摘要
针对前车换道意图识别预测问题,提出基于动态栅格地图与混合高斯隐马尔科夫模型结合的识别预测方法,建立以自车车头中心点为零点的动态栅格地图描述前车的位置信息,基于语义分割的方法获取前方车辆的轮廓特征与车道线,并分析前方障碍车辆的相关运动状态信息。将动态栅格地图与混合高斯隐马尔科夫模型相结合,识别预测前车换道行为。利用可观测的前车行驶信息推测出其隐藏的换道行为,并提取NGSIM数据集的相关观测参数,训练验证识别预测模型,结果表明模型达到较好的识别预测效果。
Aiming at the problem of lane changing intention recognition and prediction of the front vehi⁃cle,a recognition and prediction method based on the combination of a dynamic grid map and Gaussian Mixture Hidden Markov Model was proposed.A dynamic grid map with the center point of the vehicle’s front as the zero point was used to describe the position information of the front vehicles.The contour features of the front vehicles and lane lines were obtained based on the semantic segmentation method,and the relevant motion state information of the obstacle vehicle in front was analyzed.The dynamic grid map was combined with the Gaussian Hidden Markov Model to identify and predict the lanechanging behavior of the preceding vehicle.The observable driving information was used to infer its hid⁃den lane changing behavior,and relevant observation parameters of NGSIM data sets were extracted to train and validate recognition and prediction models.The result shows that the model achieves a better recognition and prediction effect.
作者
杨杨
杨正才
蔡林
Yang Yang;Yang Zhengcai;Cai Lin(Hubei University of Automotive Technology,Shiyan 442002,China;Technical Center,Dongfeng Commercial Vehicle Co.Ltd,Shiyan 442000,China)
出处
《湖北汽车工业学院学报》
2021年第2期11-16,共6页
Journal of Hubei University Of Automotive Technology
基金
湖北省教育厅科学研究计划项目(D20181802)
汽车动力传动与电子控制湖北省重点实验室开放基金(ZDK1201401)。
关键词
动态栅格地图
混合高斯隐马尔科夫
换道意图识别预测
dynamic grid map
Gaussian Mixture Hidden Markov Model
lane change intention recogni⁃tionandprediction
作者简介
杨杨(1995-),男,硕士生,从事智能汽车行为预测方面的研究。E-mail:598084281@qq.com。
