The invention discloses a vehicle trajectory prediction method based on a spatial motion relation and a long-short time memory network, which takes the long-short time memory network as a basis of an encoder-decoder structure, and adds six-dimensional features (absolute positions, absolute speeds, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, absolute positions, according to the method, twelve-dimensional features (including relative position, relative speed and relative acceleration in the transverse direction and the longitudinal direction) under a dynamic coordinate system and six-dimensional features (relative position, relative speed and relative acceleration in the transverse direction and the longitudinal direction) under the dynamic coordinate system are obtained, and a space interaction relation graph attention network model between a target vehicle and surrounding traffic objects is established according to a braking relation between vehicles under a local view angle. And then training and extracting space influence distribution between the target vehicle and the surrounding traffic objects, thereby improving the accuracy of vehicle trajectory prediction. The vehicle dynamic space relation feature and socialization space relation extraction method is applied to vehicle track prediction, and the vehicle prediction effect in the intelligent driving environment is improved.

    本发明公开了一种基于空间运动关系和长短时间记忆网络的车辆轨迹预测方法该方法以长短时间记忆网络为编码器‑解码器结构的基础,通过从输入端加入周围车辆静态坐标系下的六维特征(横向和纵向下的绝对位置,绝对速度,绝对加速度)和动态坐标系下的六维特征(横向和纵向下的相对位置,相对速度,相对加速度)总共十二维特征,并根据车辆之间在局部视角下的制动关系建立了目标车辆和周围交通对象之间空间交互关系图注意力网络模型。之后训练并提取目标车辆和周围交通对象之间的空间影响力分配,从而可以提升了车辆轨迹预测的准确度。将车辆动态空间关系特征和社会化空间关系提取方法应用于车辆轨迹预测,实现智能驾驶环境下的车辆预测效果提升。


    Access

    Download


    Export, share and cite



    Title :

    Vehicle track prediction method based on spatial motion relation and long-short time memory network


    Additional title:

    一种基于空间运动关系和长短时间记忆网络的车辆轨迹预测方法


    Contributors:
    WANG ZELIN (author) / LI YONG (author)

    Publication date :

    2024-01-30


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung



    Vehicle track prediction method based on spatial motion relation and long-short time memory network

    XIAO JIANHUA / XIAO HONGBO / DING LIMING et al. | European Patent Office | 2024

    Free access

    Peripheral vehicle track prediction method and system based on long and short time motion track fusion

    ZHANG ZHANJUN / WANG ZHANGUO / WU RENJIE et al. | European Patent Office | 2022

    Free access

    STG-LSTM: Spatial-temporal graph-based long short-term memory for vehicle trajectory prediction

    Daniela Daniel Ndunguru / Fan Xing / Chrispus Zacharia Oroni et al. | DOAJ | 2025

    Free access


    Ego-Vehicle Speed Prediction Using a Long Short-Term Memory Based Recurrent Neural Network

    Yeon, Kyuhwan / Min, Kyunghan / Shin, Jaewook et al. | Online Contents | 2019