Cooperative Search Method for Multiple UAVs Based on Deep Reinforcement Learning.

Area of Science:

• Robotics
• Artificial Intelligence
• Operations Research

Background:

• Multi-unmanned aerial vehicle (UAV) systems face challenges in efficient task execution.
• Cooperative games with incomplete information present complexities in optimizing multi-agent systems.

Purpose of the Study:

• To propose an efficient cooperative search method for multiple UAVs.
• To enhance the collaboration and task execution efficiency of UAV swarms.

Main Methods:

• A cooperative game with incomplete information framework is utilized.
• Consensus-Based Bundle Algorithm (CBBA) is applied for task area designation.
• Independent Deep Reinforcement Learning (IDRL) is employed to achieve Nash equilibrium for improved collaboration.
• A novel reward function is developed to guide UAVs towards high-reward paths and collision avoidance.

Main Results:

• The proposed method demonstrates superior performance compared to other algorithms.
• Simulations show increased reward acquisition within a given time frame.
• Effective collision avoidance between UAVs during flight operations was achieved.

Conclusions:

• The developed cooperative search method significantly improves the efficiency of multi-UAV task execution.
• The integration of CBBA and IDRL offers a robust solution for complex UAV coordination problems.
• The proposed approach provides a valuable contribution to the field of autonomous multi-agent systems.