On Saturday, Sept. 26, 2020, Baidu Security hosted the world’s first capture the flag (CTF) competition focused on autonomous driving security. Twenty-four teams from around the world competed in AutoDriving CTF, and UCI’s ASGuard (Autonomous Systems Guard) team took the top prize.

“I was very excited to compete in the world’s first CTF hacking competition on the topic of autonomous driving security,” says Assistant Professor of Computer Science Qi Alfred Chen, who led the team. “As the development and deployment of autonomous driving technology become increasingly mature and practical, it is imperative to ensure its security against malicious actors.”

Chen’s Ph.D. student Junjie Shen served as team captain, collaborating with fellow Ph.D. students Ningfei Wang, Takami Sato, Ziwen Wan, and Yunpeng Luo and with UCI alumnus Zeyuan Chen (now a graduate student at Carnegie Mellon University). The team battled for 24 hours, facing a variety of challenges, such as confronting an AI system working to trick the car’s ability to detect obstacles. In the end, ASGuard bested well-known CTF teams from both academia and industry, securing their victory in the competition’s final hour.

“The problems in this CTF were very interesting and unique,” says Shen. For example, the “Mad Race” challenge asked the teams to implement driving agents to control the vehicles and compete in a race, while another challenge required submitting an “adversarially generated patch” for a truck blocking the road. “Being able to solve those problems was really fun and meaningful to us as they represent the real-world challenges that autonomous driving is facing.”

Chen credits the team’s research for preparing them for the competition. “I think our winning is mainly because my group is among the first few in the security community to conduct research on autonomous driving systems security, so my students have the right expertise and domain knowledge for this competition,” says Chen. Earlier this year, Chen and his group conducted a study of high-level self-driving car systems and discovered a “take-over vulnerability” that, if exploited, could result in serious accidents.

“I hope that we can have more CTF competitions on this topic in the future,” says Chen, “so that more people can be aware of — and join timely efforts in securing — emerging autonomous driving technology.”

— Shani Murray