自动驾驶汽车场景测试与评估体系：研究现状、挑战及趋势

范博; 周重位; 张思楠; 杨军; 陈艳艳; 李同飞

doi:10.3963/j.jssn.1674-4861.2025.06.002

自动驾驶汽车场景测试与评估体系：研究现状、挑战及趋势

doi: 10.3963/j.jssn.1674-4861.2025.06.002

北京工业大学城市交通学院北京 100124

基金项目:

国家自然科学基金项目 61901013

详细信息

作者简介:
范博（1989—），博士，副教授. 研究方向：车联网与智能交通系统. E-mail：fanbo@bjut.edu.cn

通讯作者:
李同飞（1990—），博士，教授. 研究方向：交通信息工程及控制、自动驾驶专用道规划等. E-mail：tfli@bjut.edu.cn

中图分类号: U461.99
计量
- 文章访问数: 24
- HTML全文浏览量: 11
- PDF下载量: 3
- 被引次数: 0
出版历程
- 收稿日期: 2025-06-15
- 网络出版日期: 2026-03-13

Scenario-based Testing and Evaluation Systems for Autonomous Vehicles: Research Status, Challenges, and Trends

College of Metropolitan Transportation, Beijing University of Technology, Beijing 100124, China

摘要

摘要: 随着自动驾驶技术加速向规模化测试与商业化应用过渡，构建系统性的测试场景与评价指标体系已成为保障其安全落地的核心前提。本文对自动驾驶汽车测试场景构建与评价指标体系的研究现状、面临挑战及未来趋势进行了综述。发现在面对车路云一体化架构与动态混合交通流带来的复杂性，传统“里程-失效”统计模式已难以满足全链条性能评估需求。在测试场景体系方面，概述了测试范式向“场景驱动”演变的历程，总结了基于ISO 34501标准及PEGASUS六层模型的场景语义描述方法与主流生成技术，并指出当前体系主要存在长尾与边缘场景覆盖不足、标准规范碎片化严重、多维量化标准缺失，以及过分局限于单车智能封闭设计而忽视车辆无线通信技术（vehicle-to-everything，V2X）网联协同要素等问题。在测试指标体系方面，从竞赛型、封闭场地-仿真结合型及理论研究型3个维度对现有评价指标体系进行了归纳，指出当前指标体系在评估自动驾驶汽车利用V2X协同信息的能力方面存在不足、指标覆盖广度有限、评价维度与流程高度离散，以及客观交互体验量化指标缺失等问题。针对上述挑战，下一代测试体系需重点聚焦于以下研究路径：①构建通用的场景描述语言与数据共享框架，确立衡量场景风险关键性与真实性的统一量化基准；②构建涵盖从标称到长尾边界的分层递进场景体系以实现全域工况覆盖；③建立融合通信时延、系统韧性与社会伦理的综合评价指标以完善多维量化基准；④引入世界模型与生成式AI技术，结合因果推理机制模拟高风险极端工况并推演未知失效场景，以深度验证系统的泛化能力。
- 汽车工程 /
- 自动驾驶汽车 /
- 综述 /
- 场景测试 /
- 测试场景体系 /
- 测试指标体系
Abstract: Autonomous driving technology is accelerating toward large-scale testing and commercial application. Consequently, constructing systematic scenario frameworks for testing and robust evaluation metrics is crucial for safe deployment. This paper reviews the research status, challenges, and future trends of these systems. The study analyzes complexities introduced by vehicle-road-cloud integration and dynamic mixed traffic. It finds that traditional"mileage-failure"statistical models are insufficient for end-to-end performance assessment. Regarding test scenarios, the paper outlines the evolution toward scenario-driven paradigms. It summarizes semantic description methods based on the ISO 34501 standard and the PEGASUS six-layer model. Mainstream scenario generation technologies are also reviewed. Current frameworks show insufficient coverage of long-tail and edge scenarios. Standards are highly fragmented. Furthermore, existing frameworks often under-represent vehicle-to-everything (V2X) collaborative elements due to an excessive focus on single-vehicle intelligence. Regarding evaluation metrics, existing methodologies are categorized into three dimensions, including competition-based, closed-track/simulation hybrid, and theory-oriented approaches. The review identifies several deficiencies in current systems. Specifically, current metrics insufficiently assess the use of V2X collaborative information. Evaluation dimensions and workflows are fragmented, and objective quantitative metrics for interactive experience are lacking. To address these challenges, next-generation testing systems should focus on four research paths. ①Unified scenario description languages and data-sharing frameworks are needed to establish benchmarks for measuring scenario risk criticality and realism. ② Hierarchical scenario systems should be built to cover nominal conditions as well as long-tail boundaries for full-domain coverage. ③Comprehensive metrics should integrate communication latency, system resilience, and social ethics. ④World models and generative AI, combined with causal inference, can simulate extreme conditions and explore unknown failure modes to validate the system's generalization capability.
- automotive engineering /
- autonomous vehicles /
- review /
- scenario testing /
- test scenario framework /
- evaluation metric system

HTML全文

表 1 竞赛型评价指标体系

Table 1. Competitive evaluation index system

竞赛名称	赛道场景	常用指标/评分焦点
美国DARPA无人驾驶挑战赛	荒漠或城市全程自主完赛	路线遵循、完赛用时、违规/安全干预次数
美国印地自动驾驶挑战赛	椭圆赛道	平均圈速、超车安全率、信号响应
中国i-VISTA智能汽车挑战赛	封闭场、小规模开放路混合	AEB、FCW、LDW等功能的得分累计
欧盟协同驾驶挑战赛	多车编队协同场景	车辆协作成功率、时延、动态稳定度
F1TENTH自动驾驶竞赛	专用反光地面紧凑赛道	最快单圈、连续无碰撞圈数、碰撞罚时
赛事公司Ro-borace	全尺寸电动赛车赛道	完成度、计时赛净成绩、AI稳定性

下载: 导出CSV

表 2 封闭场地-仿真结合型体系

Table 2. Hybrid closed-track and simulation system

体系名称	场景类型	典型指标类型
瑞典AstaZero测试场	封闭场地+数字孪生	场景通过率、残余风险、关键件日志等
中国i-VISTA	封闭/开放场地+万级场景库	智能安全、智能驾驶、智能交互、智能能效、智能泊车
西班牙IDIADA虚拟试验场	封闭场地+数字孪生	功能覆盖率、VIL故障注人通过率等
匈牙利Zala-ZONE汽车测试试验场	封闭/开放场地+数字孪生	高速工况安全余量、通信时延
韩国K-City自动驾驶测试场	实体为主、仿真配套	场景覆盖率、远程监控响应
美国Mcity测试场	封闭场地+数字孪生	加速评估、行为能力评估和极端评估三阶段的风险评分

下载: 导出CSV

表 3 理论研究类指标体系对比

Table 3. Comparative analysis of theoretical research evaluation indicator system

文献	评价维度/类别	指标维度/数量	主要贡献或特点	主要缺陷
[1]	安全、功能性、通行效率、乘坐舒适性	4维	将概率类量化指标嵌入场景测试，实现对从基础安全到乘客体验的评估	评价视角局限于单车智能、功能性评价二元化，未覆盖舒适度、类人性等过程性指标
[37]	安全、舒适、行驶性能、法律遵循、利他性能	5维14项	OMDCE场景级自适应综合得分，首次引入利他性能指标	存在单车智能测试局限性、法律遵循指标单一，未覆盖感知、安全韧性等关键维度
[38]	安全、智能、体验、能耗、效率	5维62项	多级测试用例驱动的通用指标体系	体验指标主观性较高、缺乏网联协同场景测试，感知、法律遵循等指标缺失
[39]	安全、效率、经济、智能、舒适	13项	主客观因素相结合、定性定量相融合的综合性能评价方法	评价视角局限于单车智能、主观指标过度依赖专家打分，感知、法律遵循等指标缺失
[40]	安全、类人性	2维	指标体系评估与乘员主观体验高度吻合，同时兼顾日常驾驶场景与长尾场景	缺少网联环境性能评估、评估维度单一，舒适度、效率、法规遵循等指标缺失
[41]	技术、用户、交通运行、社会影响	4维	从功能级到系统级的全链路评价，含单车、V2X、交通系统效应、安全、能耗	用户类指标过度依赖主观评价、网联协同能力评估缺失
[42]	车－路通信5项、车辆动态6项	2维11项	联合网联性能与车辆运动行为的综合指标体系	车辆动态指标单一且纵向化，通信安全、舒适、能耗与环保维度缺失
[43]	安全、舒适、效能、经济、智能	5维12项	三层结构化综合评价模型	网联协同能力评估缺失、各维度指标较少、未覆盖法规遵循与类人性等指标
[44]	安全、舒适性、智能性、效率	4维	纯数据驱动的权重与指标评分替代专家打分，实现虚拟测试的客观、自动化评价	网联协同能力评估缺失、未覆盖法规遵循与系统安全韧性等高阶维度、智能和效率指标单一
[45]	路线总分；场景数量、得分均值、标准差、最值	6项	测试场景难度的方法，评估不同自动驾驶汽车测试路线的复杂性。	网联协同能力评估缺失，感知、能耗、法律遵循、社会合规等指标缺失
[46]	安全、系统、平稳、速度性	4维14项	减少了主观性，量化评价自动驾驶汽车的综合智能性能	网联协同能力评估缺失、综合评分依赖专家打分，未覆盖类人性、安全韧性等指标
[47]	风险感知、避险操纵、接管绩效	3维13项	主客观结合、从驾驶人认知到操作的安全评价指标体系	评价维度单一，仅聚焦于安全性，未覆盖乘坐舒适性、能耗经济性、网联协同、法律遵循等指标

下载: 导出CSV

参考文献(47)

[1]	FENG S, FENG Y H, SUN G W, et al. Testing scenario library generation for connected and automated vehicles, part Ⅰ: methodology[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(3): 1573-1582. doi: 10.1109/TITS.2020.2972211
[2]	孙航, 李志军, 张琳琳, 等. 基于OEDR和ODC的自动驾驶汽车实际道路测试评价技术研究[J]. 汽车工程, 2022, 44 (6): 842-850. SUN H, LI Z J, ZHANG L L, et al. Research on the test and evaluation technique of real roads for automated driving vehicles based on OEDR and ODC[J]. Automotive Engineering, 2022, 44(6): 842-850. (in Chinese)
[3]	LEE D H. Efficient perception, planning, and control algorithm for vision-based automated vehicles[J]. Applied Intelligence, 2024, 54(17): 8278-8295.
[4]	ZHANG X R, XIONG L, ZHANG P Z, et al. Real-world troublemaker: a 5G cloud-controlled track testing framework for automated driving systems in safety-critical interaction scenarios[J]. IEEE Internet of Things Journal, 2025, 12(23): 50617-50631. doi: 10.1109/JIOT.2025.3609958
[5]	HUANG T, LIU J, ZHOU X, et al. Vehicle-to-Everything cooperative perception for autonomous driving[R/OL]. (2025-09-02)[2025-10-01]. https://arxiv.org/abs/2310.03525.
[6]	SCHLENOFF C, KOOTBALLY Z, RACHAKONDA P, et al. Standards and performance metrics for on-road automated vehicles[R/OL]. (2024-06-04)[2025-09-13]. https://doi.org/10.6028/NIST.IR.8527.
[7]	SCHOLTES M, WESTHOFEN L, TURNER L R, et al. 6-layer model for a structured description and categorization of urban traffic and environment[J]. IEEE Access, 2021(9): 59131-59147.
[8]	蒋拯民, 党少博, 李慧云, 等. 自动驾驶汽车场景测试研究进展综述[J]. 汽车技术, 2022(8): 10-22. JIANG Z M, DANG S B, LI H Y, et al. A survey on the research progress of scenario-based testing for autonomous vehicles[J]. Automobile Technology, 2022(8): 10-22. (in Chinese)
[9]	SON W, HA Y, OH T, et al. PG-based vehicle-in-the-loop simulation for system development and consistency validation[J]. Electronics, 2022, 11(24): 4073. doi: 10.3390/electronics11244073
[10]	余卓平, 邢星宇, 陈君毅. 自动驾驶汽车测试技术与应用进展[J]. 同济大学学报(自然科学版), 2019, 47(4): 540-547. YU Z P, XING X Y, CHEN J Y. Review on automated vehicle testing technology and its application[J]. Journal of Tongji University(Natural Science), 2019, 47(4): 540-547. (in Chinese)
[11]	MENZEL T, BAGSCHIK G, MAURER M. Scenarios for development, test and validation of automated vehicles[C]. 2018 IEEE Intelligent Vehicles Symposium, Changshu, China: IEEE, 2018.
[12]	朱冰, 范天昕, 赵文博, 等. 自动驾驶汽车连续测试场景复杂度评估方法[J]. 吉林大学学报(工学版), 2025, 55(2): 456-467. ZHU B, FAN T X, ZHAO W B, et al. Continuous test scenario complexity evaluation method for automated driving vehicles[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(2): 456-467. (in Chinese)
[13]	CROSATO L, TIAN K, SHUM H, et al. Social interaction-aware dynamical models and decision-making for autonomous vehicles[R/OL]. (2023-10-31)[2025-10-01]. https://doi.org/10.48550/arXiv.2310.18891.
[14]	孙乐乐, 黄松, 郑长友, 等. 自动驾驶仿真测试场景生成技术研究进展[J]. 计算机工程与应用, 2025, 61(1): 59-79. SUN L L, HUANG S, ZHENG C Y, et al. Research progress on autonomous driving simulation test scenario generation technology[J]. Computer Engineering and Applications, 2025, 61(1): 59-79. (in Chinese)
[15]	赵祥模, 承靖钧, 徐志刚, 等. 基于整车在环仿真的自动驾驶汽车室内快速测试平台[J]. 中国公路学报, 2019, 32(6): 124-136. ZHAO X M, CHENG J J, XU Z G, et al. An indoor rapid-testing platform for autonomous vehicle based on vehicle-in-the-loop simulation[J]. China Journal of Highway and Transport, 2019, 32(6): 124-136. (in Chinese)
[16]	THORN E, KIMMEL S, CHAKA M. A framework for automated driving system testable cases and scenarios[R]. Washington D C: National Highway Traffic Safety Administration, 2018.
[17]	ISO. Road vehicles-test scenarios for automated driving systems-Vocabulary: ISO 34501[S]. Geneva, Switzerland: ISO, 2022.
[18]	熊璐, 吴建峰, 冯天悦, 等. 面向自动驾驶逻辑场景评价的危险域识别方法[J]. 中国公路学报, 2024, 37(5): 371-382. XIONG L, WU J F, FENG T Y, et al. Danger field identification method for evaluation of autonomous driving logic scenarios[J]. China Journal of Highway and Transport, 2024, 37 (5): 371-382. (in Chinese)
[19]	WINNER H, LEMMER K, FORM T, et al. PEGASUS— First steps for the safe introduction of automated driving[C]. Road Vehicle Automation 5, Cham, Switzerland: Springer, 2019.
[20]	ISO. Road vehicles—safety of the intended functionality: ISO/DIS 21448[S]. Geneva, Switzerland: ISO, 2019.
[21]	李鹏辉, 董倩茹, 袁赫男, 等. 面向自动驾驶仿真测试的高覆盖切入场景库生成方法[J]. 中国公路学报, 2024, 37(7): 237-249. LI P H, DONG Q R, YUAN H N, et al. High-coverage cut-in scenario library generation for automated driving simulation testing[J]. China Journal of Highway and Transport, 2024, 37 (7): 237-249. (in Chinese)
[22]	刘法旺, 曹建永, 张志强, 等. 基于场景的智能网联汽车"三支柱"安全测试评估方法研究[J]. 汽车工程学报, 2023, 13 (1): 1-7. LIU F W, CAO J Y, ZHANG Z Q, et al. A scenario-based "three-pillar"safety testing and assessment method for intelligent and connected vehicles[J]. Chinese Journal of Automotive Engineering, 2023, 13(1): 1-7. (in Chinese)
[23]	赵祥模国家重点研发计划(2021YFB2501200)团队. 自动驾驶测试与评价技术研究进展[J]. 交通运输工程学报, 2023, 23(6): 10-77. ZHAO X M Team of National Key Research and Development Program of China(2021YFB2501200). Research progress in testing and evaluation technologies for autonomous driving[J]. Journal of Traffic and Transportation Engineering, 2023, 23(6): 10-77. (in Chinese)
[24]	庞劭荣, 张诗波, 罗龙浩, 等. 基于改进云组合赋权的智能网联驾驶场景安全性评价方法[J]. 交通信息与安全, 2023, 41(5): 35-42. doi: 10.3963/j.jssn.1674-4861.2023.05.004 PANG S R, ZHANG S B, LUO L H, et al. A method for evaluating safety of driving scenes with intelligent connected vehicles based on an improved cloud combination weighting[J]. Journal of Transport Information and Safety, 2023, 41 (5): 35-42. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.05.004
[25]	FINKELDEI F, THEES C, WEGHORN J N, et al. Scenario factory 2.0: scenario-based testing of automated vehicles with commonroad[J]. Automotive Innovation, 2025, 8(2): 207-220. doi: 10.1007/s42154-025-00360-0
[26]	ZHANG X, KHASTGIR S, TIELE J K, et al. ODD and behavior based scenario generation for automated driving systems[J]. IEEE Access, 2024, 12: 10652-10663. doi: 10.1109/ACCESS.2024.3350512
[27]	FENG S, YAN X, SUN H, et al. Intelligent driving intelligence test for autonomous vehicles with naturalistic and adversarial environment[R/OL]. (2021-02-02)[2025-10-01]. https://doi.org/10.1038/s41467-021-21007-8.
[28]	MORADLOO N, MAHDINIA I, KHATTAK A J. Safety in higher level automated vehicles: investigating edge cases in crashes of vehicles equipped with automated driving systems[J]. Accident Analysis & Prevention, 2024, 203: 107607.
[29]	杜德慧, 叶振, 郑成行, 等. 面向自动驾驶系统的场景建模及边缘关键场景生成[J]. 软件学报, 2025, 36(8): 3512-3530. DU D H, YE Z, ZHENG C H, et al. Scenario modeling and edge-critical scenario generation for autonomous driving system[J]. Journal of Software, 2025, 36(8): 3512-3530. (in Chinese)
[30]	LIN Y, RATZEL M, ALTHOFF M. Automatic traffic scenario conversion from OpenSCENARIO to commonroad[C]. IEEE 26^th International Conference on Intelligent Transportation Systems, Bilbao, Spain: IEEE, 2023.
[31]	WEI Z, ZHOU H, ZHOU R. Risk and complexity assessment of autonomous vehicle testing scenarios[J]. Applied Sciences, 2024, 14(21): 9866. doi: 10.3390/app14219866
[32]	YAN S, ZHANG X, HAO K, et al. On-demand scenario generation for testing automated driving systems[R/OL]. (2025-05-25)[2025-06-11]. https://doi.org/10.48550/arXiv.2505.14053.
[33]	刘刚, 宋健. EMB硬件在环仿真试验台[J]. 汽车工程, 2006(10): 929-932. LIU G, SONG J. Hardware-in-the-loop test bench for electromechanical brake system[J]. Automotive Engineering, 2006 (10): 929-932. (in Chinese)
[34]	HAYWARD J C. Near-miss determination through use of a scale of danger[J]. Highway Research Record, 1972, 384: 24-34.
[35]	ALLEN B L, SHIN B T, COOPER P J. Analysis of traffic conflicts and collisions[J]. Transportation Research Record, 1978, 667: 67-74.
[36]	HILLENBRAND J, SPIEKER A M, KROSCHEL K. A multilevel collision mitigation approach—its situation assessment, decision making, and performance tradeoffs[J]. IEEE Transactions on Intelligent Transportation Systems, 2006, 7(4): 528-540. doi: 10.1109/TITS.2006.883115
[37]	ZHOU J, WANG L, WANG X. Scalable evaluation methods for autonomous vehicles[J]. Expert Systems with Applications, 2024, 249: 123603. doi: 10.1016/j.eswa.2024.123603
[38]	修海林. 有条件自动驾驶汽车测试与综合评价研究[D]. 重庆: 重庆大学, 2019. XIU H L. Study on the test and comprehensive evaluation of conditional autonomous vehicles[D]. Chongqing: Chongqing University, 2019. (in Chinese)
[39]	WANG W, WU L, LI X, et al. An evaluation method for automated vehicles combining subjective and objective factors[R/OL]. (2023-01-01)[2025-10-01]. https://doi.org/10.3390/machines11060597.
[40]	ZHANG P, ZHU B, ZHAO J, et al. Performance evaluation method for automated driving system in logical scenario[J]. Automotive Innovation, 2022, 5(3): 299-310. doi: 10.1007/s42154-022-00191-3
[41]	ROESENER C, SAUERBIER J, ZLOCKI A, et al. A comprehensive evaluation approach for highly automated driving[C]. 25^th International Technical Conference on the Enhanced Safety of Vehicles (ESV), Detroit, USA: NHTSA, 2017.
[42]	PEREIRA J, PREMEBIDA C, ASVADI A, et al. Test and evaluation of connected and autonomous vehicles in real-world scenarios[C]. 2019 IEEE Intelligent Vehicles Symposium, Paris, France: IEEE, 2019.
[43]	王书辉. 基于仿真的自动驾驶汽车测试评价研究[D]. 西安: 长安大学, 2024. WANG S H. Research on testing and evaluation of autonomous vehicles based on simulation[D]. Xi'an: Chang'an University, 2024. (in Chinese)
[44]	ZHAO S J, WEI Z R, WANG P Y, et al. An objective evaluation method for automated vehicle virtual test[J]. Expert Systems with Applications, 2022, 206: 117940. doi: 10.1016/j.eswa.2022.117940
[45]	LIN D J, LIU H H, CHOU A C, et al. Evaluation of difficulty for autonomous vehicles testing roads based on multiple criteria decision analysis[C]. SAE Technical Paper Series, Detroit, USA: SAE International, 2024.
[46]	李茹, 马育林, 田欢, 等. 基于熵值和G1法的自动驾驶车辆综合智能定量评价[J]. 汽车工程, 2020, 42(10): 1327-1334. LI R, MA Y L, TIAN H, et al. Comprehensive intelligent quantitative evaluation of autonomous vehicle based on entropy and G1 methods[J]. Automotive Engineering, 2020, 42 (10): 1327-1334. (in Chinese)
[47]	李振龙, 潘梦妞, 曲彦菘, 等. 基于IAHP-EWM-LDM的L3级自动驾驶接管过程安全性评价方法[J]. 交通信息与安全, 2023, 41(4): 14-23, 100. doi: 10.3963/j.jssn.1674-4861.2023.04.002 LI Z L, PAN M N, QU Y S, et al. A method for evaluating the safety over the takeover process of the level 3 automated vehicles based on IAHP-EWM-LDM[J]. Journal of Transport Information and Safety, 2023, 41(4): 14-23, 100. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.04.002