基于手机打车软件使用的驾驶分心行为研究

发布时间：2019-05-28 03:50

【摘要】：随着道路运输事业的快速发展,各国机动车保有量迅速增加,道路交通事故数量与日俱增。然而90%的道路交通事故与驾驶员因素有关,其中驾驶分心往往是造成交通事故的主要诱因。近年来,打车软件的使用人数飞速增长,驾驶中操作打车软件的分心现象也愈来愈普遍,由此诱发的驾驶安全问题也备受关注。因此,本文对驾驶中操作手机打车软件诱发的驾驶分心行为进行研究,建立驾驶分心检测模型,对改善道路交通安全状况,降低交通事故发生频率具有重大意义。本文首先介绍了驾驶安全与驾驶分心行为的关系,阐明驾驶分心行为是引发交通事故,导致驾驶安全问题的重要原因。系统梳理了手机打车软件使用特点,以《中华人民共和国道路交通安全法实施条例》等法律为依据,结合本校驾驶模拟器的实际情况,筛选出本文研究的四类分心操作:蓝牙通话、与乘客交谈、查看信息、信息输入,手机支架的摆放位置为仪表台。随后,在西南交通大学自主研发的大型驾驶模拟器上完成了基于手机打车软件使用的驾驶模拟实验。对实验数据进行分析,结果表明:常规场景中,有油门开度、纵向速度、纵向加速度、横向加速度、方向盘转角、方向盘转角角速度六项数据标准差在正常驾驶与四类分心驾驶情况下表现出显著差异,同时驾驶员个体差异对六项指标数据结果没有显著影响;非常规场景中,驾驶员在进行驾驶分心操作时,其对危险刺激的反应时显著增长,同时驾驶员个体差异也会对反应时的结果造成显著影响。驾驶员反应时的变化会直接导致避撞方式的改变,但是通过改变避撞方式的补偿行为并未降低交通事故率,驾驶员进行驾驶分心操作时的事故率都显著上升。最后,将驾驶状态分成了三类,即正常驾驶状态、认知分心驾驶状态、视觉分心驾驶状态;基于车辆横纵向控制指标差异性分析结果,构造了驾驶分心检测参数集;运用支持向量机理论搭建了驾驶分心检测模型。随后,随机选取训练样本集和测试样本集对模型进行训练与验证,研究结果显示:该模型对正常驾驶状态、认知分心驾驶状态、视觉分心驾驶状态的平均正确检测率较高,分别为82.5%、80%、92.5%;将视觉分心驾驶状态与认知分心驾驶状态误判为正常驾驶状态的概率都低于8%;模型总体检测效果较好,可用于驾驶分心检测。
[Abstract]:With the rapid development of road transportation, the number of motor vehicles in various countries is increasing rapidly, and the number of road traffic accidents is increasing day by day. However, 90% of road traffic accidents are related to driver factors, in which driving distraction is often the main cause of traffic accidents. In recent years, the number of people using ride-hailing software has increased rapidly, and the distractions of driving software are becoming more and more common, and the driving safety problems caused by this phenomenon have also attracted much attention. Therefore, this paper studies the driving distraction behavior induced by mobile phone hailing software in driving, and establishes the driving distraction detection model, which is of great significance to improve the road traffic safety and reduce the frequency of traffic accidents. This paper first introduces the relationship between driving safety and driving distraction behavior, and expounds that driving distraction behavior is an important cause of traffic accidents and driving safety problems. This paper systematically combs the characteristics of the use of mobile phone ride-hailing software, based on the regulations of the people's Republic of China on the implementation of the Road Traffic Safety Law and other laws, combined with the actual situation of the driving simulator in our school. Four kinds of distractions studied in this paper are screened out: Bluetooth conversation, talking to passengers, viewing information, information input, and the location of mobile phone bracket is the dashboard. Then, the driving simulation experiment based on mobile phone ride hailing software is completed on the large driving simulator developed by Southwest Jiaotong University. The experimental data are analyzed. The results show that there are throttle opening, longitudinal velocity, longitudinal acceleration, transverse acceleration and steering wheel angle in the conventional scene. The standard deviation of the six data of steering wheel angular speed showed significant difference between normal driving and four types of distracted driving, and the individual difference of driver had no significant effect on the results of the six indexes. In the unconventional scene, when the driver is distracted, his response to dangerous stimuli increases significantly, and the individual difference of the driver also has a significant impact on the response time results. The change of driver's reaction time will directly lead to the change of collision avoidance mode, but the compensation behavior of changing collision avoidance mode does not reduce the traffic accident rate, and the accident rate of driver when driving distraction operation is significantly increased. Finally, the driving state is divided into three categories, namely, normal driving state, cognitive distracted driving state and visual distracted driving state. Based on the difference analysis results of vehicle horizontal and longitudinal control indexes, the parameter set of driving distraction detection is constructed. Based on the theory of support vector machine (SVM), a driving distraction detection model is built. Then, the training sample set and the test sample set are randomly selected to train and verify the model. The results show that the average correct detection rate of the model for normal driving state, cognitive distracted driving state and visual distracted driving state is high. It was 82.5%, 80% and 92.5% respectively. The probability of misjudging both visual distracted driving state and cognitive distracted driving state to normal driving state is less than 8%. The overall detection effect of the model is good and can be used for driving distraction detection.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.254

【相似文献】