We consider the problem of engineering robust direct perception neural networks with the output being regression. Such networks take high dimensional input image data, and they produce affordances such as the curvature of the upcoming road segment or the distance to the front vehicle. Our proposal starts by allowing a neural network prediction to deviate from the label with tolerance Δ. The source of tolerance can be either contractual or from limiting factors where two entities may label the same data with slightly different numerical values. The tolerance motivates the use of a non-standard loss function where the loss is set to 0 so long as the prediction-to-label distance is less than Δ. We further extend the loss function and define a new provably robust criterion that is parametric to the allowed output tolerance Δ, the layer index $l$ where perturbation is considered, and the maximum perturbation amount $k$. During training, the robust loss is computed by first propagating symbolic errors from the l-th layer (with quantity bounded by $k$) to the output layer, followed by computing the overflow between the error bounds and the allowed tolerance. The overall concept is experimented in engineering a direct perception neural network for understanding the central position of the ego-lane in pixel coordinates.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Towards Robust Direct Perception Networks for Automated Driving


    Contributors:


    Publication date :

    2020-10-19


    Size :

    859034 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Fault-tolerant environmental perception architecture for robust automated driving

    Grubmuller, Stephanie / Stettinger, Georg / Sotelo, Miguel Angel et al. | IEEE | 2019


    NeurAll: Towards a Unified Visual Perception Model for Automated Driving

    Sistu, Ganesh / Leang, Isabelle / Chennupati, Sumanth et al. | IEEE | 2019


    ADAPTIVE RATIONALIZER FOR VEHICLE PERCEPTION SYSTEMS TOWARD ROBUST AUTOMATED DRIVING CONTROL

    SHAHRIARI MOHAMMADALI / ZARRINGHALAM REZA / DALZELL HALEY MARIE et al. | European Patent Office | 2023

    Free access

    Adaptive rationalizer for vehicle perception systems toward robust automated driving control

    SHAHRIARI MOHAMMADALI / ZARRINGHALAM REZA / DALZELL HALEY MARIE et al. | European Patent Office | 2024

    Free access

    Towards Realistic Evaluation of Collective Perception for Connected and Automated Driving

    Volk, Georg / Delooz, Quentin / Schiegg, Florian A. et al. | IEEE | 2021