Anomaly detection in roads with a data mining approach

Abstract

Road condition has an important role in our daily live. Anomalies in road surface can cause accidents, mechanical failure, stress and discomfort in drivers and passengers. Governments spend millions each year in roads maintenance for maintaining roads in good condition. But extensive maintenance work can lead to traffic jams, causing frustration in road users. In way to avoid problems caused by road anomalies, we propose a system that can detect road anomalies using smartphone sensors. The approach is based in data-mining algorithms to mitigate the problem of hardware diversity. In this work we used scikit-learn, a python module, and Weka, as tools for data-mining. All cleaning data process was made using python language. The final results show that it is possible detect road anomalies using only a smartphone.