News and views for the UB community
Campus News
Team Cacti places fifth at Baidu Autodriving Capture the Flag competition
Team Cacti placed fifth worldwide and first in the U.S. in the DEFCON AutoDriving Capture the Flag competition organized by Baidu. Team members are, from left, Assistant Professor Ziming Zhao, Zheyuan Ma, Nishant Vishwamitra, Sabiha Shaik, Xi Tan, Feng Wei, Qiqing Huang and Associate Professor Hongxin Hu. Photo: Madison Dailley
By NICOLE CAPOZZIELLO
Published November 23, 2021
A team of computer science and engineering graduate students placed fifth worldwide and first in the U.S. in the DEFCON AutoDriving Capture the Flag (CTF) competition organized by Baidu.
Over 90 teams took part in the event, which focuses on autonomous driving security.
“This was our first time participating in the AutoDriving CTF and we are very proud of the team,” says Ziming Zhao, assistant professor in the Department of Computer Science and Engineering, and one of the team’s advisers.
“Congratulations to the whole team,” adds Jinhui Xu, professor and chair of the Department of Computer Science and Engineering. “High-impact international competitions like these are a great way to showcase the talented students in our department and at the University at Buffalo.”
With safety as the No.1 priority for autonomous driving, the contest encouraged participants to exploit and defend autonomous driving vehicles by all means, including, but not limited to, robust physical adversarial attacks and defenses in the form of camouflage stickers and road graffiti; adversarial road patches; sensor spoofing/hijacking; and multi-sensor fusion manipulations.
“Competing alongside some of the leading cybersecurity researchers from around the globe in these CTFs competitions helps build our confidence as cybersecurity researchers,” says Nishant Vishwamitra, a PhD student in the Department of Computer Science and Engineering and leader of Team Cacti.
Simulation of an autonomous vehicle (AV) traveling on a road. The goal was to make the AV deviate laterally to create a dangerous road situation. The figure shows how this can be done using several techniques ranging from simple re-drawing of lane lines to complex adversarial machine learning road patch attacks.
In preparation for the competition, the team met regularly to study and practice on example challenges and the 2020 version of the competition.
“On the day of the CTF, our team had a clear division of tasks, which helped us be successful,” says Zheyuan Ma, a PhD student and member of Team Cacti.
“A key part of studying security is putting skills to the test in practice. Hacking challenges known as CTF competitions are a great way to do this,” says Zhao.
“These competitions push us to think as efficient cybersecurity researchers, as hackers and as forensic investigators," adds Vishwamitra. “We learn transferrable skills that we can apply to any cybersecurity research area.”
Hongxin Hu, associate professor and Team Cacti’s other adviser, notes that for students, this kind of competition “is a really good chance for them to gain hands-on skills. It can also help them find new research problems.”
One student in the lab is already working on a research direction directly motivated by the CTF competition.
In the future, Team Cacti plans to participate in more software, hardware and CTF competitions. “We hope to recruit more undergraduate students and underrepresented students to our team,” Hu adds.
Prior to this competition, the team placed ninth during the spring semester in the MITRE Embedded CTF competition.
In addition to Vishwamitra and Ma, other team members include PhD students Xi Tan and Qiqing Huang, and master’s students Shaik Sabiha and Keyan Guo.
The competition was held in August as part of DEFCON 29. It was sponsored by BCTF (Baidu Capture The Flag) and the University of California at Irvine.