智能汽车人机共享控制研究综述  

Review on Human-machine Shared Control of Intelligent Vehicles

作　　者：杨俊儒 褚端峰[1] 陆丽萍[2] 王金湘[3] 吴超仲[1] 殷国栋[3] YANG Junru;CHU Duanfeng;LU Liping;WANG Jinxiang;WU Chaozhong;YIN Guodong(Intelligent Transportation Systems Center,Wuhan University of Technology,Wuhan 430063;School of Computer Science and Artificial Intelligence,Wuhan University of Technology,Wuhan 430063;School of Mechanical Engineering,Southeast University,Nanjing 211189)

机构地区：[1]武汉理工大学智能交通系统研究中心,武汉430063 [2]武汉理工大学计算机与人工智能学院,武汉430063 [3]东南大学机械工程学院,南京211189

出　　处：《机械工程学报》2022年第18期31-55,共25页Journal of Mechanical Engineering

基　　金：国家自然科学基金(52172393);湖北省重点研发计划(2020BAB096);重庆市工程研究中心开放课题(21AKC44)资助项目。

摘　　要：智能汽车人机共享控制由人类和机器共同完成驾驶任务,通过人机智能混合增强,保障行车安全,提升驾驶性能。对当前人机共享控制的研究现状及其概念进行梳理;从客观风险评估指标和考虑驾驶员因素论述人机共享控制权决策方法,分析直接式和间接式2种共享控制方式的特点和应用范围,讨论5种共享控制方法的优点和局限性,并总结人机共享控制性能评价指标;指出人机共享控制存在的问题和未来研究方向。分析表明,将机器学习与现有的基于模型的方法相结合,综合考虑驾驶员信息、车辆状态、动态环境对行车风险影响,是未来人机共享控制的研究方向。此外,建立健全自动驾驶预期功能安全测试标准和评价体系,通过预期功能安全认证,保障行车安全,是商业化应用的关键。A shared control of intelligent vehicles consists of a human driver and an automation to complete the driving task. Through human-machine intelligence hybrid enhancement, driving safety is guaranteed and driving performance is improved. According to the current research status of human-machine shared control, the concept is mentioned. The decision-making method of human-machine shared control can be divided into objective risk assessment indicators and driver factors. The characteristics and application scope of direct and indirect shared control frameworks are analyzed. Five shared control methods are reviewed, and the advantages and limitations of each control strategy are discussed. The evaluation indexes of human-machine shared control system are summarized.Finally, the existing problems and future research direction of human-machine shared control are pointed out. The analysis shows that the future research direction is to combine machine learning with existing model-based methods and comprehensively consider the impact of driver information, vehicle state and dynamic environment on driving risk. Simultaneously, the safety test standards and evaluation systems for the safety of the intended functionality of autonomous driving should be established. It is the key of commercial application to ensure driving safety through safety certification of the intended functionality.

关 键 词：智能汽车 人机共享控制 控制权决策 共享控制方法 性能评价 

分 类 号：U46[机械工程—车辆工程]