Flexible Mobility MADDPG Algorithm for UAV Swarm Unknown Region Searching
Heng Huang, Jianfeng Guo, Kun Zhu, Banghuang Peng · 2023
The application of UAV swarm has been very wide, including rescue, search, confrontation, assistance communication, path search, etc. Usually, when UAV swarm performs a task, it needs to solve many problems, including but not limited to the requirement of high efficiency in the execution of the task, the interference of the execution environment to the UAV, and the degree of cooperation between each individual UAV. However, some tasks are unknown and need an exploration process, and the exploration process can be repeated to mine the cooperation between UAV clusters, so as to improve the efficiency of the task execution. In order to improve the efficiency of UAV clusters searching unknown areas, this paper proposes an efficient UAV cluster search algorithm based on FM-MADDPG. The FM-MADDPG proposed in this paper optimizes the detection efficiency of UAV clusters in unknown areas, and the flexible assistance between multiple UAVs improves the efficiency of UAV clusters in performing tasks. We establish the UAV kinematics model and task model, and conduct simulation experiments. FM-MADDPG has better performance than MADDPG and DDPG algorithms in convergence speed and reward value.