Pattern-aware multiobjective optimization with multimodal representations for UAV reconnaissance and task offloading

M. Yu

doi:10.1016/j.patcog.2026.114589

Pattern-aware multiobjective optimization with multimodal representations for UAV reconnaissance and task offloading

2026-08-01

0

PRE

AI

DOI：10.1016/j.patcog.2026.114589

Original

Original request for help

Save

Abstract

En 中文

Cooperative reconnaissance and task offloading in heterogeneous multi-UAV systems form a coupled multiobjective optimization problem, where target coverage, Dubins-constrained flight, communication-computation service, threat exposure, and energy consumption must be jointly coordinated. Existing studies usually optimize reconnaissance planning and task offloading separately, making it difficult to capture the interactions among heterogeneous mission patterns and service requirements. To address this problem, this paper formulates a pattern-aware multiobjective optimization model with multimodal mission representations. The model represents point-like, line-like, and region-like targets, spatial threat fields, and task-service demands in a unified framework, and jointly optimizes task assignment, visiting sequence, Dubins path transition, communication delay, computation delay, threat exposure, and energy consumption. The main technical novelty lies in the proposed entropy-guided adaptive multiobjective particle swarm optimization algorithm, termed EGA-MOPSO. Unlike standard MOPSO, EGA-MOPSO introduces an entropy-guided leader selection mechanism to regulate objective-space diversity, an adaptive differential perturbation term to enhance decision-space exploration, and an intuitionistic-fuzzy-entropy-based mutation control strategy to identify evolutionary states and avoid search stagnation. These mechanisms work jointly to improve convergence guidance, diversity preservation, and robustness in strongly coupled multiobjective search. Experiments on ZDT, DTLZ, and WFG benchmark suites show that EGA-MOPSO achieves competitive convergence, diversity, and Pareto solution quality. Heterogeneous multi-UAV simulations further demonstrate that the proposed framework generates feasible task-allocation and trajectory-planning schemes under different target patterns and threat configurations, providing diverse tradeoff solutions for cooperative reconnaissance and task offloading.

Keywords:

Multimodal mission representation

Pattern-aware multiobjective optimization

Cooperative reconnaissance

Task offloading

Heterogeneous multi-UAV systems

Pattern-aware multiobjective optimization with multimodal representations for UAV reconnaissance and task offloading

Abstract

Journal

Organization

Cited Papers