Influence of Vehicle Make on Accuracy of Real-time Road Anomaly Identification

Abstract

As road infrastructure in the United States is aging, road anomalies such as cracks, potholes, and other abnormalities are becoming much more prevalent. Currently there is no real-time understanding of the conditions of roads, thus we developed a machine-learning algorithm developed and trained to identify road conditions in real time based on data collected by smartphones. Since there are a multitude of different vehicles on the roads and locations where phones can be placed in the vehicle, creating a classification algorithm that can work regardless of the vehicle type and phone placement is incredibly important. Doing a comparative study on the different vibrations received at different locations in different vehicles will provide a baseline for future development of a universal algorithm that uses crowd sourced data from cell phones to allow for real-time awareness of changing road conditions. This in turn provides a way to identify and fix dangerous road anomalies quickly.