A multi-UAV cooperative path planning method based on HCA-MAPPO deep reinforcement learning

Key Points

• The proposed method improves cooperative path planning efficiency for multi-UAV systems, enhancing exploration in sparse reward scenarios.
• Key improvements include a hierarchical cross-attention network that processes heterogeneous information, boosting situational awareness.
• Modeling the planning problem as a partially observable Markov decision process allows effective algorithmic solutions.
• Simulation experiments validate the effectiveness and superiority of the proposed approach in real-world applications.

Abstract