Answers are grounded in this paper and cite their sources.
Cooperative area coverage by multiple unmanned aerial vehicles (UAVs) requires efficient exploration while avoiding collisions and maintaining communication within each group. Existing coverage methods are mainly rule-based or learning-based. Rule-based methods depend heavily on handcrafted heuristics and have limited adaptability to complex environments, whereas learning-based methods often couple exploration planning with motion control in a single policy, increasing the difficulty of policy learning. As the UAV fleet size increases, group-based organization becomes increasingly important for scalable coordination. However, most existing methods still treat UAVs as a flat swarm and lack explicit mechanisms to distinguish intra-group interactions from inter-group coordination.To address these limitations, we develop a topology-aware hierarchical coverage framework tailored to fixed-group multi-UAV exploration. First, a planning--execution hierarchy decouples frontier-based target assignment from low-level motion control, thereby reducing the policy-learning burden. Second, the execution layer employs a group-aware hierarchical topology encoder. Its sequential Graphormer modules construct UAV-centric action representations by first encoding intra-group communication dependencies and then inter-group exploration relationships, enabling complementary exploration across groups while preserving intra-group communication. We evaluate the proposed method and the baselines on 200 unseen randomly generated maps, comprising 50 Easy, 100 Medium, and 50 Hard maps. On the Hard maps, our method achieves a final coverage rate (FCR) of 91.33%, exceeding the strongest reinforcement-learning-based baseline by 14.98 percentage points, while limiting the collision rate (CR) and communication violation rate (CVR) to 1.81% and 9.32%, respectively, with the lowest CVR among all compared methods.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: