Detailed Pavement Inspection of Airports Using Remote Sensing UAS and Machine Learning of Distress Imagery

Historically, airfield pavement has and continues to be important to the nation’s commerce. It is clear that annual detailed pavement inspections of airport pavements are a critical part of airfield pavement management programs (PMP). In the 1970s, inspections were done with pen, paper, and rulers. In the 1990s, detailed inspections were conducted using tablet computers, GPS, and georeferenced photography. In the 2000s, pavement inspections were conducted using trucks with cameras, laser crack scanners, and even LIDAR with onboard inertial reference systems. The next leap in technology is using remote sensing UAS (UAS) imagery with high resolution data (sub 2 mm) and artificial intelligence (AI) processing of this data finding distresses to identify and catalog distress types, severity, size, and GIS data. We have collected UAS high resolution imagery for 33 airports in 14 states and in all nine FAA Regions. We have developed proprietary machine artificial intelligence neural network deep learning tools to teach the computer to identify pavement distresses and calculate lengths and areas, and establish a polygon of distress. After the AI model has identified each distress and assigned a confidence level, pavement analysts review the results and confirm the distress ID, size, shape, and severity followed by an update to the AI model. The authors have also developed digital elevation models to show the depth of 3D distresses such as rutting and depressions. All of this data is then be used to calculate a pavement condition index using the methodology of ASTM D5340. We have delivered imagery data and GIS location and PCI calculation for over 33 airports measuring 1,001,069,776 sq ft of pavement at a cost of $0.002 sq ft for 100% data collection. In contrast, traditional boots on the ground projects of these same airports may only have measured 10% of the pavement (100,106,977 sq ft) at a reported cost of $0.04 sq ft. We have successfully compared the UAS distress data with traditional data collection methods with world leading engineering firms, and the UAS results are repeatable, measurable, provided faster, and cost 88% less in a direct per square foot comparison and 1,000 times faster.