A Multi-Agent System Architecture for Distributed Air Traffic Control
A Multi-Agent System Architecture for Distributed Air Traffic Control
复制标题
分布式空中交通管制的多代理系统架构
DOI：
--
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
R. Frost
中科院分区：
文献类型：
--
作者：
J. Hill;J. Archibald;W. Stirling;R. Frost
下一代空中交通管制将需要自动化，以满足在空中交通密度不断增加的环境中对安全性、可靠性、灵活性和稳健性的要求。然而，对于协同式分布式空中交通管制系统的设计来说，优化是一个不充分的范例。这个问题源于冯·诺伊曼-摩根斯坦公用事业的一个根本局限性，该公用事业没有考虑到情景利他主义等复杂的社会行为。社会效用函数克服了这一局限，允许决策者通过有条件的效用将他们的兴趣范围扩大到超越自我。满意博弈提供了一种决策策略，允许决策者为了实现个人和群体目标而妥协，并为复杂的合作多智能体社会的设计提供了一个数学框架。在各种几何场景下的仿真结果表明，即使在高流量密度的情况下，该算法在效率和灵活性方面也具有良好的性能。
The next generation of air traffic control will require automation in order to meet safety, reliability, flexibility, and robustness demands in an environment of steadily increasing air traffic density. Optimization, however, is an inadequate paradigm for the design of a cooperative distributed air traffic control system. The problem stems from a fundamental limitation of von Neumann-Morgenstern utilities, which do not account for sophisticated social behavior such as situational altruism. Social utility functions overcome this limitation by permitting decision makers to expand their spheres of interest beyond the self via conditional utilities. Satisficing game theory provides a decision strategy that permits decision makers to compromise in the interest of achieving both individual and group goals and presents a mathematical framework for the design of sophisticated cooperative multiagent societies. Simulation results in a variety of geometric scenarios show promising performance in terms of efficiency and flexibility even with high traffic densities.