Answers are grounded in this paper and cite their sources.
In low-altitude applications, Unmanned Aerial Vehicles (UAVs) have gained widespread adoption, expanding their roles across diverse domains such as search and rescue, communication coverage, and security surveillance. In complex and remote environments, UAVs are frequently deployed as communication relays to support ground search units (GUs). However, a critical challenge persists in ensuring effective communication coverage for GUs during penetration missions while simultaneously minimizing interception risk. To address this challenge, we propose a multi-agent reinforcement learningbased cooperative communication coverage algorithm with low probability of interception (RLC³-LPI) for multi-UAV trajectory planning. Our approach employs a centralized training and decentralized execution paradigm, eliminating the UAV swarm's reliance on global environmental awareness and enhancing scalability in real-world scenarios. Within RLC³-LPI, the low-intercept coverage problem is formulated as a partially observable Markov decision process, and a federation-enhanced deep reinforcement learning algorithm is developed, incorporating meticulously designed reward functions to optimize connectivity, coverage efficiency, and interception avoidance. Extensive experimental results demonstrate that RLC³-LPI outperforms existing benchmarks, achieving higher coverage and connectivity rates while significantly reducing interception probability.
No takes yet. Share an insight, caveat, or question.
Zhao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: