作者:刘彪，王恒阳，宋昱鹏，马晓宇
Authors:LIU Biao,WANG Heng-yang,SONG Yu-peng,MA Xiao-yu
摘要:摘要:针对复杂动态交通环境下，智能网联汽车做出安全合理的变道决策需要大量数据为支撑的问题，对车辆变道行为特征进行深入研究。首先对真实交通信息数据集NGSIM（next generation Simulation）的驾驶数据进行平滑去噪处理，再识别出车辆变道行为的产生和结束时刻，得到3200个包含交通环境时序信息的驾驶行为片段；基于8车交通模型对本车道及相邻车道前、后、旁车的相对速度、相对纵向距离以及本车变道时间等车辆周围的微观交通特性进行聚类分析，归纳出车辆变道的原因及行为特征。结果表明，寻求更高的驾驶速度和规避前方慢车以避免碰撞是车辆采取变道行为的两个主要原因，车辆更倾向于往车道行驶车速快和变道安全距离长的车道变道；车辆的变道行为在纵向空间维度具有前瞻性。这为微观交通仿真、驾驶模拟、无人驾驶车辆换道决策数据集的建立提供了理论基础。
Abstract:Abstract:Aiming at the problem that intelligent networked vehicles need a large amount of data to make safe and reasonable lane-changing decisions in complex dynamic traffic environments, the characteristics of vehicle lane-changing behaviors are deeply studied.Firstly, the driving data of NGSIM (Next Generation Simulation) were smoothed and de-noised.Then, the occurrence and end moments of lane change behavior were identified, and 3,200 lane change behavior fragments containing temporal information of traffic environment were obtained.Based on the 8-car traffic model, the micro-traffic characteristics around the vehicle, such as the relative speed, relative longitudinal distance and lane change time of the vehicle before, behind and beside the vehicle in the lane and adjacent lane, were analyzed by clustering, and the reasons and behavior characteristics of the vehicle lane change were summarized.The results show that seeking higher driving speed and avoiding the slow vehicle ahead to avoid collision are the two main reasons for vehicles to change lanes.Lane-changing behavior is forward-looking in the longitudinal spatial dimension.This provides a theoretical basis for the establishment of data sets for micro-traffic simulation, driving simulation, and unmanned vehicle lane-changing decision-making.

PDF全文下载地址:

可免费Download/下载PDF全文