Methods for Automated Identification of Roadway Drainage-Related Features from Mobile LiDAR Data

Light Detection and Ranging (LiDAR) systems have been increasingly used in project planning, project development, construction, operations, maintenance, and asset management. Typical data collected by a LiDAR system include slant distance, incidence angle, and reflectivity measurements. This research focuses on mobile LiDAR systems (MLSs). Processing of large amounts of data collected by MLSs remains tedious and time-consuming. For MLSs to be used efficiently in roadway drainage inventory and condition assessment, automated methods are needed to identify key features that affect drainage. The aim of this research is to develop computational methods for automated identification of such features from data collected through MLSs. The specific objectives of this research are to a) detect pavement surface type, b) detect the presence of driveways and underlying pipes and extract count, width, elevation difference and material cover and c) detect roadside features such as grass-cover area, curb location, and curb height based on the data collected using a SICK LMS-5XX series LiDAR scanner and hardware and software by Road Doctor. Reflectivity, measured as a logarithmic index of power level called received signal strength indicator (RSSI), is used to develop an algorithm to detect surface type based on statistical analysis of RSSI. Cross-sectional geometry, along with material identification, is used to identify driveways and underlying pipes. RSSI distribution and material identification techniques are used to detect roadside grass areas. Elevation distribution and filter template technique are used to detect curbs. Each method was tested and validated using data from actual road sections in Texas. The ability to detect aforementioned features reliably using automated means is an initial step to further the cause of MLS acceptance and implementation. Generally, the accuracies of pavement and grass detection methods were at least 83%. The effect of reflectivity attenuation is pronounced for roadside. Therefore, in order to develop a reliable grass detection method, attenuation correction is required. It is possible to detect driveways and distinguish them from topographical features using a combination of elevation cross sections, material detection, and surface smoothness. It is possible to identify curbs using filter template technique