基于LSTM网络预测智能车辆变道的路径规划研究

A path planning method of intelligent vehicle lane change prediction based on LSTM network

为了实现自主车辆的避障和变道,提出了一种基于LSTM网络预测终点位置的变道轨迹规划方法。首先,对NGSIM数据集中的变道规则进行重新定义,采用指数加权移动平均法(EWMA)对轨迹点进行平滑处理。利用LSTM网络来预测换道的终点位置,使用均方根误差(RMSE)作为模型的评价指标,在相同指标下模型精度提升约14.25%。同时,定义了车辆换道的安全区域来验证预测轨迹的可行性。最后,基于样条曲线理论,建立了变道轨迹数学模型,综合考虑安全性与舒适性选择最优的变道路径。仿真结果表明,提出的轨迹规划算法可以为自主车辆生成安全无碰撞的最优路径。

In order to realize obstacle avoidance and lane change of autonomous vehicles,a lane change path planning method based on LSTM network is proposed.Firstly,the lane changing rules in NGSIM data set are redefined,and the exponential weighted moving average(EWMA)method is used to smooth the trajectory points.The LSTM network is used to predict the end position of lane changing,and the root mean square error(RMSE)is used as the evaluation index of the model.Under the same index,the accuracy of the model is improved by about 14.25%.At the same time,the safety area of vehicle lane changing is defined to verify the feasibility of the predicted trajectory.Finally,based on the spline curve theory,the mathematical model of lane changing trajectory is established,and the optimal lane changing path is selected by comprehensively considering safety and comfort.Simulation results show that the proposed trajectory planning algorithm can generate a safe and collision free optimal path for autonomous vehicles.