摘要
多机器人对抗系统中,攻防双方在合作联盟形式下共同追求各自的最高预期收益,最终收敛于博弈均衡状态。提出一种面向多机器人系统可计算的对抗博弈均衡求解方法。首先,形式化描述多机器人对抗博弈系统的状态迁移过程,并给出机器人博弈策略的表示及推理方法;然后设计一个DOAN算法,并以Stackelberg安全博弈为例求取均衡解;最后,通过模拟实验表明,提出的方法在可接受的时间内收敛于均衡解。
In multi-motor countermeasure system,the players of attack and defense pursue their highest expected revenue in the form of cooperative alliance,and it eventually converges to the equilibrium state of the game.A computational equilibrium solving method against game oriented to multi-robot system is proposed.The state transition model of multi-robot system against game is described formally,and the representation and seasoning method of robot game strategy is given.A DOAN(double-Oracle based algorithm for allocating resources on nodes)is designed,and Stackelberg security game is taken as an example for equilibrium solving.The results of simulation experiment show that the proposed method can converge to the equilibrium solution within the acceptable time.
作者
朱泽宇
陆建波
杨洲
钟智
滕玲
ZHU Zeyu;LU Jianbo1;YANG Zhou;ZHONG Zhi;TENG Ling(School of Computer and Information Engineering,Guangxi Teachers Education University,Nanning 530023,China;College of Information Engineering,Yangzhou University,Yangzhou 225127,China)
出处
《现代电子技术》
北大核心
2018年第23期165-168,共4页
Modern Electronics Technique
基金
国家科技支撑计划课题(2015BAH55F02)
广西科学研究与技术开发计划项目(桂科攻14124005-2-7)~~
