考虑前车状态的智能网联车交叉口行为决策

为使智能网联汽车（intelligent connected vehicle, ICV）在复杂交通环境下高效、安全地通过信号交叉口，在车联网实时获取信号灯和前车状态信息的基础上，建立了智能网联汽车通过信号交叉口的驾驶行为决策框架. 通过跟驰模型推导智能网联汽车和前方车辆在未来的行驶状态，预测得到前方车辆是否要通过交叉口的行为，进一步分别对智能网联汽车是领头车和跟随车时通过交叉口停止线的条件进行判断；将换道加入到驾驶方式中来寻求更高的通行效率，用基于换道时间模型的方法判断智能网联汽车换道后的通过条件；仿真对比分析了所提出模型和现有模型的决策能力，讨论了影响决策过程的关键因素. 研究结果表明:相比于现有模型，综合信号灯和前车行驶意图的决策方法能够提高智能网联汽车对通行条件判断的准确性，从而进行更合理的行为选择，随着单位绿灯剩余时间的增加，车辆决策通过交叉口的概率可提高20%，当前车道的车辆位置对决策结果影响显著. 

Behavior Decision of Intelligent Connected Vehicles Considering Status of Preceding Vehicles at Intersections

To enable intelligent connected vehicles (ICV) pass through signalized intersections efficiently and safely in complex traffic environments, on the basis of real-time acquisition of signal lights and preceding vehicle status information in the internet of vehicles, the driving behavior decision framework of ICV through signalized intersections in the ICV environment was established.The car-following model was used to derive the future driving status of ICV and vehicles ahead,then to predict whether the preceding vehicles will pass through the intersection, and further to judge the conditions for passing the intersection stop line when the ICVis the leader or the follower.The lane change was addedin the driving mode toseek higher traffic efficiency, and the method based on the lane change time model was used to judge the passing conditions after the ICV changesa lane.Through simulation comparison, the decision-making ability of the proposed model and the existing model was analyzed, the key factors that affect the decision-making process were discussed. Compared with the existing model, the decision method of combining signal lights and the driving intention of the preceding vehicle can improve the judgmentaccuracy of ICVs on passing conditions and realize reasonable decision-making.With the increasing of the remaining time per unit green light, the probability of vehicles passing through the intersection can be increased by 20%, and vehicle positions in the current lane have significant influence on decision result.