摘要
详细调研无人艇的发展现状,分析其走向自主智能化的现实瓶颈,并揭示无人艇平台的缺失及智能博弈对抗能力差这2个制约智能无人艇落地的根本问题。针对这2个问题,结合数字孪生技术,给出水面无人艇群体智能研究平台设计方案,同时也提出对基于深度学习的多智能体博弈对抗的相关可行性设想,为实现无人艇在复杂海洋环境下的自主航行、群体决策和博弈对抗等目标打开新思路,有望为我国智能无人艇的工程实践落地提供技术支持。
The development status of unmanned craft is investigated in detail,the bottleneck of autonomous intelligence is analyzed,and the two fundamental problems that restrict the landing of intelligent unmanned craft are revealed,the lack of unmanned craft platform and the poor ability of intelligent game confrontation.To solve these two problems,combined with digital twin technology,the design scheme of swarm intelligence research platform for surface unmanned craft,and also proposes the feasibility assumption of multi-agent game confrontation is present.based on deep learning,which opens a new idea for autonomous navigation,group decision making and game confrontation of unmanned craft in complex Marine environment.It is expected to provide technical support for the engineering practice landing of intelligent unmanned craft.
作者
平洋
刘文斌
缪正元
葛品
黄琮凯
庄正浩
PING Yang;LIU Wenbin;MIAO Zhengyuan;GE Pin;HUANG Congkai;ZHUANG Zhenghao(Institute of War,Academy of Military Science,Beijing 100091,China;Digital Intelligence Technology Research Center,Nanhu Laboratory,Jiaxing 314000,Zhejiang,China)
出处
《船舶工程》
CSCD
北大核心
2023年第2期61-69,共9页
Ship Engineering
关键词
无人艇
数字孪生
强化学习
群体博弈对抗
unmanned surface vehicle
digital intelligence twin
reinforcement learning
group game confrontation