The invention discloses a behavior modal division method, and a training method and a reasoning method of a multi-modal trajectory prediction model, and the method comprises the steps: dividing a behavior modal of a vehicle trajectory based on a basic driving rule, and constructing a generalization modal domain; reasonable space guidance is constructed based on modal refinement of road map elements; and outputting multi-mode track prediction results corresponding to different high-level control instructions through a multi-mode decoder according to the coding result of the intelligent agent. According to the multi-modal trajectory prediction model, basic driving rules are used for generalization modal division, so that training samples with similar driving behaviors are aggregated, more visual, wider and more effective modal classification is realized, and the model training efficiency is improved. And secondly, a high-fine-granularity lane-level modal refinement method is designed, so that the correlation of information fusion is improved, the manual design of various candidate priori is avoided, and a richer and more effective trajectory prediction modal result is naturally guided and generated.

    本发明公开了一种行为模态划分方法、多模态轨迹预测模型的训练方法与推理方法,该方法包括:基于基本驾驶规则划分车辆轨迹的行为模态,构建泛化模态域;基于道路地图元素的模态细化,构建合理的空间引导;智能体编码结果通过多模态解码器输出对应不同高级别控制指令的多模态轨迹预测结果。本发明的多模态轨迹预测模型将基本驾驶规则用于泛化模态的划分,从而聚合具有相似驾驶行为的训练样本,实现更加直观、广泛、有效的模态分类,提升模型训练效率。其次设计了高细粒度的车道级模态细化方法,提升了信息融合的相关性,避免手工设计繁多的候选先验,自然引导生成更加丰富、有效的轨迹预测模态结果。


    Access

    Download


    Export, share and cite



    Title :

    Behavior modal division method, and training method and reasoning method of multi-modal trajectory prediction model


    Additional title:

    一种行为模态划分方法、多模态轨迹预测模型的训练方法与推理方法


    Contributors:
    CAI YINGFENG (author) / LU ZIHENG (author) / WANG HAI (author) / DONG ZHAOZHI (author) / CHEN LONG (author) / LIU ZE (author)

    Publication date :

    2024-08-30


    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 / B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion



    Multi-modal trajectory prediction method

    JIANG WENJUAN / JIN ZHI / WANG REN et al. | European Patent Office | 2023

    Free access

    Multi-modal multi-agent trajectory prediction

    SUN PEI / ZHAO HANG / MCCAULEY ALEXANDER et al. | European Patent Office | 2024

    Free access

    MULTI-MODAL MULTI-AGENT TRAJECTORY PREDICTION

    SUN PEI / ZHAO HANG / MCCAULEY ALEXANDER et al. | European Patent Office | 2022

    Free access

    Vehicle multi-modal trajectory prediction method based on semi-supervised model

    TIAN WEI / WANG SONGTAO | European Patent Office | 2023

    Free access

    Multi-modal vehicle trajectory prediction method based on graph interaction mechanism

    GAO JIANPING / YANG YU / LIU PAN et al. | European Patent Office | 2024

    Free access