RoadProbe: A Machine Learning-based System for Predictive Road Maintenance

Machine learning (ML) is enabling a wide range of IoT applications. In this paper, we explore the use of crowdsourcing and machine learning for more cost-effective alternatives to road maintenance. Specifically, we present RoadProbe, a system that uses machine learning for the detection and prediction of road irregularities (e.g., potholes, cracks). RoadProbe uses crowdsourcing to collect data through a mobile application. The collected data is first stored on Google’s Firebase. We then use the collected data to build a LSTM model to predict road irregularities. To assist in decision making, we present the obtained predictions on Google Maps. Our evaluation shows that the model has a high prediction accuracy.