Robust Trajectory Prediction against Adversarial AttacksDownload PDF

Published: 10 Sept 2022, Last Modified: 12 Mar 2024CoRL 2022 OralReaders: Everyone
Keywords: Adversarial Attack, Trajectory Prediction, Autonomous Driving
TL;DR: We propose a new adversarial training framework for training robust trajectory prediction systems by addressing domain specific challenges.
Abstract: Trajectory prediction using deep neural networks (DNNs) is an essential component of autonomous driving (AD) systems. However, these methods are vulnerable to adversarial attacks, leading to serious consequences such as collisions. In this work, we identify two key ingredients to defend trajectory prediction models against adversarial attacks including (1) designing effective adversarial training methods and (2) adding domain-specific data augmentation to mitigate the performance degradation on clean data. We demonstrate that our method is able to improve the performance by 46\% on adversarial data and at the cost of only 3\% performance degradation on clean data, compared to the model trained with clean data. Additionally, compared to existing robust methods, our method can improve performance by 21\% on adversarial examples and 9\% on clean data. Our robust model is evaluated with a planner to study its downstream impacts. We demonstrate that our model can significantly reduce the severe accident rates (e.g., collisions and off-road driving).
Student First Author: yes
Supplementary Material: zip
Website: https://robustav.github.io/RobustTraj/
Code: https://github.com/kikacaty/RobustTraj
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 1 code implementation](https://www.catalyzex.com/paper/arxiv:2208.00094/code)
5 Replies

Loading