Robust Trajectory Prediction against Adversarial Attacks

Yulong Cao, Danfei Xu, Xinshuo Weng, Zhuoqing Mao, Anima Anandkumar, Chaowei Xiao, Marco Pavone
Proceedings of The 6th Conference on Robot Learning, PMLR 205:128-137, 2023.

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).

Cite this Paper

BibTeX
@InProceedings{pmlr-v205-cao23a, title = {Robust Trajectory Prediction against Adversarial Attacks}, author = {Cao, Yulong and Xu, Danfei and Weng, Xinshuo and Mao, Zhuoqing and Anandkumar, Anima and Xiao, Chaowei and Pavone, Marco}, booktitle = {Proceedings of The 6th Conference on Robot Learning}, pages = {128--137}, year = {2023}, editor = {Liu, Karen and Kulic, Dana and Ichnowski, Jeff}, volume = {205}, series = {Proceedings of Machine Learning Research}, month = {14--18 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v205/cao23a/cao23a.pdf}, url = {https://proceedings.mlr.press/v205/cao23a.html}, 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).} }
Endnote
%0 Conference Paper %T Robust Trajectory Prediction against Adversarial Attacks %A Yulong Cao %A Danfei Xu %A Xinshuo Weng %A Zhuoqing Mao %A Anima Anandkumar %A Chaowei Xiao %A Marco Pavone %B Proceedings of The 6th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Karen Liu %E Dana Kulic %E Jeff Ichnowski %F pmlr-v205-cao23a %I PMLR %P 128--137 %U https://proceedings.mlr.press/v205/cao23a.html %V 205 %X 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).
APA
Cao, Y., Xu, D., Weng, X., Mao, Z., Anandkumar, A., Xiao, C. & Pavone, M.. (2023). Robust Trajectory Prediction against Adversarial Attacks. Proceedings of The 6th Conference on Robot Learning, in Proceedings of Machine Learning Research 205:128-137 Available from https://proceedings.mlr.press/v205/cao23a.html.

Related Material