All in all, we utilized an embedded system which is based on Raspberry pi and integrates sensors including an accelerometer and a GPS module to collect vibration data of the infrastructure. The infrastructure is divided into 12 different road conditions and we collect the corresponding data for thresholds determination. In data analysis module, we used the differential method to minimize the mistake made by the drift of the sensor.
In the future, this project could continue to develop the corresponding machine learning algorithm and make the system more perfectly.