【作者】 李航； 谭联群； 曾隽芳； 刘禹；

【Author】 Li Hang;Tan Lianqun;Zeng Junfang;Liu Yu;Unit 91776 of PLA;Communication & Repair Factory of Navy;Institute of Automation,Chinese Academy of Sciences;

【机构】 中国人民解放军91776部队； 海军青岛通信修理厂； 中国科学院自动化研究所；

【摘要】 兵棋推演能够较好地模拟战争进程和结局,辅助研究军事领域决策、提升战场指挥能力。随着深度学习、强化学习算法的巨大突破,如何将人工智能技术集成到计算机兵棋推演系统中形成军事智能博弈对抗系统,在复杂多变的战场环境下制定高效的军事策略并进行及时调整成为新的研究方向。本文分析智能兵棋推演的特征,研究智能兵棋推演系统总体框架,进而针对AI智能体进行了分析设计,其重点是人机对战和自主学习,并给出了系统开发组件。更多还原

【Abstract】 Wargaming can simulate the process and outcome of war, assist decision-making in military field and improve battlefield command ability. With the great breakthrough of deep learning and reinforcement learning algorithm, how to integrate AI technology into computer wargame and simulation system to form Military Intelligent Wargame System(MIWS), develop efficient military strategies in the complex and changeable battlefield environment has become an extremely valuable challenge. This paper analyses the features of the wargaming, studies the general framework of the intelligent wargaming system, and analyzes and designs the AI agent as well, which focuses on man-machine combat and autonomous learning. And system development components are presented.更多还原