and objective-weight screening runs, 270 stagnation-parameter screening runs, and 1740 standard-function implementation-check runs. The results support configuration-specific failure analysis within the tested synthetic environments. The main contribution is a controlled evaluation and diagnostic workflow for assessing added search strategies under geometric constraints.