FlyTrap: Physical Distance-Pulling Attack Towards Camera-based Autonomous Target Tracking Systems
Shaoyuan Xie
Network and Distributed System Security (NDSS) Symposium 2026 · Day 3 · Evasion Attacks
FlyTrap is the first **distance-pulling attack** against camera-based autonomous drone tracking systems. By printing adversarial patterns on an ordinary **umbrella** ($20 on Amazon), an attacker can trick a tracking drone into progressively moving closer to the target by spoofing a smaller bounding box -- causing the drone's control loop to misinterpret the target as moving farther away. The attack achieves **100% success rate** at pulling drones within 0.5 meters in closed-loop experiments and transfers to three commercial drones: **DJI Mini4 Pro**, **DJI Neo**, and **Hover Air X1**.
AI review
A beautifully executed physical adversarial attack that turns a $20 umbrella into a drone countermeasure. The progressive distance-pulling design that simulates the entire closed-loop control process during optimization is technically elegant, the 100% success rate at 0.5m is decisive, and the transfer to three commercial DJI/Hover Air drones makes this immediately real. This is adversarial ML at its best: a clear threat model, a novel attack vector, working physical-world demos, and validated against deployed commercial systems.