Scholarlink

<p>In large passenger aircraft, fuel management is automated, but in small trainer aircraft, pilots must manually configure the fuel selector valve (FSV). Forgetting to open the FSV during pre-flight checks is a critical human-factor risk that can lead to engine failure after takeoff. Currently, small training aircraft lack a dedicated warning system for this specific oversight. This study presents a novel, non-invasive warning system to mitigate this hazard. The system uses a cockpit-mounted camera and an embedded NVIDIA Jetson Nano to run a YOLOv5-based computer vision model that detects the FSV’s position in real-time. If a hazardous state (CLOSED or FAULTY) is identified, the pilot receives simultaneous audible and vibrotactile alerts. When tested on Cessna 172 and Viper SD4 aircraft, the system demonstrated high operational reliability. A frame-by-frame analysis of real-time demonstrations showed an overall accuracy of 90.9%. Most importantly, for the safety–critical CLOSED position, the system achieved perfect precision (100%) and a recall of 96.9%. This study provides a practical and cost-effective solution to a persistent safety gap in general aviation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Image processing-based warning system for preventing the fuel selector valve from remaining closed in small trainer aircraft

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月25日

查看全文
  • İbrahim Halil Çoban,
  • Fatih Alpaslan Kazan

摘要

In large passenger aircraft, fuel management is automated, but in small trainer aircraft, pilots must manually configure the fuel selector valve (FSV). Forgetting to open the FSV during pre-flight checks is a critical human-factor risk that can lead to engine failure after takeoff. Currently, small training aircraft lack a dedicated warning system for this specific oversight. This study presents a novel, non-invasive warning system to mitigate this hazard. The system uses a cockpit-mounted camera and an embedded NVIDIA Jetson Nano to run a YOLOv5-based computer vision model that detects the FSV’s position in real-time. If a hazardous state (CLOSED or FAULTY) is identified, the pilot receives simultaneous audible and vibrotactile alerts. When tested on Cessna 172 and Viper SD4 aircraft, the system demonstrated high operational reliability. A frame-by-frame analysis of real-time demonstrations showed an overall accuracy of 90.9%. Most importantly, for the safety–critical CLOSED position, the system achieved perfect precision (100%) and a recall of 96.9%. This study provides a practical and cost-effective solution to a persistent safety gap in general aviation.