基于多和驾驶员行为的汽车追尾预警模型基于多和驾驶员行为的汽车追尾预警模型.pdf

2009海峡两岸智能型 运输系统学术研讨会 基于多Agent和驾驶员行为的汽车追尾预警模型 Rear-end CollisionAlarmModeIandSimulationBasedon and Behavior Multi—AgentDriving Junl 梁军Liang 姚明YaoMin92 摘要 现有汽车追尾预警系统以车间距离测量作为碰撞发生的唯一决定因素，针对这个不足之处提 出了以MAS为架构，以驾驶员行为为核心，以Bayes决策为手段的基于多Agent和驾驶员行为的 CarRear-end Basedon and 汽车追尾预警模型(Modelingof warning Multi-AgentSystems 性和实时性，体现道路交通中驾驶员地位的主体性，以及追尾预警各方面信息的不确定性，采用 基于Bayes决策的追尾预警算法计算碰撞概率实现实时预防追尾碰撞事故发生。仿真结果表明 MCRMASB模型在系统预警的自主性和抗干扰方面比传统的预警系统有优势。 关键词：追尾预警、行为、agent技术、仿真 Abstract Aimtothe betweenvehiclesmeasurement船the Shortcomingsofdistance onlydeterminingfactor inCarRear-endAlarm decision model rear-end system．砂mear坫ofBayestheory,aofcar warning basedonMilSand the decision are behavior(MCRMASB)andcorrespondingalgorithmproposed wh{chis onDBandis MAS口sits aimistomakeu哼e the centering having framework．The full of characteristicsthe inMAS as of agents such andreal-timeto driver’ interaction,autonomyembody inroad andto the in subjectivity this tra．fflc improveuncertainO,ofinformationeveryaspeaofnCAS．In current to anewidea the RCAS．Thismodels w。晴the research耐tempts for studyof provide androbustnesshavebeen thesimulated effectiveness confirmedby experiments． 转4