Using electric vehicle onboard data for pavement quality assessment and management

Abstract: Modern vehicles are increasingly equipped with onboard diagnostic systems that continuously collect data on driving conditions and vehicle performance. While primarily designed for vehicle monitoring, this data holds significant potential for infrastructure evaluation, particularly in assessing pavement quality. Traditional methods of pavement condition assessment rely on specialized, expensive instrumented vehicles that provide high-accuracy measurements but are not scalable for widespread, continuous monitoring.
This study investigates the feasibility of using onboard data from electric vehicles (EVs) to assess pavement quality in Minnesota. Drawing inspiration from Denmark's Live Road Assessment (LiRA) project and its open-source methodologies, we apply similar techniques to Minnesota's road infrastructure. By integrating onboard diagnostic data (OBD-II) from EVs with pavement quality data from the Minnesota Department of Transportation (MnDOT), we train machine learning algorithms to predict pavement conditions, focusing on measurements such as the International Roughness Index (IRI) and friction. The data utilized includes low-resolution vehicle dynamics attributes collected as EVs traverse the road network.
The key advantage of this approach is the widespread availability and low cost of data collection, as EVs routinely travel the roadways during normal operations. We evaluate the accuracy of pavement condition predictions derived from EV onboard data and assess the potential for broader implementation across Minnesota. Our findings aim to establish a framework for future studies and practical applications in pavement quality management, contributing to more efficient and cost-effective infrastructure maintenance strategies.