Long-Endurance Collaborative Search and Rescue Based on Maritime Unmanned Systems and Deep-Reinforcement Learning.

Area of Science:

• Robotics and Autonomous Systems
• Maritime Operations
• Search and Rescue Technology

Background:

• Maritime search and rescue (SAR) missions face challenges in complex environments.
• Effective collaboration between unmanned aerial vehicles (UAVs) and unmanned surface vehicles (USVs) is crucial for SAR success.
• Existing systems often lack long-endurance capabilities and efficient coordination.

Purpose of the Study:

• To propose a long-endurance collaborative SAR scheme for maritime environments.
• To leverage the complementary strengths of UAVs and USVs for enhanced SAR operations.
• To address energy constraints and prolong the operational time of unmanned systems.

Main Methods:

• A multi-agent reinforcement learning (MARL) approach for cooperative first-phase search by UAVs.
• Utilizing probability maps for efficient search area refinement.
• Implementing a multi-USV charging scheduling method based on MARL to extend UAV endurance.

Main Results:

• The proposed scheme effectively integrates UAVs and USVs for collaborative SAR.
• Demonstrated enhanced search capabilities through multi-phase operations.
• Verified the long-endurance performance of the system via extensive simulations.

Conclusions:

• The developed collaborative SAR scheme significantly improves maritime search and rescue efficiency.
• The MARL-based coordination and charging strategy ensure extended operational endurance.
• This approach offers a robust solution for complex maritime SAR scenarios.