汽车工程 ›› 2023, Vol. 45 ›› Issue (12): 2222-2233.doi: 10.19562/j.chinasae.qcgc.2023.12.004

所属专题: 智能网联汽车技术专题-控制2023年

• • 上一篇    下一篇

面向密集障碍规避的人车共享转向控制系统

严凉1,吴晓东2(),胡川2   

  1. 1.上海交通大学机械与动力工程学院,上海  200240
    2.上海交通大学,汽车动力与智能控制国家工程研究中心,上海  200240
  • 收稿日期:2023-05-31 修回日期:2023-06-29 出版日期:2023-12-25 发布日期:2023-12-21
  • 通讯作者: 吴晓东 E-mail:xiaodongwu@sjtu.edu.cn
  • 基金资助:
    国家自然科学基金(51775331)

Human-Vehicle Shared Steering Control System for Dense Obstacle Avoidance

Liang Yan1,Xiaodong Wu2(),Chuan Hu2   

  1. 1.School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai  200240
    2.Shanghai Jiao Tong University,National Engineer Research Center of Automotive Power and Intelligent Control,Shanghai  200240
  • Received:2023-05-31 Revised:2023-06-29 Online:2023-12-25 Published:2023-12-21
  • Contact: Xiaodong Wu E-mail:xiaodongwu@sjtu.edu.cn

摘要:

车辆底盘技术的创新与发展为智能驾驶系统的设计注入新的活力。结合线控转向系统的物理解耦特性,提出了一种面向密集障碍物避让场景的人车共享转向控制系统方案。首先,基于车辆动力学和蒙特卡洛树搜索算法,采用滚动时域转向场直方图法实现车辆的避障动作决策与规划;其次,搭建了基于门控循环单元网络的驾驶人行为短时预测模型,可直接输出驾驶人转角预测时序信号;在此基础上,结合车辆转向特性计算预期碰撞时距,据此建立风险评估模型;最后,构建了车辆转向控制权的动态分配策略,基于MATLAB/Simulink搭建的联合仿真环境开展了硬件在环驾驶实验。对比多工况下的共享驾驶与手动驾驶的结果表明,共享转向控制方法能有效减少碰撞、提升驾驶安全性,且在保证通行效率的情况下减轻驾驶人的操作负担。

关键词: 间接共享控制, 人机协同, 避障规划, 驾驶人行为预测

Abstract:

The innovation and development of vehicle chassis technology have injected new vitality into the design of intelligent driving systems. Combining the physical decoupling characteristics of steer-by-wire (SBW) systems, a human-machine shared steering control system for dense obstacle avoidance scenarios is proposed. Firstly, based on vehicle dynamics and the Monte Carlo tree search (MCTS) algorithm, a receding horizon steering field histogram (RH-SFH) method is utilized to achieve the real-time obstacle avoidance decision-making and planning. Secondly, a short-term driver behavior prediction model based on gated recurrent unit (GRU) network is established, which can directly output temporal signal of steering command. Then, expected time-to-collision is calculated referring to the vehicle steering characteristics, and the risk evaluation model is established accordingly. Finally, a dynamic allocation strategy for vehicle steering control rights is constructed, and hardware-in-the-loop experiments are conducted in the joint simulation environment based on MATLAB/Simulink. The results show that the shared steering control method can effectively reduce collision, improve the driving safety, and reduce the operational burden of drivers while ensuring the traffic efficiency compared with manual driving under multiple working conditions.

Key words: indirect shared control, human-machine coordination, collision avoidance planning, driver behavior prediction