交通微观仿真中的驾驶员视觉感知模型  

Drivers’ Visual Perception Model in Traffic Microscopic Simulation

作　　者：杨建国[1] 肖永剑[1] 王兆安[1] 

机构地区：[1]西安交通大学电气工程学院,西安710049

出　　处：《系统仿真学报》2005年第10期2437-2441,共5页Journal of System Simulation

摘　　要：驾驶员视觉感知模型是交通微观仿真中机动车模型的一部分,其作用是表现出实际驾驶活动中,驾驶员对外界的视觉感知能力,也就是哪些物体可以被驾驶员看到,哪些不能。通过驾驶员视觉关注焦点的移动规律、视场范围和障碍物遮挡等3个方面的研究,提出了一个新的驾驶员视觉感知模型,并注意了该模型的运算速度问题。通过计算机模拟实际场景,对比受试者和模型标注的感知对象,完成了模型有效性验证。对10位受试者进行了194次测试,受试者标注了528个目标,而感知机模型标注了516个目标,其中吻合率为84.30%,漏检率17.61%,误检率15.34%。对出现差异的情况进行细致地分析,可以发现这些差异全部来源于受试者的感觉误差,而且受试者在看到视觉感知模型的输出后,没有一例表现出异议。此外,该模型在避免以往的理想全局观察模式的带来的缺点的同时,还能使模型的计算量大为减少。Driver＇s visual perception model is one part of the vehicle models in traffic microscopic simulation. And the purpose of the visual perception model is to represent the visual perception ability of the real driver, which is to answer what they can see and what they can not. Anew driver＇s visual perception model was introduced after carefully studying on the movement of driver＇s visual focus, the visual field, and obstacle shroud. And special attention was paid to the calculation ability of this model. In order to validate the validity of this model, a contrastive experiment was done by comparing the perception objects of real drivers and the model in a simulated scene, 10 testees were invited to this experiment for 194 tests. The testees marked 528 targets, while the model marked 516 targets, among which 84.30% were accordant, 17.61% were missed, and 15.34% were error-detected. After carefully analyzing those test results one by one, the difference between the model＇s output and the testees＇ selection was found to be the result of human＇s perception error, and after they saw the output of visual perception model, no demurral was advanced. Besides avoiding the limitation caused by the perfect global vision, this model reduced the calculation for the simulation.

关 键 词：交通微观仿真 建模 视觉感知 关注焦点 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]