Estimation of Gravel Roads Ride Quality Through an Android-Based Smartphone

Abstract

This study demonstrated the ability of smartphone sensors in evaluating gravel roads conditions. Seventy gravel roads with various conditions, surface materials, and geometric features were included in this study. The analysis was based on signal demodulation and wavelet transformation to reduce the effect of many external factors (i.e., speed dependency, engine vibrations, and suspension system) that may affect the obtained measurements. It was found that the acquired signals from a smartphone accelerometer can reflect the actual conditions of a gravel road. In addition, the location and the severity of surface deteriorations such as potholes could be identified. A regression model (R² = 0.78) based on the acquired signals from smartphones was developed to predict the overall rating of the gravel road condition according to the Riding Quality Rating Guide (RQRG) system. An initial validation analysis, conducted on 35 new gravel roads, showed that this model was able to return reasonable ratings. Also, the statistical analysis showed that any difference between the predicted and the actual ratings of <1.3 was not significant. The proposed methodology can be considered as a baseline for building a low cost crowdsourcing platform that helps local agencies in managing their inventory of gravel roads.