The invention relates to a driving control method based on self-supervised imitation learning, and belongs to the technical field of automatic driving reinforcement learning. Comprises: generating an automatic driving task data set; determining a t moment potential representation of an anchor node in the observation image and a potential representation of a sample related to the anchor node; integrating the potential representation z < = t smaller than the t moment in the potential space, and generating the potential representation ct of the context at the t moment; retaining effective information between the observation image st + k at the t + k moment and the ct by using a density comparator; and after different automatic driving scene observation images pass through a trained encoder, calculating a cost function, and after the cost function is updated, transmitting the updated cost function into a simulation learning target formula to update a strategy network. According to the method, dimension disasters in a high-dimensional data learning strategy can be relieved through dimension reduction mapping, high-density information with higher correlation with tasks in image observation data is extracted, and the sampling efficiency of model samples is improved.

    本发明涉及一种基于自监督模仿学习的驾驶控制方法,属于自动驾驶强化学习技术领域。包括:生成自动驾驶任务数据集;确定所述观测图像中锚节点的t时刻潜在表示,及与所述锚节点相关样本的潜在表示;将潜在空间中小于t时刻的潜在表示z≤t进行整合,产生所述t时刻上下文潜在表示ct;利用密度比器保留t+k时刻的观测图像st+k和所述ct之间的有效信息;将不同自动驾驶场景观测图像经由训练好的编码器后计算代价函数,所述代价函数更新后传入到模仿学习目标公式中进行策略网络的更新。本发明能够通过降维映射来缓解高维数据学习策略中所面临的维度灾难,提取图像观测数据中与任务相关性更强的高密度信息,提升模型样本的采样效率。


    Access

    Download


    Export, share and cite



    Title :

    Driving control method based on self-supervised imitation learning


    Additional title:

    一种基于自监督模仿学习的驾驶控制方法


    Contributors:
    LI XIN (author) / ZANG HONGYU (author) / LIU CHEN (author) / YUAN YAN (author) / HUANG JIE (author)

    Publication date :

    2023-06-30


    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



    Self-Driving Car Using Supervised Learning

    Collins, Ryan ;Kumar Maurya, Himanshu ;Ragul, Raj S.R. | Trans Tech Publications | 2023


    Dynamic Conditional Imitation Learning for Autonomous Driving

    Eraqi, Hesham M. / Moustafa, Mohamed N. / Honer, Jens | IEEE | 2022


    Self-driving commercial vehicle formation driving decision-making method based on semi-supervised learning

    HU WEIMING / LIU YAN / ZHOU JINYING et al. | European Patent Office | 2024

    Free access

    Graph-Based Adversarial Imitation Learning for Predicting Human Driving Behavior

    Konstantinidis, Fabian / Sackmann, Moritz / Hofmann, Ulrich et al. | IEEE | 2024


    Parallelized and Randomized Adversarial Imitation Learning for Safety-Critical Self-Driving Vehicles

    Yun, Won Joon / Shin, MyungJae / Jung, Soyi et al. | ArXiv | 2021

    Free access