汽车工程 ›› 2023, Vol. 45 ›› Issue (3): 361-371.doi: 10.19562/j.chinasae.qcgc.2023.03.003

所属专题: 智能网联汽车技术专题-控制2023年

• • 上一篇    下一篇

考虑前车运动不确定性的多目标自适应巡航控制

张紫微1,郑玲1,2(),李以农1,2,乔旭强1,郑浩1,王戡1   

  1. 1.重庆大学机械与运载工程学院,重庆 400044
    2.重庆大学,机械传动国家重点实验室,重庆 400044
  • 收稿日期:2022-10-12 修回日期:2022-11-08 出版日期:2023-03-25 发布日期:2023-03-22
  • 通讯作者: 郑玲 E-mail:zling@cqu.edu.cn
  • 基金资助:
    国家自然科学基金面上项目(51875061)

A Multi-objective Adaptive Cruise Control Strategy for Autonomous Vehicle Considering Uncertain Movements of Preceding Vehicle

Ziwei Zhang1,Ling Zheng1,2(),Yinong Li1,2,Xuqiang Qiao1,Hao Zheng1,Kan Wang1   

  1. 1.College of Mechanical Engineering,Chongqing University,Chongqing  400044
    2.Chongqing University,State Key Laboratory of Mechanical Transmissions,Chongqing  400044
  • Received:2022-10-12 Revised:2022-11-08 Online:2023-03-25 Published:2023-03-22
  • Contact: Ling Zheng E-mail:zling@cqu.edu.cn

摘要:

考虑前车运动状态不可控所带来的性能下降,提出一种基于高斯过程的随机模型预测多目标自适应巡航控制方法。基于车间运动关系对跟驰系统进行集成建模,综合考虑车辆安全、经济、舒适等多维诉求,确定跟驰系统目标函数与性能约束;引入径向基核描述样本间的关系,通过极大似然法获取预测模型超参数,根据历史交通数据,对前车运动轨迹进行短期预测;考虑预测结果存在的偏差,引入概率约束,建立不确定环境下的随机预测模型以保障系统在随机扰动下的整体性能最优;通过切入、加速跟驰、减速避撞等典型场景对算法的有效性与优越性进行验证。研究结果表明:所提出的方法具有良好的工况适应性,可快速消除跟踪误差与前车运动保持一致,使车辆对交通环境的反应更加敏捷。

关键词: 自适应巡航控制, 随机模型预测控制, 智能汽车, 高斯过程, 前车运动

Abstract:

Considering the performance degradation caused by the uncontrollable movement of the preceding vehicle, this paper proposes a stochastic model predictive control strategy based on the Gaussian process for adaptive cruise control. Firstly, an integration model of the car-following system is constructed based on the kinematic relationship between the vehicles. And objective functions and performance constraints of the car-following system are formulated considering comprehensively the multi-dimensional demand of vehicle security, fuel economy, ride comfort, etc. Then, the radial basis function kernel is introduced to describe the relationship among samples and hyperparameters are obtained via the maximum-likelihood method. Based on historical traffic data, the trajectory of the preceding vehicle is predicted in a short term. Subsequently, in consideration of the error between prediction results and its actual values, probability constraints are introduced to establish the stochastic predictive control model under uncertain environment to ensure the optimal overall performance of the system in the presence of stochastic disturbance. Finally, the superiority and effectiveness of the algorithm are verified by typical scenarios such as cut-in, acceleration for car following, and deceleration for collision avoidance. The results show that the proposed strategy possesses good adaptability to working conditions, which can quickly eliminate the tracking errors and keep consistent with the movement of the preceding vehicle. Thus, it makes the vehicle respond more quickly to the highly dynamic traffic environment.

Key words: adaptive cruise control, stochastic model predictive control, autonomous vehicle, Gaussian process, movement of preceding vehicle