Smartphone road surface monitoring system

Abstract

In view of the EU’s long-term vision of zero fatalities in road transport by 2050, transport authorities require accurate, up-to-date, and easy-to-use technologies to help them address road anomalies. Smartphone devices offer a compelling solution because of their availability and various inbuilt sensors. With this in mind, this study shows how smartphone devices can help identify road anomalies. In our study, we used the smartphone’s inertial measurement unit (IMU) and camera sensors to track road anomalies. This study uses two identical smartphone devices to capture the data simultaneously. The first hosts the IMU sensing application, whilst the second hosts the object detection component. The IMU road anomaly detection is based on a supervised machine learning model. Meanwhile, the vision solution employs a lightweight object detection architecture to detect anomalies in real-time. In addition, the GPS data provides the backbone to synchronise both applications and also pinpoint the anomaly locations. Finally, we use a simple data fusion technique to merge the IMU and vision results into one reporting system. This simple yet powerful reporting solution allows operators to assess the accuracy of both systems and also provides a richer reporting data set. Our study finds that a dual system yields more conclusive results than conventional mono-sensing systems, as the system provides a straightforward method for verifying both the IMU and vision predictions.