多车道复杂环境下驾驶倾向性状态辨识  

Dynamic Recognition of Driver's Propensity under Multi-lane Complex Conditions

作　　者：王克刚[1] 王健[2] 王晓原[1] 王晓龙[1] 张敬磊[1] 孙亮[1] 

机构地区：[1]山东理工大学交通与车辆工程学院,山东淄博255091 [2]山东理工大学理学院,山东淄博255091

出　　处：《交通信息与安全》2014年第5期130-137,共8页Journal of Transport Information and Safety

基　　金：国家自然科学基金项目(批准号:61074140);山东省自然科学基金项目(批准号:ZR2010FM007;ZR2011EEM034)

摘　　要：驾驶倾向性是汽车行驶过程中操控者情感偏好等特征的动态测度,是车辆安全驾驶辅助系统,特别是其碰撞预警系统中汽车驾驶人意图等心理、意识计算必须考虑的核心参数。以3车道场景为例,分析目标车位于不同车道时周边的交通态势(主要指车辆集群编组关系,重点以目标车位于中间车道为例),设计实验采集人、车、环境等相关动态信息,获取不同态势下驾驶倾向性特征数据,利用动态贝叶斯网络建立时变环境下驾驶倾向性动态辨识模型。实验验证表明,所建模型对驾驶人倾向性类型的辨识准确率可达到91%以上,能够适应多车道情况下驾驶人倾向性类型的动态识别,为以人为中心的个性化汽车主动安全系统的实现提供理论基础。Vehicle-driving propensity is the dynamic measure of driver＇s emotional preference characteristics in driv- ing process. It is a core parameter to compute driver＇s intention, consciousness in safety driving assisted systems, espe- cially vehicle collision warning systems. With three-lane condition as an example, the traffic situation （considering the ve- hicle groups, and focusing on the target vehicle in the middle lane here）, is analyzed when the target vehicle runs in differ- ent lanes. Experiments are designed to collect the dynamic information related to the driver, vehicle, environment and so on, and to get characteristic data o~ drivels propensity in different situations. Then, a dynamic recognition model of driv- er＇s propensity is established in time varying environment through Dynamic Bayesian Network （DBN）. The identification accuracies of driver＇s propensity types are all above 91%. Experimental results indicate that the model presented is adapta- ble to realize the dynamic recognition of driver＇s propensity type in multilane conditions, which provides a theoretical basis for realizing the human-centered and personalized automobile active safety systems.

关 键 词：情感辨识 驾驶倾向性 动态贝叶斯网络 时变环境 车辆编组关系 

分 类 号：U491[交通运输工程—交通运输规划与管理]