一句话:提出误行为预测方法,用于运行时预测自动驾驶系统中 DNN 组件的错误行为,弥补离线测试的不足。
深度神经网络(DNN)是现代自动驾驶系统的核心组件,但其难以在所有驾驶条件下正确泛化。现有测试技术多为离线方案,通过识别对抗样本或边缘案例来改进训练阶段,无法在运行时应对未预见的情况。本文提出误行为预测(misbehaviour prediction)方法,旨在运行时预测自动驾驶系统中 DNN 组件可能出现的错误行为,从而为在线缓解措施提供依据。该方法关注 DNN 预测的可靠性,评估在自动驾驶场景下进行,以检验预测的有效性。本文的工作转向运行时误行为预测,为自动驾驶系统的安全运行提供了新思路,弥补了离线测试的不足。
Deep Neural Networks (DNNs) are the core component of modern autonomous driving systems. To date, it is still unrealistic that a DNN will generalize correctly to all driving conditions. Current testing techniques consist of offline solutions that identify adversarial or corner cases for improving the training phase.
DOI 原文 ·
@inproceedings{paperbot5595,
title = {Misbehaviour prediction for autonomous driving systems},
author = {Andrea Stocco and Michael Weiss and Marco Calzana and Paolo Tonella},
booktitle = {ICSE},
year = {2020},
doi = {10.1145/3377811.3380353}
}