Abstract

Due to the rapid development of drone manufacturing technologies and their ease of use, many people can own and use drones for services such as photography, entertainment, scientific, security surveillance, and so on. However, terrorists and criminals have exploited drones in attacks on important places and vital institutions due to a lack of protection from anti-drone systems. This paper presents an anti-drone system based on computer vision using the You Only Look Once (YOLO11) algorithm to detect, track, and destroy unauthorized drones. The proposed system leverages computer vision techniques to detect drones in real-time with high accuracy and low latency. The system integrates the YOLO11 algorithm with automatic weapon control, enabling precise targeting and destruction of detected drones. The authors designed a new mathematical synchronization method to align firearm movements with drone coordinates for improved tracking performance. We implemented a prototype of the proposed project that can detect and track an unknown drone using a Raspberry Pi as a microcomputer, issue an audible alarm upon detection using an Arduino, track the drone, and destroy it using a robot arm as an automatic weapon that moves based on coordinates from the detection data. The practical experiments demonstrated high detection accuracy (precision: 95.7%, recall: 91.5%, mAP50: 91.1%, and FPS: 29.6), outperforming previous YOLO versions and other existing anti-drone solutions.