基于Yolov8和RK3588的无人机侵入行为监测、告警与干扰系统

A Drone Intrusion Monitoring, Alert, and Jamming System Based on YOLOv8 and the Rockchip RK3588 Platform

摘要: 系统采用“感知–决策–执行”一体化和“视觉引导射频”协同作战架构：多目全景视觉系统结合SURF拼接与去雾算法，实时生成180˚无缝全景视频感知；RK3588 NPU部署优化的YOLOv8 (DCNv2)与重匹配跟踪算法，实现无人机精准识别与锁定；自主设计的便携式多频段射频干扰枪进行高效反制，并集成广和通L610 LTE模块实现远程告警与物联网联动。该系统实现从无人机发现到智能反制的闭环处置，适用于关键基础设施、大型活动安保等，具有显著应用价值。

Abstract: In response to the security threats posed by unauthorized flights of low-altitude, low-speed, and small unmanned aerial vehicles and the limitations of traditional security methods, this project develops an intelligent low-altitude defense system named “SkyShield Intelligence” based on the Rockchip RK3588 chip. The system adopts an integrated “perception - decision - execution” architecture and a collaborative strategy of vision-guided radio frequency countermeasures. A multi-camera panoramic vision system combined with Speeded-Up Robust Features image stitching and image dehazing algorithms generates a real-time 180-degree seamless panoramic video for environmental perception. The neural processing unit of the Rockchip RK3588 deploys an optimized You Only Look Once Version 8 model integrated with Deformable Convolutional Networks Version 2 and a re-identification tracking algorithm to achieve accurate detection and locking of unmanned aerial vehicles. A self-designed portable multi-band radio frequency jamming gun is used for efficient countermeasures, and a Fibocom L610 Long Term Evolution communication module is integrated to enable remote alarms and Internet of Things linkage. The system forms a closed-loop response from unmanned aerial vehicle detection to intelligent countermeasures, making it suitable for the protection of critical infrastructure and security for large-scale events, demonstrating significant practical application value.