PhySense: Defending Physically Realizable Attacks for Autonomous Systems via Consistency Reasoning

This repository hosts the source code for the paper "PhySense: Defending Physically Realizable Attacks for Autonomous Systems via Consistency Reasoning". The paper has been accepted by the 31st ACM Conference on Computer and Communications Security (CCS), 14-18 October 2024.

 

PhySense is a defense-in-depth system against physical realizable adversarial attacks. The key approach relies on reasoning, empowered by statistical modeling, robust physical rules, and pipelining techniques to ensure reliable and timely defense. PhySense not only detects malicious objects but also provides the potential true labels to correct misclassifications.

 

If you find our project useful, please cite us at: