汽车工程学报2025，Vol.15Issue(5)：707-717,11.DOI:10.3969/j.issn.2095-1469.2025.05.07
基于实车关键场景的自动驾驶前向视觉感知测评
Forward Visual Perception Evaluation for Autonomous Driving Based on Critical Real-Vehicle Scenarios
摘要
Abstract
Existing methods for evaluating visual perception performance in autonomous driving systems fail to accurately reflect algorithm behavior in real-world scenarios and lack targeted test data.Therefore,this paper proposes a forward visual perception evaluation method based on critical scenarios in real vehicles.Firstly,critical scenarios are defined according to the operational design domain(ODD)of the autonomous driving system.Then,a data collection platform is built,and a scenario-extraction method is used to extract the key scenario test segments.Finally,a designed evaluation tool is used to analyze the perception performance of the system under test and generate a quantitative evaluation report.Using the adaptive cruise control as an example,the method evaluates two intelligent front-view perception systems across different scenarios and evaluation intervals.The experimental results confirm the effectiveness of the proposed method.关键词
感知测评/关键场景/自动驾驶/实车测评Key words
perception evaluation/key scenarios/autonomous driving/real-vehicle evaluation分类
交通工程引用本文复制引用
高嵩,高博麟,王鹤,李润泽,张家峰,徐月云..基于实车关键场景的自动驾驶前向视觉感知测评[J].汽车工程学报,2025,15(5):707-717,11.基金项目
国家自然科学基金项目(52172389) （52172389）
