考虑驾驶人交通环境感知的车辆安全势场及跟驰行为建模

Vehicle safety potential field and car-following model based on traffic environment perception

安全势场能够描述车辆驾驶过程中周围安全风险的空间分布。针对既有模型重点关注车辆自身运动状态而忽视驾驶人环境感知信息的问题,围绕车辆安全势场模型改进以及其在跟驰模型中的应用展开研究。引入相对状态影响因子和道路交通状态影响因子对既有模型进行改进,强化车辆间相对速度和所处道路交通状态对行车安全性的影响;利用车型系数对实际空间的距离进行修正,研究多车型混合环境下车型差异对行车安全性的影响;利用感知安全势场将前车运动状态与后车跟驰行为建立联系,得到基于感知安全势场的车辆跟驰模型;采用遗传算法对本文所建模型和智能驾驶人跟驰模型、安全势场跟驰模型进行标定。结果表明,上述3个模型在测试集上的均方根误差分别为6.124、8.515、7.248,证明该模型误差最小,能够更为精确地描述车辆跟驰行为。研究成果能为行车安全风险评估和车辆驾驶行为决策提供理论依据。

The safety potential field is utilized to characterize the distribution of safety risks around a vehicle during the driving process.However,when analyzing the safety potential field formed by moving vehicles,the existing models only focus on the vehicle motion but ignore the traffic environment information perceived by drivers.This study focuses on the construction of an improved safety potential field model and its application to the car-following model.Herein,the relative state influence factor is introduced to strengthen the influence of relative speed among vehicles,and the traffic state influence factor is introduced to reflect its influence on driving safety.Moreover,the vehicle type coefficient is introduced to adjust the distance to reflect its influence on driving safety in mixed vehicle type traffic.The car-following model is developed by using the preceptive safety potential field to establish the relationship between the motion state of the front vehicle and the behavior of the following vehicle.Furthermore,the genetic algorithm is employed to calibrate the proposed model,the intelligent driver model,and the car-following model based on the safety potential field.The results show that the root mean square errors of these three models mentioned before are 6.124,8.515 and 7.248 respectively,which proves that the model proposed in this paper can describe car-following behavior more accurately.Therefore,this study can provide theoretical support for driving risk evaluation and vehicle control under a complex environment.