The invention discloses a multi-view attention mechanism end-to-end automatic driving behavior decision-making method, and belongs to the technical field of automatic driving. The method comprises the following steps: obtaining a multi-moment scene image of each view angle based on video stream information obtained by multiple cameras; the method comprises the following steps: performing dimension reduction on each visual angle to obtain each visual angle feature, extracting each visual angle feature through a multi-visual angle attention network module, fusing weight information of different objects of other visual angles, and acting the weight information back to original input to obtain a fusion feature of each visual angle; after pooling processing, inputting the data into a long-short-term memory network, and obtaining time sequence characteristics of each view angle; and finally, in combination with the real-time navigation information, predicting a turning angle control quantity and a speed control quantity of the vehicle as final decision information. Based on multi-view attention, the position information and the time information of the target around the vehicle body are extracted by the visual sensors, so that the attention information around the vehicle body can be observed in the prediction process, and the prediction result is more accurate.

    本发明公开了一种多视角注意力机制端到端自动驾驶行为决策方法,属于自动驾驶技术领域。该方法基于多摄像头获取的视频流信息,得到各视角的多时刻场景图像;将其分别降维后得到各视角特征,通过多视角注意力网络模块提取出各视角特征融合了其它视角不同事物的权重信息,再将该权重信息作用回原输入,得到各视角的融合特征;再分别经过池化处理后输入长短时记忆网络,获取各视角的时序特征;最后结合实时导航信息,预测车辆的转角控制量和速度控制量,作为最终的决策信息。本发明基于多视角注意力,综合各个视觉传感器提取车身周围目标位置信息和时间信息,使得预测过程更能观察到车身周围应注意信息,从而使预测结果更加准确。


    Access

    Download


    Export, share and cite



    Title :

    End-to-end automatic driving behavior decision-making method of multi-view attention mechanism


    Additional title:

    一种多视角注意力机制的端到端自动驾驶行为决策方法


    Contributors:
    ZHAO YANG (author) / LI JIE (author) / LI SHUGUANG (author) / PENG ZHINAN (author) / HUANG RUI (author) / CHENG HONG (author)

    Publication date :

    2022-09-23


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    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 / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06V



    Automatic driving behavior decision-making method for view shielding

    ZHOU JINCAO / ZHANG GUOWEI / FU WEIPING et al. | European Patent Office | 2023

    Free access

    Automatic driving behavior decision-making method and system based on multiple attention mechanisms and medium

    ZHANG PIN / LEI DI / CUI ZHIBIN et al. | European Patent Office | 2023

    Free access

    Automatic driving behavior decision-making system and method based on driving preference

    ZHANG SHUO / REN FAN / WEN TAO et al. | European Patent Office | 2022

    Free access

    Automatic driving vehicle behavior decision-making system and method

    CEN MING / YANG CHENG / YANG RUN et al. | European Patent Office | 2023

    Free access

    Interaction behavior decision-making method and device of automatic driving vehicle and automatic driving vehicle

    LIU JINXIN / LI WENBO / PENG LIANG et al. | European Patent Office | 2023

    Free access