Best Demo Award at the 22nd ACM Conference on Embedded Networked Sensor Systems (SenSys 2024)

Assoc Prof Tan Rui (right) and Guo Dongfang, CCDS researcher (middle) at the award ceremony together with the award presenter – SenSys 2024 demo co-chair, Asst Prof Guo Xiuzhen (Zhejiang University). Inset photo: the other two team members are Dai Yimin (left) and Wu Yuting (right).

NTU CCDS Researchers Guo Dongfang, Wu Yuting, Dai Yimin and Associate Professor Tan Rui, in collaboration with Zhou Pengfei from University of Pittsburgh and Lou Xin from Singapore Institute of Technology, won the Best Demo Award at the 22nd ACM Conference on Embedded Networked Sensor Systems (SenSys 2024) held in Hangzhou, China. 

Guo Dongfang (left) demonstrating the system to the audience.

The award-winning demo, titled "Invisible Adversarial Stripes against Traffic Sign Recognition in Autonomous Driving," showcases their system, published earlier this year in ACM MobiSys 2024. This research designs GhostStripe, a physical adversarial attack system that utilizes light-emitting diodes and exploits the rolling shutter effect of cameras to create adversarial colour stripes that can consistently mislead AI-driven traffic sign recognition, while remaining imperceptible to human observers in the physical world.

Guo Dongfang presenting the one-minute madness talk.

At the conference, lead author Guo Dongfang gave a one-minute madness presentation, demonstrated the GhostStripe system, and finally won the Best Demo Award. This recognition highlights the team's research in autonomous driving perception, drawing attention to potential vulnerabilities in both camera sensors and AI perception models, while advancing understanding of robust and secure intelligent systems.

SenSys is a top conference focused on the challenges and innovations in networked sensing and actuation systems, which enable high-density, high-fidelity sensing for diverse applications.

Details of the work:

• Demo abstract:

  Demo: Invisible Adversarial Stripes against Traffic Sign Recognition in Autonomous Driving.
  Dongfang Guo, Yuting Wu, Yimin Dai, Pengfei Zhou, Xin Lou, Rui Tan.
  The 22nd ACM Conference on Embedded Networked Sensor Systems (SenSys), November 4-7, 2024, Hangzhou, China.
  https://dl.acm.org/doi/10.1145/3666025.3699401

• Full paper:

  Invisible Optical Adversarial Stripes on Traffic Sign against Autonomous Vehicles.
  Dongfang Guo, Yuting Wu, Yimin Dai, Pengfei Zhou, Xin Lou, Rui Tan.
  The 22nd ACM International Conference on Mobile Systems, Applications, and Services (MobiSys), June 3-7, 2024, Tokyo, Japan.
  https://dl.acm.org/doi/10.1145/3643832.3661854    
  

Read more on ACM website.