Reinforcement learning-based drone-assisted collection system for infection samples in IoT environment

Since infectious disease surveillance and control rely on efficient sample collection, it is important to research the infection sample collection system. The combination of Internet of Things (IoT) and drone technology provides an emerging solution to this issue. This paper designs a drone-assisted collection system for infection samples (DASS) that provides safe, intelligent, and efficient sample collection services. In this system, flexible collector drones gather infection samples from remote users and return to designated transit points to unload. Meanwhile, deliverer drones shuttle between the testing center and transit points, transporting all packaged infection samples to the testing center. However, the moment when users post collection requests is unknown. This dynamism and uncertainty present new challenges for the routing and scheduling of heterogeneous drones. To address this issue, this paper proposes a deep reinforcement learning-based dynamic sample collection (RLDSC) scheme. Considering the differences in infection samples, minimizing age of samples (AoS) is introduced as an objective. Simulation results indicate that the RLDSC scheme outperforms existing solutions in both effectiveness and efficiency.