Knee Point-Guided Online Cooperative Path Planning for UAVs in Dynamic Multi-Objective Environments
针对动态威胁环境下无人机在线路径规划需权衡路径长度与安全风险的问题，提出一种膝点引导的动态多目标规划方法，包含双机协同干扰、动态角色切换和膝点引导的变更响应机制，实验证明能快速生成可行高效路径。
Online path planning for unmanned aerial vehicles requires dynamic trade-offs between path length and safety risks in complex and dynamically changing threat environments. However, most of recent researches either do not consider multi-UAV cooperative operations or do not incorporate online responses to dynamic threats. To address these limitations, a knee point-guided dynamic multi-objective path planning method is proposed in this paper. The main contributions are as follows: 1). A dual-UAV cooperative jamming mechanism is proposed to actively respond to threat risks and improve mission success rates; 2). A dynamic role-switching mechanism is developed under a leader-follower framework to enhance the system’s flexibility in responding to threats; 3). A knee point-guided change response mechanism is constructed, which integrates neighborhood sampling and imbalanced transfer learning to adaptively adjust the response strategy based on the degree of environmental change. Experimental results demonstrate that the proposed mothed can effectively and rapidly respond to environmental changes, enabling UAVs to generate feasible and efficient online flight paths under complex dynamic threat conditions.