Authors: Xiao Li*, Department of Geography, Texas A&M University, Daniel W Goldberg, Department of Geography, Texas A&M University, Da Huo, Department of Geography, Texas A&M University
Topics: Transportation Geography, Geographic Information Science and Systems, Urban Geography
Keywords: Pothole Detection, Crowdsourcing, Mobile Devices, Wavelet Analysis
Session Type: Paper
Start / End Time: 3:05 PM / 4:45 PM
Room: Roosevelt 5, Marriott, Exhibition Level
Presentation File: No File Uploaded
Road surface condition assessment plays an essential role in traffic network maintenance and management. Potholes, as one of the most important criteria in road surface assessment, not only impact road quality, fuel consumption but even traffic safety. Many previous studies have been successfully carried out for identifying potholes; however, the measurement of pothole size using mobile devices is still unexplored. How to identify damaging potholes (large enough to cause traffic risk) is more practical and meaningful for road maintenance requiring more attention.
In the study, we created a labor-saving and low-cost crowdsensing system for pothole detection and measurement. A smartphone application PotholeAnalyzer was designed to collect, process measurements and calculate potential potholes. Meanwhile, we innovatively proposed a two-phase pothole detection framework, which combines threshold techniques and wavelet analysis. First, the potential potholes were extracted by performing threshold techniques using mobile devices. Then, the detected potential potholes and raw data were fed to wavelet analysis. Each potential pothole was used as the center to clip a 50-sample snippet from the raw acceleration dataset. Wavelet analysis was implemented to check each acceleration snippet to verify the detected potholes and measure their sizes. Field test results demonstrated that our innovatively proposed two-phase pothole detection framework is able to effectively detect the potholes and measure their sizes. By implementing a crowdsensing solution and fuzzy logic, the proposed method can accurately locate the damanging potholes and continuously monitor the pothole growth.