Cheating Stereo Matching in Full-scale: Physical Adversarial Attack against Binocular Depth Estimation in Autonomous Driving

Researchers have developed the first physical adversarial attack targeting stereo-based depth estimation in autonomous vehicles, using 3D camouflaged objects that can fool binocular vision systems. The attack employs global texture patterns and a novel merging technique to create nearly invisible threats that cause stereo matching models to produce incorrect depth information.

• →First demonstrated physical adversarial attack specifically targeting stereo-based binocular depth estimation in autonomous driving systems.
• →Uses 3D adversarial examples with global camouflage texture rather than traditional 2D patches for better stealth and effectiveness.
• →Introduces new 3D stereo matching rendering module to handle disparity effects between stereo cameras.
• →Novel merging attack seamlessly blends adversarial objects into environments for enhanced stealth.
• →Successfully demonstrated ability to fool stereo models into producing erroneous depth information critical for autonomous vehicle safety.