Fooling Perception via Location: A Case of Region-of-Interest Attacks on Traffic Light Detection in Autonomous Driving

End-to-End Attack Demos:

Experimental Setup:

• Baidu Apollo Version: r5.0.0

  • Enabled Modules: Localization, Perception, Prediction, Planning, Routing, Control, Transform, Dreamview.

  • Attack Implementation: implemented as an independent module in Baidu Apollo to intercept GPS messages and apply spoofing distances.

• LGSVL Simulator Version: 2019.10

  • HD Map: Shalun

    • Red Light Detection Scenario: The current intersection is in red light and the next intersection is in green light.

    • Green Light Detection Scenario: The current intersection is in green light and the next intersection is in red light.

  • AV Model: Lincoln MKZ 2017

Research Paper:

[AutoSec 2021] Fooling Perception via Location: A Case of Region-of-Interest Attacks on Traffic Light Detection in Autonomous Driving

Kanglan Tang, Junjie Shen, Qi Alfred Chen

Appeared in the Third International Workshop on Automotive and Autonomous Vehicle Security (AutoSec) 2021.

BibTex for citation:

@inproceedings{autosec:2021:kanglan:roiattack,

title={{Fooling Perception via Location: A Case of Region-of-Interest Attacks on Traffic Light Detection in Autonomous Driving}},

author={Kanglan Tang and Junjie Shen and Qi Alfred Chen},

booktitle={Third International Workshop on Automotive and Autonomous Vehicle Security (AutoSec) 2021},

year={2021},

month = {February}

}