This paper proposes a new approach to optimize the coordination of drone swarms in Smart Cities. It combines distributed optimization and deep reinforcement learning (DRL) to improve navigation, sensing, and recharging of drones. The distributed optimization provides cost-effective, high-quality drone operations, while DRL enables long-term planning for better efficiency and effectiveness. This approach addresses the limitations of short-term optimization and the scalability issues of long-term DRL. The proposed solution has shown outstanding performance in traffic monitoring compared to baseline methods.
Publication date: 20 Nov 2023
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2311.09852
