网联车辆并线预测与巡航控制的研究

为检测旁车道车辆驾驶员的并线意图,提升网联车辆巡航跟车的主动安全性,提出了一种基于NAR神经网络学习的迭代循环预测算法。NAR神经网络的训练样本由实际交通环境中的车辆并线数据获得,通过训练的网络预测未来一段时间内旁车的横向行驶轨迹,并根据划定的监控区域计算旁车的切入概率。同时,提出了一种考虑并线概率的跟车距离策略,并应用到网联车辆CACC系统中。结果表明,所提出的并线预测算法能精确计算出旁车的横向换道轨迹,所提出的跟车策略可提升车辆的跟车安全性。

Research on Merging Prediction and Cruise Control for Connected Vehicles

For detecting the driver s merging intention of the vehicle in adjacent lane and enhance the cruising active safety of connected vehicles,an iterative loop prediction algorithm based on nonlinear autoregressive(NAR)neural network learning is proposed.The training samples of NAR neural network are obtained from the merging data of vehicles in real traffic environment,the trained network is used to predict the lateral trajectory of the adjacent vehicle in a certain time-segment of future,and the cut-in probability of adjacent vehicle is calculated according to the designated monitoring area.Meanwhile,a follow-up distance strategy considering merging probability is also proposed and applied to the connected vehicle CACC system.The results show that the merging prediction algorithm proposed can accurately calculate the lateral lane change trajectory of adjacent vehicle,and the follow-up strategy proposed can enhance the follow-up safety of vehicle.