摘要
准确估计周围车辆的驾驶员合作程度能提高无人驾驶车辆在密集交通场景下的换道安全性与效率,特别是对于自身灵活性和稳定性较差、对周围车辆安全威胁较大的重型卡车。因此,提出一种基于周围车辆驾驶员合作程度预测以及非对称风险评估的无人驾驶重型卡车换道决策规划方法。该方法基于高斯混合模型对周围车辆进行运动轨迹预测,结合当前驾驶环境估计目标车道后车的驾驶员合作程度,并用于构建非对称风险模型;基于轨迹预测结果,采用效用理论建模无人驾驶重型卡车当前和未来的车道选择概率,综合当前和未来的风险评估,输出最终的驾驶行为决策;设计多目标代价函数用于从多项式候选轨迹中选取最优轨迹。基于自然驾驶数据集的仿真试验表明,提出的方法可以准确地预测目标车道后车的驾驶员合作程度以及对周围车辆的风险等级,使无人驾驶重型卡车在密集交通流下也能安全高效地执行换道决策和轨迹规划。
The lane-change safety and efficiency in the dense traffic will be greatly improved if the driver cooperativeness of the surrounding vehicles can be estimated for the autonomous heavy truck,which has inferior flexibility and dynamic stability,as well as greater destructiveness.Therefore,this study propose a lane-change decision-making and planning method based on the predicted driver cooperativeness of the surrounding vehicles and the asymmetrical risk assessment.The driver cooperativeness of the following vehicle in the target lane is estimated by considering the driving environment and the trajectories predicted by Gaussian mixture model,which is also used to construct the asymmetrical risk model.The utility theory is used to describe the probability of selecting the target lane,and then the driving decision will be made by combining the risk level of the target lane in the prediction horizon.Finally,a multi-objective cost function is designed to select the optimal trajectory from the candidates generated by the polynomial curve.The results of simulation using the naturalistic driving dataset show that the proposed method can accurately predict the driver cooperativeness and correctly recognize the risk level.Sequentially,the autonomous heavy truck can make the lane-change decision more safely and efficiently in the dense traffic.
作者
胡文
邓泽健
张邦基
曹东璞
杨彦鼎
曹恺
李深
HU Wen;DENG Zejian;ZHANG Bangji;CAO Dongpu;YANG Yanding;CAO Kai;LI Shen(College of Mechanical and Vehicle Engineering,Hunan University,Changsha 410082;University of Waterloo,Waterloo,N2L3G1,Canada;Intelligent Transportation System Research Center,Zhejiang University City College,Hangzhou 310015;School of Vehicle and Mobility,Tsinghua University,Beijing 100084;Dongfeng USharing Technology Co.Ltd,Wuhan 430058;School of Civil Engineering,Tsinghua University,Beijing 100084)
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2023年第12期332-342,共11页
Journal of Mechanical Engineering
基金
国家自然科学基金资助项目(52202498)。
关键词
无人驾驶重型卡车
轨迹预测
驾驶员合作程度
风险评估
决策规划
autonomous heavy truck
trajectory prediction
driver cooperativeness
risk assessment
decision-making and planning
作者简介
胡文,男,1992年出生,博士研究生。主要研究方向为无人驾驶车辆风险评估、社会认知导向的驾驶行为决策与路径规划。E-mail:huxiaowen@hnu.edu.cn;通信作者:张邦基,男,1968年出生,教授,博士研究生导师。主要研究方向为车辆智能底盘、无人驾驶车辆路径跟踪控制。E-mail:zhangbj@zucc.edu.cn。
