A method for virtual lane lines fitting based on traffic environment information was proposed to address the problem of lane lines missing caused by wear, ice and snow coverage, etc. on certain road sections during vehicle driving. Firstly, a probability model for the lateral offset of lane lines under lane-keeping conditions of traffic vehicles was established based on the highD naturalistic driving dataset to determine the lateral position of virtual lane lines. Then, a traffic vehicle intention recognition and trajectory prediction model based on bi-directional long short-term memory networks was established to realize the condition judgment and shape generation of virtual lane lines generation. Finally, based on the information of the above, the statistical characteristics of lateral lane deviation of traffic vehicles that meet the fitting conditions were transformed into the risk field of virtual lane lines in the main vehicle coordinate system to realize the virtual lane lines fitting. Simulation and real vehicle data validation results show that this method can generate safe and effective virtual lane lines with a small error compared to the true value of lane lines, providing a reference driving space for the main vehicle and ensuring driving safety.
因此,本文提出了一种基于交通环境信息的虚拟车道线拟合方法,充分挖掘并利用交通环境中丰富的交通车运动信息构建虚拟车道线。首先,基于highD自然驾驶数据集建立交通车辆在车道保持状态下的车道线横向偏移概率模型,以确定虚拟车道线横向位置;其次,基于双向长短期记忆网络(Bi-directional long short-term memory,Bi-LSTM)建立交通车辆意图识别与轨迹预测模型,结合交通车辆在车道保持时获得的历史和预测轨迹,实现虚拟车道线产生条件判别和形状生成;最后,根据上述信息,将满足拟合条件的交通车车道线横向偏移统计学特征转化为主车坐标系下的虚拟车道线风险场,实现虚拟车道线拟合,并进行仿真和实车数据测试验证。该方法能够在车道线短暂消失情况下,充分利用交通环境信息生成可信赖的虚拟车道线,为主车提供可参考行驶空间,使自动驾驶系统持续安全运行,确保行车安全。
1 问题定义和总体框架
本文待解决的问题为:在车道线短暂缺失情况下,如何充分利用交通环境信息,拟合主车视角下可参考和使用的虚拟车道线。本文所述的交通环境信息主要指主车周围交通车的运动学信息[8-10]。研究的具体场景如图1所示,在单向3车道道路中,车道线存在的正常场景和车道线缺失的异常场景,两种场景中的行驶车辆按位置可划分为两侧车道车辆和中间车道车辆。当车道线缺失时,两侧车道的车辆分别以各自一侧的道路边界(如路缘石、护栏、树木等)为参考进行车道保持运动;而中间车道的车辆与左右两侧道路边界距离均较远,缺少可参考的行驶空间。这种车道线信息的缺失使自动驾驶系统功能失效,甚至面临与两侧交通车碰撞的行车风险。因此,本文将重点研究主车位于中间车道且左右两侧均有交通车场景下的虚拟车道线拟合问题。定义主车为EV,其左右两侧的交通车辆为LV i 和RV j (i,j=1,2,…,n)。在t时刻,主车观察到的左侧i辆和右侧j辆交通车辆分别由主车坐标系中的相对运动学特征和表示,即:
DengTian-min, XiePeng-fei, YuYang, et al. Method of lane detection based on adaptive fusion of double branch features[J/OL]. [2024-05-03].
[5]
TabeliniL, BerrielR, PaixaoT M, et al. Keep your eyes on the lane: Real-time attention-guided lane detection[J/OL]. [2024-05-03]
[6]
JohnV, KarunakaranN M, GuoC Z, et al. Free space, visible and missing lane marker estimation using the PsiNet and extra trees regression[C]∥2018 24th International Confernece on Pattern Recognition, Beijing, China, 2018: 189-194.
[7]
JohnV, LiuZ, GuoC Z, et al. Real-time lane estimation using deep features and extra trees regression[M]. Cham: Springer International Publishing, 2015.
[8]
WangX L, QianY Q, WangC X, et al. Map-enhanced ego-lane detection in the missing feature scenarios[J]. IEEE Access, 2020, 8: 107958-107968.
[9]
YaoH C, ChenC, LiuS D, et al. Lane marking detection algorithm based on high-precision map and multisensor fusion[J]. Concurrency and Computation: Practice and Experience, 2022, 34(8): No.e5797.
QiChun-yang. Research on energy management strategy for hybrid electric vehicle with incorporation of traffic information[D]. Changchun: College of Automotive Engineering, Jilin University,2023.
[12]
王如月. 基于交通信息的汽车制动动力学研究[D].北京:清华大学车辆与运载学院, 2021.
[13]
WangRu-yue. Research on vehicle braking dynamics based on traffic information[D]. Beijing: School of Vehicle and Mobility, Tsinghua University, 2021.
[14]
ZhangY, ZhangG. Research on the cooperative car-following behavior based on traffic information iteration[J]. International Journal of Modern Physics C, 2023, 34(1): 1-12.
[15]
DingJ Y, WangJ Q, LiuC C, et al. Driver steering behavior model based on lane-keeping characteristics analysis[C]∥17th international IEEE conference on intelligent transportation systems (ITSC), Qingdao, China, 2014: 623-628.
[16]
TianY, PeiH X, YangJ X, et al. An improved model of driving risk field for connected and automated vehicles[C]∥2021 IEEE International Intelligent Transportation Systems Conference (ITSC), Indianapolis, USA, 2021: 285-291.
HeRen, ZhaoXiao-cong, YangYi-bin, et al. Man⁃machine shared driving model using risk⁃response[J]. Journal of Jilin University (Engineering and Technology Edition), 2021, 51(3): 799-809.
[19]
SongX, SunY, TaoP. A dynamic Bayesian network model for real-time risk propagation of secondary rear-end collision accident using driving risk field [J]. IEEE Access, 2022, 10: 72429-72443.
[20]
MaY L, DongF Q, YinB Q, et al. Real-time risk assessment model for multi-vehicle interaction of connected and autonomous vehicles in weaving area based on risk potential field[J]. Physica A, 2023, 620: No. 128725.
[21]
LiL H, GanJ, YiZ, et al. Risk perception and the warning strategy based on safety potential field theory[J]. Accident Analysis and Prevention, 2020, 148: No. 105805.