SlowTrack: Increasing the Latency of Camera-Based Perception in Autonomous Driving Using Adversarial Examples
Summary
In Autonomous Driving (AD), real-time perception is a critical component responsible for detecting surrounding objects to ensure safe driving. While researchers have extensively explored the integrity of AD perception due to its safety and security implications, the aspect of availability (real-time performance) or latency has received limited attention. Existing works on latency-based attack have focused mainly on object detection, i.e., a component in camera-based AD perception, overlooking the entire camera-based AD perception, which hinders them to achieve effective system-level effects, such as vehicle crashes. In this paper, we propose SlowTrack, a novel framework for generating adversarial attacks to increase the execution time of camera-based AD perception. We propose a novel two-stage attack strategy along with the three new loss function designs. Our evaluation is conducted on four popular camera-based AD perception pipelines, and the results demonstrate that SlowTrack significantly outperforms existing latency-based attacks while maintaining comparable imperceptibility levels. Furthermore, we perform the evaluation on Baidu Apollo, an industry-grade full-stack AD system, and LGSVL, a production-grade AD simulator, with two scenarios to compare the system-level effects of SlowTrack and existing attacks. Our evaluation results show that the system-level effects can be significantly improved, i.e., the vehicle crash rate of SlowTrack is around 95% on average while existing works only have around 30%.
Autonomous Driving (AD) Visual Perception
AD visual perception consists of object detection and object tracking
Research Gap
Autonomous Driving (AD) Perception is safety-critical
Many prior works have studied its security, especially on integrity: e.g., object evasion attack
Cause traffic rule violations or crashes, which is called system-level effect
However, availability aspect has been relatively underexplored
While some prior works have studied availability in object detection, they do not encompass entire AD perception, including both object detection and tracking
Research gap: their proposed attack strategies may not be effective enough to conduct system-level effects.
Problem Formulation & Tracking Analysis
SlowTrack Attack Design
Attack Effectiveness Evaluation
End-to-end Simulation Evaluation Results
Research Paper
Chen Ma∗, Ningfei Wang∗ (co-first authors), Qi Alfred Chen, Chao Shen
To appear in the 38th Annual AAAI Conference on Artificial Intelligence (AAAI 2024)
BibTex for citation:
@inproceedings{ma2024slowtrack,
title={{SlowTrack: Increasing the Latency of Camera-Based Perception in Autonomous Driving Using Adversarial Examples}},
author={Ma, Chen and Wang, Ningfei and Chen, Qi Alfred and Shen, Chao},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
volume={38},
number={5},
pages={4062--4070},
year={2024}
}
Team
Chan Ma, Master student, Xi'an Jiaotong University
Ningfei Wang, Ph.D. student, University of California, Irvine
Qi Alfred Chen, Assistant Professor, University of California, Irvine
Chao Shen, Professor, Xi'an Jiaotong University
Acknowledgments
This research was supported in part by National Key R&D Program of China (2020AAA0107702); the NSF under grants CNS-1929771, CNS-2145493, and CNS-1932464; National Natural Science Foundation of China (U21B2018, 62161160337, 62132011, 62376210, 62006181, U20B2049); Shaanxi Province Key Industry Innovation Program (2021ZDLGY01-02); and Fundamental Research Funds for the Central Universities under grant (xtr052023004, xtr022019002).