Classifying road surface conditions using vibration signals.

The paper aims to classify road surface types and conditions by characterizing the temporal and spectral features of vibration signals gathered from land roads. In the past, road surfaces have been studied for detecting road anomalies like bumps and potholes. This study extends the analysis to detect road anomalies such as patches and road gaps. In terms of temporal features such as magnitude peaks and variance, these anomalies have common features to road anomalies. Therefore, a classification method based on support vector classifier is proposed by taking into account both the temporal and spectral features of the road vibrations as well as factor such as vehicle speed. It is tested on a real data gathered by conducting a smart phone-based data collection between Thailand and Cambodia and is shown to be effective in differentiating road segments with and without anomalies. The method is applicable to undertaking appropriate road maintenance works.