University of California, Irvine researchers have identified a significant security flaw in autonomous drones that use camera-based target-tracking technology. The team demonstrated that by using an ordinary umbrella covered with a specific visual pattern, attackers can manipulate drones to approach closely enough to be captured or caused to crash.
The research group developed an attack framework named FlyTrap, which exploits weaknesses in the AI-driven tracking systems commonly found in commercial drones. These systems, often referred to as “active track” or “dynamic track,” are increasingly used for border control, security surveillance, and law enforcement purposes.
The findings will be presented at the Network and Distributed System Security Symposium in San Diego this week.
“Autonomous target tracking represents both tremendous potential and significant risk,” said Alfred Chen, assistant professor of computer science at UC Irvine. “While law enforcement and security agencies are adopting this technology for border patrol and public safety, it’s also being misused by criminals for stalking and other malicious purposes. Our work is the first comprehensive security study of this widely deployed technology.”
The research revealed what is described as a distance-pulling attack. In this scenario, the drone’s neural network misinterprets the umbrella’s pattern as a person moving away. To maintain its programmed distance from the target, the drone moves closer to the umbrella holder until it can be caught or crashed. Unlike attacks that only disrupt tracking, this method allows for physical elimination of drones through capture or collision.
Tests were conducted on three commercial drones: DJI Mini 4 Pro, DJI Neo, and HoverAir X1. The FlyTrap attack was successful in drawing these models close enough for capture or causing crashes. Vulnerabilities were disclosed to manufacturers DJI and HoverAir.
The paper notes that such attacks could allow criminals to evade detection by law enforcement drones or interfere with border patrol operations. Conversely, individuals targeted by unwanted drone surveillance could use similar methods to disable harassing aircraft.
“Our findings highlight urgent needs for security improvements in [autonomous target-tracking] systems before wider deployment in critical infrastructure,” said Shaoyuan Xie, lead author and graduate student researcher at UC Irvine. “If it’s that easy to seize control over an autonomous drone, operating them in public or in critical security or law enforcement settings should be reconsidered.”
The FlyTrap method works without external signaling or wireless data connections and is effective under various weather and lighting conditions due to its progressive manipulation of tracking algorithms through simple physical actions.
Comprehensive documentation about FlyTrap—including datasets, demonstration videos, metrics, social media content, and an extended paper—is available to support future improvements in drone security. All testing was completed before December 22, 2025.
In addition to Chen and Xie, contributors included former doctoral students Ningfei Wang and Takami Sato; current graduate students Mohamad Habib Fakih, Junchi Lu, Fayzah Alshammari; postdoctoral scholar Halima Bouzidi; and Professor Mohammad Abdullah Al Faruque from electrical engineering and computer science. Funding was provided by NASA and the National Science Foundation.
UC Irvine operates within the University of California system as a public research university located in Irvine since 1965 (official website). The university is recognized for very high research activity (official website) with accredited programs (official website), offering undergraduate and graduate degrees across various fields (official website). It engages globally through partnerships (official website) while emphasizing inclusive excellence (official website).
For further information on UC Irvine news or resources for journalists visit https://news.uci.edu/media-resources.