Foreign object debris detection in lane images using deep learning methodology.

Area of Science:

• Computer Science
• Aerospace Engineering
• Artificial Intelligence

Background:

• Foreign object debris (FOD) poses a significant safety risk to aircraft operations.
• FOD-related damage incurs substantial annual costs exceeding $4 billion.
• Current FOD detection methods, like radar and camera surveillance, are costly and labor-intensive.

Purpose of the Study:

• To develop a cost-effective, video-based deep learning methodology for foreign object debris detection.
• To improve the accuracy and efficiency of identifying and locating FOD on airport runways.
• To reduce the financial and operational burden associated with traditional FOD clearance.

Main Methods:

• A two-module deep learning system was proposed for FOD detection: object classification and object localization.
• The classification module identifies specific types of foreign objects.
• The object localization module precisely pinpoints detected FOD within video frames.

Main Results:

• The video-based system demonstrated improved accuracy and robustness in experimental tests with a large dataset.
• The methodology enables rapid detection and removal of foreign objects.
• The system offers a more efficient and potentially less expensive alternative to current FOD detection technologies.

Conclusions:

• The proposed deep learning approach offers a viable and effective solution for FOD detection.
• Implementing this system can significantly enhance airport runway safety and operational efficiency.
• This technology has the potential to reduce the economic impact of FOD-related damages.