交通运输系统工程与信息 ›› 2022, Vol. 22 ›› Issue (1): 85-97.DOI: 10.16097/j.cnki.1009-6744.2022.01.010

• 智能交通系统与信息技术 • 上一篇    下一篇

基于安全势场的网联自主车辆跟驰行为决策及模型

贾彦峰,曲大义*,赵梓旭,王韬,宋慧   

  1. 青岛理工大学,机械与汽车工程学院,山东 青岛 266520
  • 收稿日期:2021-09-17 修回日期:2021-11-01 接受日期:2021-11-09 出版日期:2022-02-25 发布日期:2022-02-23
  • 作者简介:贾彦峰(1992- ),男,山东菏泽人,博士生。
  • 基金资助:
    国家自然科学基金;山东省重点研发计划

Car-following Decision-making and Model for Connected and Autonomous Vehicles Based on Safety Potential Field

JIA Yan-feng, QU Da-yi* , ZHAO Zi-xu, WANG Tao, SONG Hui   

  1. School of Mechanical and Automotive Engineering, Qingdao University of Technology, Qingdao 266520, Shandong, China
  • Received:2021-09-17 Revised:2021-11-01 Accepted:2021-11-09 Online:2022-02-25 Published:2022-02-23
  • Supported by:
    National Natural Science Foundation of China(51678320); Key Research and Development Program of Shandong Province(2019GGX101038)

摘要: 为改善网联自主车辆(CAV)的跟车安全和效率,针对CAV通过对周围环境进行感知进而进行自主决策的特点,首先,建立包含车道线势场、道路边界势场和车辆势场的安全势场模型,系统地刻画CAV在行驶过程中面临的安全风险,在安全势场模型的建立过程中,针对现有车辆势场函数存在引力和斥力表达式分割独立的缺陷,借鉴分子间相互作用关系建立统一的基于LennardJones势的车辆相互作用势场函数,并将加速度参数引入到车辆势场中,加速度的变化直接影响车辆势场的分布,能够有效地反映车辆在不同运行状态下安全势场的动态变化趋势;然后,将安全势场应用于CAV跟驰行为决策,并通过上海自然驾驶数据集标定模型参数;最后,选择与现有经典的智能驾驶人IDM和可变车头时距VTH模型进行仿真对比。结果表明:与其他两种模型相比,该模型在所设计的3种交通场景中有更平滑的响应曲线来改善跟车安全和效率,验证了模型的有效性。研究成果可为CAV的上层控制设计提供理论支撑,也为CAV安全技术的研究提供了 独特的途径。

关键词: 智能交通, Lennard-Jones势, 安全势场, 网联自主车辆, 车辆跟驰

Abstract: To improve the safety and efficiency of Connected and Autonomous Vehicles (CAV) following behavior, this paper investigates the characteristics of CAV autonomous decision making when perceiving the surrounding environment. A safety potential field model is proposed to describe the safety risk of CAV in driving, which includes lane marking potential field, road boundary potential field, and vehicle potential field. In the model, the existing vehicle potential field function has independent gravitational and repulsive expressions. A unified function of vehicle potential field based on Lennard Jones potential is also included considering the relationship of intermolecular interaction, and the parameter of vehicle's acceleration. The statistical analysis of the parameter reveals that the change of acceleration directly affects the distribution of vehicle potential field and reflect the dynamic trend of vehicle's safety potential field under different driving states. The safety potential field is applied to the car- following behavior of CAV, and the model's parameters are calibrated by Shanghai natural driving data-set. Compared with the existing classic Intelligent Driver Model(IDM)and Variable Time Headway (VTH) models, the simulation results show the proposed model produces smoother response curves in the three car-following scenarios designed to improve the safety and efficiency, which proves the effectiveness of the model. The research results can provide theoretical support for the upper control design of CAV, and this study provide a unique way for CAV safety technology research.

Key words: intelligent transportation, Lennard-Jones potential, safety potential field, connected and autonomous vehicle, car following

中图分类号: