UAV and AI Application for Runway Foreign Object Debris (FOD) Detection

There are several ways in which Foreign Object Debris (FOD) are detected on runways. Some of these methods include Radar, infrared technologies, and stationary cameras mounted on the runway and use image processing tools to find these FODs. Radar technology is highly accurate when finding FODs but is highly inaccurate with small FOD items causing a high false-positive rate. Stationary based RGB camera-based methods also have a high false-positive rate prompting the shutdown of runways, creating disruptions for both the airport and the airline carriers. The paper presents a new method of detection by using an Unmanned Aerial Vehicle (UAV) to fly above the runway at a low altitude (e.g. <; 30 m) to find FOD in. We developed a system that combines a UAV, an RGB camera, an Artificial Intelligence (AI) detector trained using deep learning methods and locally collected images over a runway. The classes specifically looked at were paper, metal, bolts, plastic, and plastic bottles. Different lighting conditions of both full sunlight and cloudy weather were taken into consideration when the images were collected. The detector was trained with various data augmentation techniques including resize, rotate, and colour augmentation. Results have concluded that there is a potential use for UAV's as a method of FOD detection, with a high rate of accuracy in the detections. This could lead to shorter timeframes and fewer disruptions where runways are closed.