摘要
针对城市道路交叉路口智能车周围转弯车辆的轨迹预测问题,提出一种基于三阶贝塞尔曲线和运动模型的转弯车辆轨迹预测方法。首先,在智能车位于道路交叉口附近时,结合感知模块输出的障碍物信息和高精度地图信息对智能车周围车辆进行行为识别。然后,对行为识别结果为转弯状态的车辆,利用高精度地图中的拓扑关系,结合车辆状态信息选取目标点;在选定目标点之后,使用三阶贝塞尔曲线预测车辆未来4 s内的轨迹,同时结合车辆状态信息,应用恒定转率和加速度模型(CTRA model)预测车辆未来4 s内的轨迹。最后,使用权重函数加权得到最终的预测轨迹。实车实验结果表明,道路交叉口处转弯车辆4 s内的轨迹预测平均误差为2.34 m,较CTRA模型预测误差减小了3.86 m,单个转弯车辆轨迹预测平均耗时为0.14 ms,验证了本文所提方法的有效性、准确性以及实时性。
To solve the trajectory prediction problem of intelligent vehicles around urban road intersection,this paper proposes a method for predicting turning vehicle trajectory based on three order Bessel curve and motion model.Firstly,when the intelligent vehicle is located near the road intersection,it carries out the behavior recognition of the vehicle around the intelligent vehicle by combining the obstacle information output by the perception module and the high-precison map information.Then,it identifies the vehicle with the turning behavior by using the topological relationship of the high precision map and the vehicle state information.After selecting the target point,it predicts the trajectory of the vehicle in the future 4 S with three order Bessel curve,and predicts the trajectory of the vehicle in the future 4 S by using the constant rate and acceleration model(CTRAmodel)combined with the vehicle state information.Finally,it weighs the final predicted trajectory by using weight function.The actual vehicle experiment results show that the average trajectory predition error of turning vehicles at the intersection is 2.34 m,which is 3.86 m less than that of CTRA model,and the average trajectory prediction time of single turning vehicle is 0.14 ms.The validity,accuracy and real-time of the proposed method are verified.
作者
谢枫
李永乐
苏致远
白睿
徐友春
XIE Feng;LI Yongle;SU Zhiyuan;BAI Rui;XU Youchun(Fifth Team of Cadets,Army Military Transportation University,Tianjin 300161,China;Institute of Military Transportation,Army Military Transportation University,Tianjin 300161,China)
出处
《军事交通学院学报》
2019年第11期78-83,共6页
Journal of Military Transportation University
基金
国家重点研发计划项目(2016YFB0100903).
作者简介
谢枫(1996—),男,硕士研究生.
