A multi-agent three-dimensional path planning method based on the multi-objective Pareto optimization framework
TL;DR
At the framework level, this paper completes a methodological upgrade from scalar weighted-sum to multi-objective Pareto optimization, providing a more scientific decision-support system for multi-agent cooperative search.
Abstract
Aiming at the multi-objective path planning problem when multiple agents cooperatively search for maneuvering targets in complex three-dimensional environments, this paper proposes an improved 3D Artificial Bee Colony algorithm (MOABC) based on the multi-objective Pareto optimization framework. The proposed method treats path length, detection probability, threat exposure, and mission time as four optimization objectives. A Pareto dominance relationship combined with an external archive mechanism filters the non-dominated solution set, and a PSO guidance mechanism accelerates the convergence of employed-bee search. A rolling-window replanning strategy is embedded into the MOABC framework to achieve real-time tracking of maneuvering targets. Simulation results on a typical 3D battlefield environment show that, compared with the scalar weighted-sum method, the MOABC framework reduces threat exposure by 22.6%, achieves a 98.5% target detection probability, and limits planning time to within 2 seconds. Statistical significance is confirmed by 30 Monte Carlo runs with p < 0.01 for all performance metrics. At the framework level, this paper completes a methodological upgrade from scalar weighted-sum to multi-objective Pareto optimization, providing a more scientific decision-support system for multi-agent cooperative search.