Using electric vehicle onboard data for pavement quality assessment and management

Project summary:

Today's vehicles are constantly collecting onboard diagnostics data to assess the driving conditions and internal component states. While these sensors and data are usually used to monitor and optimize vehicle performance, they may soon be useful to also monitor the infrastructure by implementing novel machine learning techniques to assess pavement quality based on vehicle driving characteristics as it passes over different pavement sections. The benefit is that these data are already being collected while vehicle fleets traverse the roadway network as part of their typical daily operations.

Inspired by the recent Live Road Assessment project in Denmark funded by the Danish Road Ministry, researchers propose to implement their open-source methods on a MnDOT vehicle (or small fleet of vehicles) to perform a feasibility analysis on conducting pavement monitoring using onboard vehicle data from fleets of vehicles in Minnesota. This will include identifying possible vehicles for use, decoding onboard data, recalibrating the machine learning models for the selected vehicle, and comparing obtained pavement quality estimates to established values from regular MnDOT pavement assessment measurements.

Access to the one-of-a-kind MnROAD research facility and its wealth of information is of critical importance in successfully implementing this advanced technology to obtain a more accurate and timely evaluation of the condition of the pavement network.