Road irregularities detection and signalling with sonar and accelerometer using arduino

Road anomalies represent one of the most significant causes of accidents for vulnerable 2-wheeler users. Many actions have been taken to reduce the level of risk, from new regulations and laws to analyzing roads to point authorities to the most damaged road sections. However, nothing has been changed to mitigate the risk directly in the vehicle equipment. In this thesis, we create an experimental set-up in a city bicycle using accelerometers and a sonar sensor mounted on Arduino boards to detect and analyze the road surface, eventually mapping any anomalies. Starting from the creation of the system by reasoning about the hardware characteristics of the bicycle, we proceed to collect data in an intelligent way to create different datasets with different properties covering different types of road surfaces. Once we understand the behavior of the collected measures, we analyze the dataset with an offline approach, extrapolating some metrics to discern road anomalies from smooth segments; in particular, we obtain good results using the variance of the dataset as a threshold, proposing an adaptive window comparison with local variance. Next, keeping the focus on creating a real-life use case, we try to detect road anomalies directly on board using Arduino, with all its limitations. However, the results show that this approach can be viable and, with some improvements, can really be adopted in a Smart-City context to improve the safety of vulnerable road users.