机构地区: 中国科学院研究生院
出 处: 《模式识别与人工智能》 2004年第4期405-411,共7页
摘 要: 提出一种分布式多主体规划算法.在此算法中,主体规划之间的冲突检测与协调通过多主体协商来解决,即主体规划之间约束一致性的分布式判断.在确定环境中该算法是可靠的.由于算法中主体之间交换的只是与冲突有关的动作、因果链和约束.因此具有通信量小、安全性高的优点. This paper presents a distributed multi-agent planning algorithm based on constraints propagating. In our algorithm, conflicts are detected and resolved through a special type of negotiation among agents, i. e. consistency checking for constraints. The algorithm is sound under deterministic circumstance. In our algorithm, agents only exchange information about actions, casual links and constraints related to conflicts. It is of less communication, higher security than that exchanging the whole plan.
领 域: [自动化与计算机技术] [自动化与计算机技术]