Author(s): Yu Zhe; Lu Wenjing
Linked Author(s):
Keywords: Deep reinforcement learning; Unmanned aerial vehicle swarm; Flood remote sensing monitoring; Efficiency evaluation
Abstract: UAV remote sensing technology is widely utilized in urban supervision, flood monitoring, and various other fields. The key of flood emergency monitoring is to obtain the early warning information of dam leakage in time. Although the UAV swarms can better cope with the adverse weather conditions and collect more comprehensive data information in the flood early warning and monitoring. However, currently the allocation of UAV swarms is mainly based on expert experience, which needs to be allocated well in advance, and it is difficult to effectively control in real time, which affects the operation efficiency of UAV. In this paper, a UAV swarms efficiency evaluation method is proposed, which concentrates on two indices of task completion time and data completeness. The above two indices are used as the basis for designing the reward value function of the deep reinforcement learning method, which is used to train the UAV quantity allocation agent. This method integrates the prior knowledge of experts and the autonomous learning of agents, which is well-suited to dynamic changes in the environment and is expected to enhance the efficiency of collaborative monitoring by UAV swarms.
Year: 2024