Vehicle-sensor-based pavement surface condition monitoring based on an optical fibre computing framework

Abstract

This study is the primary research towards physical optical reservoir computing for road surface monitoring. Road surface monitoring traditionally utilizes specialized equipment on a road assessment vehicle to monitor road conditions, such as the profile, the roughness, the cracks, and the friction of the road pavement. These specific road assessment vehicles cannot provide up-to-date road surface conditions between two survey tasks, which limits the decision-making by using road data. Acceleration data collected by the sensors on the vehicles to achieve low-cost and high-efficiency road surface data acquisition based on the mobile vehicle platforms showing potential for road surface monitoring as an alternative approach. In this study, a physical optical fibre computing framework was utilized for road pavement surface fluctuation detection with acceleration data of the vehicles. Specifically, the acceleration data was obtained by an accelerometer on the vehicle for a road distance of 10 km and the corresponding road pavement surface fluctuation data was calculated from the road profile collected by a road assessment vehicle with a laser. The data was extracted from a road dataset. A time-delay optical fibre computing framework uses the normalized acceleration data in the z direction as the input signals and uses the dynamics of the physical optical fibre computing structure to calculate the temporary signals to be trained with linear regression to predict the road pavement surface fluctuations. The results showed that the physical optical computing structure achieved road surface fluctuation prediction while the parameters of the physical optical computing structure have a great influence on the prediction of the road surface data. This study provides insights into the development of physical computing for road surface monitoring in the future.