Safety in Multi-Agent Systems: Reputation Based on Dossier.

The Florida AI Research Society(2015)

引用 0|浏览19
暂无评分
摘要
A multi-agent architecture possesses versatility when solving complex problems in various research fields. In open systems, in which agents are unknown to each other a priori, their trust relationship is an essential subject. Hence, we present a reputation model based on dossier, which grants authenticity and reliability of exchanged information to its participants. This model has been applied to a classical problem in the railway sector, e.g., to define economical driving plans for freight trains. At least three agents are placed for each station, to plan, operate and manage plans. Planning and management skills are conducted by stationary agents, and the operating skills are conducted by mobile agents. Each locomotive has in its on-board computer a container for accommodating agents and transport mobile agents, either to complete an assignment or return to the station of origin. Empirical results show that sharing experiences improves the efficiency of the generation of plans, and that the use of the reputation based on dossier guarantees the veracity of the testimonies, therefore decreasing the need to centralize the information.
更多
查看译文
关键词
safety,reputation,systems,multi-agent
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要