AI Enhances Foreign Object Debris (FOD) Detection

“In 2010, detection technology was the human eyeball,” according to Steve Boyle, CEO of Essential Aero, which develops AI-enabled aerial and ground-based inspection platforms. “People would walk and scan pavement looking for FOD.”

Unfortunately, manual FOD detection has multiple built-in inefficiencies. Monitoring a sprawling network of runways, taxiways and aprons is time-consuming and labor-intensive. Adverse weather conditions can reduce visibility and hinder the visual detection of small debris.

Transforming FOD Detection

Addressing these challenges requires the adoption of innovative technologies, such as artificial AI-supported drone-based surveillance and advanced sensor systems.

“More recently, new technologies were introduced, first using computer vision and now using artificial Intelligence and machine learning (ML) to detect FOD present in images captured using LiDAR and visual range cameras,” says Boyle.

By harnessing the power of algorithms and computer vision technologies, airports can enhance FOD detection accuracy, speed up response times, and minimize the risk of potential runway hazards. For instance, advanced sensor systems integrated into drones can employ optical cameras for real-time FOD detection, allowing for accurate identification and classification of potential hazards.

Algorithms for FOD Detection

AI-based frameworks have been developed to provide highly accurate and timely detection of FOD, enabling swift removal from airport surfaces.

The Feature-Fusion YOLO (FF-Yolo) real-time object detection algorithm is a prime example of AI innovation in FOD detection. This intelligent computer vision system accelerates the identification and localization of FOD on airport surfaces.

In addition, ML models can be utilized to identify anomalies in the visual data captured by drones. By learning the normal patterns of runway surfaces, the models can detect any deviations that may indicate the presence of FOD, enabling proactive identification of potential hazards.

Training Data

The integration of AI with drone technologies equipped with optical cameras and microcontrollers opens up exciting possibilities for revolutionizing FOD detection at airports, enhancing surveillance capabilities, identifying FOD in real-time, and streamlining the removal process with greater efficiency.

However, the key component to AI FOD detection is not the AI itself, but the “training data” fed into it, according to Boyle. “This consists of images with labels – an image of a bolt that is labeled as ‘bolt’ and ‘metal’ can be used to train an ML model to detect a bolt on the pavement and then correctly classify it to an object type and material. The entity with the richest training data will have the best detection product.”

The dataset should cover a variety of weather conditions and lighting levels to ensure the model’s robustness under different environmental scenarios. Moreover, the training data should also include images that depict runway and taxiway backgrounds, as well as different perspectives and angles of FOD objects.

Future Perspectives

As airports continue to prioritize safety measures and efficiency in their operations, new technologies will expand the role of artificial intelligence in FOD detection

For instance, integrating data from multiple sensors, such as optical cameras, LiDAR, radar, and hyperspectral imaging can enhance detection accuracy and reliability while minimizing false alarms. Leveraging autonomous drone technology can enable agile and efficient FOD detection across large airport areas, with minimal human oversight.

The integration of detection algorithms, drone fleets and other cutting-edge technologies into FOD control systems represents a game-changer for airport safety. With AI-powered solutions driving innovation, airports can stay ahead of potential hazards, maintain operational excellence, and uphold the highest standards of safety and security in the aviation industry.