We consider the challenge of automated steering angle prediction for self driving cars using egocentric road images. In this work, we explore the use of feudal networks, used in hierarchical reinforcement learning (HRL), to devise a vehicle agent to predict steering angles from first person, dash-cam images of the Udacity driving dataset. Our method, Feudal Steering, is inspired by recent work in HRL consisting of a manager network and a worker network that operate on different temporal scales and have different goals. The manager works at a temporal scale that is relatively coarse compared to the worker and has a higher level, task-oriented goal space. Using feudal learning to divide the task into manager and worker sub-networks provides more accurate and robust prediction. Temporal abstraction in driving allows more complex primitives than the steering angle at a single time instance. Composite actions comprise a subroutine or skill that can be re-used throughout the driving sequence. The associated subroutine id is the manager network's goal, so that the manager seeks to succeed at the high level task (e.g. a sharp right turn, a slight right turn, moving straight in traffic, or moving straight unencumbered by traffic). The steering angle at a particular time instance is the worker network output which is regulated by the manager's high level task. We demonstrate state-of-the art steering angle prediction results on the Udacity dataset.


    Access

    Download


    Export, share and cite



    Title :

    Feudal Steering: Hierarchical Learning for Steering Angle Prediction


    Contributors:

    Publication date :

    2020


    Remarks:

    InThe IEEE/CVFConference on Computer Vision and Pattern Recognition(CVPR) Workshops, June 2020
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp. 1002-1003. 2020


    Type of media :

    Preprint


    Type of material :

    Electronic Resource


    Language :

    English




    VARIABLE STEERING ANGLE STEERING DEVICE

    TANI HIDEYUKI | European Patent Office | 2019

    Free access

    PREDICTION STEERING ANGLE CALCULATION DEVICE

    SAIGO SHINTARO / MANABE SHUHEI / INOUE SATOSHI | European Patent Office | 2021

    Free access

    VARIABLE STEERING ANGLE STEERING DEVICE

    TANI HIDEYUKI | European Patent Office | 2017

    Free access

    STEERING ANGLE ESTIMATION DEVICE, STEERING DEVICE, AND STEERING ANGLE ESTIMATION METHOD

    SUZUKI TERUHIKO | European Patent Office | 2021

    Free access

    Variable steering angle steering device

    TANI HIDEYUKI | European Patent Office | 2019

    Free access