Synthesized Control for In-field UAV Moving Target Interception via Deep Reinforcement Learning and Fuzzy Logic

This paper presents significant advancements to our previously proposed deep reinforcement learning and fuzzy logic-based algorithm for dynamic target interception using multiple UAVs. Key components of the original framework have been redesigned to enhance real-world safety and adaptability. In support of practical deployment, we developed a comprehensive cross-platform simulation environment integrating MATLAB, ROS, and PX4, allowing for robust testing and validation. Additionally, a fully programmable drone has been constructed to bridge the gap between simulation and reality. The improved algorithm has undergone rigorous evaluation through both simulation and real-world flight tests in complex, dynamic conditions, demonstrating its effectiveness and reliability for autonomous UAV applications.