基于自然驾驶数据的危险评估算法研究  

A Study on Risk Assessment Algorithm Based on Natural Driving Data

作　　者：吴斌[1,2] 朱西产[1] 沈剑平[2] 李霖 

机构地区：[1]同济大学汽车学院,上海201804 [2]国家机动车产品质量监督检验中心(上海),上海201805 [3]上海国际汽车城(集团)有限公司,上海201805

出　　处：《汽车工程》2017年第8期907-914,共8页Automotive Engineering

基　　金：上海市科委研发平台项目(16DZ2291000)资助

摘　　要：本文中提出了一种基于自然驾驶数据的汽车碰撞危险评估算法。首先分析了在自然驾驶工况下的驾驶员制动和转向的输入特性,研究了表征驾驶员的制动避撞和转向避撞紧急程度的行为特征参数,建立了紧急制动和紧急转向的驾驶员数学模型;在此基础上,分析了不同车速下制动避撞和转向避撞的特征,提出了基于自然驾驶数据的危险评估算法。仿真结果表明,本文中提出的危险评估算法可同时兼顾制动避撞和转向避撞,并与驾驶员在真实交通环境下的紧急避撞行为相对应,可用于汽车避撞系统控制策略的开发。A vehicle collision risk assessment algorithm based on natural driving data.is proposed in this paper.Firstly the driver input characteristics for braking and steering under natural driving condition are analyzed,the feature parameters of driver behavior characterizing the emergency extent of collision avoidance by braking and by steering are studied,and the driver models for emergency braking and emergency steering are established.Then on this basis,the features of collision avoidance by braking and by steering are analyzed.Finally a risk assessment algorithm based on natural driving data is proposed.The simulation results show that the risk assessment algorithm proposed can concurrently consider collision avoidance by both braking and steering and correspond to emergency collision avoidance behavior of driver under real traffic environment,so can be used for the development of control strategy for vehicle collision avoidance system.

关 键 词：自然驾驶数据 紧急避撞 驾驶员行为 危险评估 

分 类 号：U463.6[机械工程—车辆工程]