Condition Assessment of Ancillary Roads from Aerial Imagery

Satellite and pavement cracking

Background

Pavement condition is typically characterized by roughness, surface friction, and distresses such as cracking, rutting, patching, raveling, etc. State transportation agencies use manual or automated techniques, or a combination of both, to record and keep track of these surface distresses over the years. Manual data collection involves trained personnel rating the surface distresses, resulting on a labor intensive and time-consuming approach that is also highly subjective In the automated process, a Pavement Management System (PMS) is typically coupled with Global Positioning System (GPS) where a van equipped with high-resolution electronic sensors runs along the roads at a driving speed and records georeferenced condition data. These data go through extensive post processing as it involves complicated analytical models and algorithms and requires substantial technical expertise. Despite of being an automated process, this remains an expensive and time-consuming form of data collection due to the amount of driving and post processing required at a network level.

Due to these high costs, DOT’s often limit the pavement condition assessment to major highways and limited information is available on the condition of ancillary roads such as ramps. There is therefore a need to identify alternative and cost-effective solutions to assess the condition of ancillary roads. In this regard, recent advances on remote sensing technologies and the availability of high-resolution aerial imagery provides a cost-effective and innovative solution to assess pavement condition.

Research Objective

The goal of this project is to develop a predictive tool to assess the condition of ancillary roads from high-resolution aerial imagery. The concrete prediction task of the predictive model will be as follows: given an aerial image of an ancillary road, the model will predict the pavement condition.

Research Methods

As part of this project, the research team collected Synthetic Aperture Radar (SAR) images and pavement condition data from the MnDOT Pavement Management Systemto develop predictive models based on machine-learning to estimate the condition of pavements from SAR images.

Contributions

  • SAR-C: A software application to estimate pavement condition from SAR images.
  • Bashar, M.; Torres-Machi, C. (2023) Quantifying the Value of Satellite-Based Pavement Monitoring in Partially Observable Stochastic Environments. Journal of Computing in Civil Engineering, 37(3): 04023004. DOI: 10.1061/JCCEE5/CPENG-5108
  • Bashar, M.; Torres-Machi, C. (2022) Deep Learning for Estimating Pavement Roughness using Synthetic Aperture Radar Data. Automation in Construction, 142, 104504, DOI: 10.1016/j.autcon.2022.104504
  • Bashar, M.; Torres-Machi, C. (2022) Exploring the Capabilities of Optical Satellite Imagery to Evaluate Pavement Condition. Construction Research Congress (CRC 2022), Arlington, VA, USA, March 9-12, pp. 108-115. Organized by: American Society of Civil Engineers (ASCE).
  • Bashar, M.; Torres-Machi, C. (2021) Performance of Machine Learning Algorithms in Predicting Pavement International Roughness Index. Transportation Research Record: Journal of the Transportation Research Board, 2675(5) 226–237, DOI: 10.1177/0361198120986171.

Funding

Minnesota Department of Transportation (MnDOT): 2020-2022.

 

Research Team