汽车工程 ›› 2021, Vol. 43 ›› Issue (8): 1168-1176.doi: 10.19562/j.chinasae.qcgc.2021.08.007

• • 上一篇    下一篇

自动驾驶汽车乘员个性化乘坐舒适性辨识方法

兰凤崇1,李诗成1,陈吉清1(),沈宗卯2   

  1. 1.华南理工大学机械与汽车工程学院,广州 510640
    2.广州小鹏自动驾驶科技有限公司,广州 510640
  • 收稿日期:2021-03-22 修回日期:2021-04-28 出版日期:2021-08-25 发布日期:2021-08-20
  • 通讯作者: 陈吉清 E-mail:chenjq@scut.edu.cn
  • 基金资助:
    国家自然科学基金(51775193);广东省科技计划(2015B01037002);广东省自然科学基金(2018A030313727)

Identification Method for Occupant Personalized Ride Comfort of Autonomous Vehicles

Fengchong Lan1,Shicheng Li1,Jiqing Chen1(),Zongmao Shen2   

  1. 1.School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou 510640
    2.Guangzhou Xiaopeng Motors Technology Co. ,Ltd. ,Guangzhou 510640
  • Received:2021-03-22 Revised:2021-04-28 Online:2021-08-25 Published:2021-08-20
  • Contact: Jiqing Chen E-mail:chenjq@scut.edu.cn

摘要:

针对自动驾驶车辆轨迹规划控制算法无法满足乘员个性化舒适性问题,结合自然驾驶数据和乘员乘坐舒适性需求,建立乘员个性化舒适性辨识方法。首先确定主观舒适性评价方式,基于标准ISO2631搭建频域和时域加权滤波函数,提取自动驾驶汽车乘员舒适性主客观特征参数,辨识乘员个性化舒适性与自动驾驶车辆行驶规划参数关系;随后搭建自然驾驶数采平台,采集影响舒适性的行驶参数和主客观参数;利用因子分析对行驶参数降维,得到三向运动(横向冲击、纵向加速、垂向振动)、行驶风险和效率影响因子;最后运用加权分析方法辨识模型,并通过卡尔曼滤波算法快速准确识别乘员个性化需求,得到舒适度加权方均根阈值。辨识结果表明:乘员主客观舒适度相关性达85.8%;三向运动因子对乘员舒适性影响大于行驶风险和效率因子;乘员个性化舒适性辨识率高达93.9%。本研究可为搭建考虑乘员舒适性的个性化轨迹规划控制算法提供理论支持。

关键词: 自动驾驶, 乘员舒适性, 辨识, 评价, 个性化

Abstract:

For the problem that autonomous vehicle trajectory planning control algorithm cannot meet the personalized comfort of occupants, combining natural driving data and occupant comfort requirements, a method for identifying occupants’ personalized comfort is established. Firstly, the subjective comfort evaluation method is determined. Based on the standard ISO2631, frequency domain and time domain weighted filter functions are built. Subjective and objective characteristic parameters of occupant comfort of autonomous vehicles are extracted and the relationship between occupant’s personalized comfort and autonomous vehicle driving planning parameters is identified. Then, a natural driving data acquisition platform is established to collect the driving parameters and subjective and objective parameters that affect comfort. Factor analysis is used to reduce the dimensions of driving parameters to obtain three?way motion (lateral impact, longitudinal acceleration, and vertical vibration), driving risk and efficiency influencing factors. Finally, the weighted analysis method is used to identify the model, and the Kalman filter algorithm is applied to quickly and accurately identify the individual needs of the occupant, and the weighted root?mean?square threshold of comfort is obtained. The identification results show that the correlation between the subjective and objective comfort of the occupant reaches 85.81%; the three?way motion factor has a greater impact on the occupant comfort than the driving risk and efficiency factors; the identification rate of occupant personalized comfort is as high as 93%. The study can provide theoretical support for constructing personalized trajectory planning control algorithm considering occupant comfort.

Key words: autonomous driving, occupant comfort, identification, evaluation, personalization