Research on Intelligent Maneuvering Decision Making Based on Deep Reinforcement Learning

At present, the maneuvering environment is complex and changeable, and unmanned platforms are faced with a variety of uncertain factors when carrying out multiple tasks. In order to improve the efficient and safe maneuvering performance of unmanned platforms in complex scenarios, this paper proposed building a maneuvering decision network system by using deep reinforcement learning network DDPG network for guiding agents execute maneuvering decisions according to tasks and environments to achieve the purpose of safe and efficient maneuvering. The paper tested the maneuvering decision performance in dynamic changing scenarios and verified the autonomous intelligent maneuvering performance of network guided agents, which could effectively avoid threats and choose better paths, and had good robustness and generalization.