Resilience-Oriented Real-Time Decision-Making for Autonomous Recovery in Multi-UAV Systems

Autonomous real-time decision-making for multiple unmanned aerial vehicles (multi-UAV) mission planning is crucial yet challenging due to unexpected disruptions. In this brief, we propose a resilience-oriented decision-making framework that enables UAVs to autonomously generate recovery strategies. To quantify the performance of multi-UAV systems in mitigating disruptions and restoring operations, we introduce a resilience metric that considers the mission completion rate and the resources required for maintenance. On this basis, we formulate a joint optimization model that incorporates component maintenance and task rescheduling as recovery strategies to maximize system resilience. To enable UAVs to make real-time decisions based on local observations, we conceive a multiagent deep Q-network (MADQN) algorithm assited by a pigeon flocks swarm intelligence mechanism to find the optimal recovery policy with enhanced learning efficiency. Simulations validate the effectiveness and superiority of the proposed methods.