Asphalt Road Layer Detection for Construction Progress Monitoring

Abstract

Transportation construction projects consistently underperform, with an estimated $82.6 billion globally in annual cost overruns. Logistical challenges associated with the size and location of transportation construction sites is a contributing factor, as is the inefficiency of current progress monitoring practice. A method that leverages the rich 3D information available in Civil Infrastructure Models (CIMs) and accurate 3D reality capture technologies, like LIDAR or photogrammetry, could address these shortfalls. An essential task in implementing such a progress monitoring approach is the detection of relatively thin design surface layers in 3D as-built data. This paper proposes a method for accomplishing this detection and presents experimental results on as-built data collected during the construction of a small residential road in Cambridge, UK. A total of 640 experiments were run for different combinations of parameters and classification rules, producing a peak accuracy of 86.62%, peak precision of 80.65%, and peak recall of 92.50%. The most balanced combination of parameters and classification rule produced an accuracy of 86.50%, precision of 68.17%, and recall of 60.99%.