摘要
自动驾驶测试场景具有无限丰富、极其复杂、不可穷尽的特点,传统的测试手段难以对自动驾驶系统安全性进行较为全面的评估。仿真测试具有效率高、成本低、测试过程安全等特点,现已成为自动驾驶测试和验证的重要手段,测试和量化自动驾驶系统的可靠性,理论上需要完成所有场景的测试,但是完成对所有的场景的穷举和测试并不现实。基于仿真测试场景,利用高效的响应面构造方法及抽样方法对自动驾驶系统进行评估为研究自动驾驶系统的可靠性提供了一种新的方向和思路。本文基于一汽集团某车型自动紧急制动(AEB)算法,搭建仿真测试系统,建立了直道CCRm仿真场景,并以前方匀速行驶车辆追尾测试(CCRm)为例,运用克里金法建立了场景变量与输出之间的高质量响应面,通过方差法量化了场景变量的敏感度;并基于合肥某条高架道路车速分布,结合拉丁超立方抽样,得到直道CCRm场景下自动紧急制动系统作用时的两车最小相对距离的分布以及自动紧急制动系统的失效概率,研究结果可为自动紧急制动系统评估提供有力支撑。
Autonomous driving test scenarios are infinitely rich, extremely complex and inexhaustible,it is difficult for traditional testing methods to comprehensively evaluate the safety of autonomous driving system. Simulation test has the characteristics of high efficiency, low cost and safe test procedure, and has become an important means for autonomous driving test and verification, to test and quantify the reliability of the autonomous driving system, in theory, it is necessary to test all the scenarios, but it is not realistic to exhaust and test all scenarios. Based on the autonomous emergency braking(AEB) algorithm of a certain car, the simulation test system is built, and straight CCRm simulation scenarios are established. This paper takes the Car-to-Car Rear Moving test(CCRm)simulation scenario as example, the Kriging method is used to establish high-quality response surface between scenario variables and system output, and the sensitivity of the scenario variables is quantified through the variance method. Based on the speed distribution of an elevated highway in Hefei, combined with Latin hypercube sampling, the distribution of the minimum relative distance between two vehicles when the autonomous emergency braking system acts under the straight CCRm scenario and the failure probability of the autonomous emergency braking system are obtained. The research results can provide strong support for the evaluation of the autonomous emergency braking system.
作者
刘涛
周忠贺
迟霆
LIU Tao;ZHOU Zhonghe;CHI Ting(Intelligent Connected Vehicle Development Institute,China FAW Group Company Limited,Changchun 130013,China)
出处
《汽车实用技术》
2022年第24期33-39,共7页
Automobile Applied Technology
