Dynamic Deception: When Pedestrians Team Up to Fool Autonomous Cars

Masoud Jamshidiyan Tehrani, Marco Gabriel, Jinhan Kim, Paolo Tonella · 2026

Read Dynamic Deception: When Pedestrians Team Up to Fool Autonomous Cars by Masoud Jamshidiyan Tehrani, Marco Gabriel, Jinhan Kim, Paolo Tonella (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-18079.txt.

Description

Masoud Jamshidiyan Tehrani and colleagues study why adversarial attacks on autonomous-driving perception often fail in a full driving stack. Such attacks tend to be spatially or temporally short-lived, since vehicle dynamics blunt them, so they rarely change vehicle behavior. The authors answer with a system-level attack in which two pedestrians carry adversarial patches on their clothing and jointly amplify the effect through coordination and motion.

They evaluate the attacks in the CARLA simulator with a state-of-the-art autonomous driving agent. Single-pedestrian attacks fail in all ten runs, while dynamic collusion by two pedestrians forces full vehicle stops in up to half the runs. Static collusion yields no successful attack at all. The results show system-level failures arise only when adversarial signals persist over time and amplify through coordinated actors, exposing a gap between model-level robustness and end-to-end safety.

Catalog record

CREATORMasoud Jamshidiyan Tehrani, Marco Gabriel, Jinhan Kim, Paolo Tonella
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-18079

Browse the full library · About the Tower of Babel