Abstract: Intelligent unmanned systems are required to make decisions quickly and stably in complex environments and have the ability to deal with unexpected states, but due to the high complexity of environment and tasks, it is often difficult for them to implement decision-making management. The reinforcement learning platform can provide a good solution to this problem. In view of the diversity, complexity, high dynamics and uncertainty of the environment in which the intelligent unmanned system is located, the decision-making management is carried out by using the reinforcement learning platform, and the environment and situation are accurately perceived and decided in the case of limited sensors, so that the agents can use self-learning and adaptive capabilities to quickly make decisions. Reinforcement learning learns the decision-making strategy through the autonomous interaction with the environment, so as to maximize the long-term cumulative reward value of the strategy. Through the connection of the reinforcement learning platform and the simulation platform, decision-making model construction and agent training are performed, and the decision-making management of intelligent unmanned systems is realized through the control of agent output strategy.

Key words: intelligentunmannedsystem, reinforcementlearning, decisionmanagement, agent