摘要
自主泊车系统是智能汽车领域的研究热点,但其缺少系统性测试评价方法的研究。针对这一问题:首先,提出了基于ALFUS(Autonomy Levels for Unmanned System)框架思想的测试用例构建体系,结合对由泊车事故数据、实际泊车场景数据构成的实际数据以及包含传感器工作原理和泊车过程的自主泊车系统工作原理的分析,生成了自主泊车系统测试环境元素集合、测试任务元素集合、两级测试时空顺序,并通过对三者进行整合得到了自主泊车系统测试用例集;其次,提出了以环境类型、任务类型进行划分的自主泊车系统三级四类分级测试方法,并在此基础上结合模糊综合评价法得到了基于自主泊车系统测试用例集的测试评价方法;然后,利用乔哈里视窗理论建立了自主泊车系统测试评价结果分析矩阵,实现了从智能汽车自主泊车系统性能局限、测试方法局限两方面对测试评价结果的分析;最后,考虑现有自主泊车系统技术条件,选用不同型号的具备自主泊车系统的智能汽车进行实车试验,对上述测试方法进行了实践,并分析获得了自主泊车系统功能及性能方面的改进意见。试验结果表明:所提智能汽车自主泊车系统测试方法不仅可对测试对象的性能提出改进建议,也能验证测试方法本身的完善性,并证明了该测试方法对于现阶段自主泊车系统的有效性。
As a hot topic of research,the automated valet parking system lacks methods necessary for systematic testing and evaluation.To address these issues,this work conducted a series of studies.First,a framework for a test scenario based on an autonomy levels for unmanned system (ALFUS) was proposed.Under this framework,a test scenario of an automated valet parking system was studied through integrated environment elements,task elements,and a spatialtemporal order,which was derived from actual parking scenario data and the working principle of a valet parking system.Then,a test and evaluation method based on the scenario was proposed using a fuzzy-synthetic evaluation and a test clustering method.The latter was divided into four types from simplicity to complexity based on different environments and tasks.Then,referring to Johari window theory,a matrix was established for analyzing the test results in terms of both the merits and demerits of the automated valet parking system and the test and evaluation method itself.Finally,based on a series of real vehicle tests,an application example of a full set of the test and evaluation method is presented ,in which some suggestions for improving both the valet parking system and the test and evaluation method are presented.Experimental results indicate that this study not only improves the performance of test objects but also verifies that the perfection of the test method itself,which is effective for use with current automated valet parking systems.
作者
蒙昊蓝
陈君毅
左任婧
余卓平
陈磊
MENG Hao-Ian;CHEN Jun-yi;ZUO Ren-jing;YU Zhuo-ping;CHEN Lei(School of Automotive Studies,Tongji University,Shanghai 201804,China;Chinese-German College,Tongji University,Shanghai 201804,China)
出处
《中国公路学报》
EI
CAS
CSCD
北大核心
2019年第6期158-168,共11页
China Journal of Highway and Transport
基金
国家重点研发计划项目(2018YFB0105101)
关键词
汽车工程
自主泊车系统
乔哈里视窗理论
测试方法
测试用例
分级测试
automotive engineering
automated valet parking system
Johari window theory
test method
test scenario
test clustering method
作者简介
蒙昊蓝(1993-),男,重庆市人,工学博士研究生,E-mail:hmeng@tongji.edu.cn;通讯作者:陈君毅(1980-),女,上海市人,讲师,工学博士,E-mail:chenjunyi@tongji.edu.cn。
