Open Access Open Access  Restricted Access Subscription or Fee Access

In Field Application of Rapid Roadway Inspection System Using Vehicle-Mounted Multi-Modal Sensing



U.S. roads are in critical need of repair. This is evident through personal driving experiences and professional organizations like the American Society of Civil Engineers, who performed a nationwide survey in 2013 that yielded a report card rating of D, which is nearly failing. Going hand in hand with this critical need for repairs, is a critical need for pavement condition surveys. These city-wide surveys provide a condition label for every street based on type, severity, and density of distresses. Since cities lack the resources to fix all roads at once, they must rely on these surveys to prioritize and develop cost effective maintenance and repair plans. Conventional pavement condition surveys require a rigorous in-field inspection by experts, which is time consuming, unsafe for inspectors, and causes traffic interruptions. Alternative vehicle-mounted sensing approaches exist, but are not economical due to expensive sensing technology and costs for experts to drive the vehicle and manually process data. The VOTERS project (Versatile Onboard Traffic Embedded Roaming Sensors) aims to develop a mobile pavement inspection system that is more affordable and can be repeated more often. This goal will be realized through the development of a compact, hidden, and fully automatic system that can be installed on service vehicles such as Post Office, UPS, FedEx, or public transportation. In order to handle the complex nature of pavement condition and the challenges of mobile sensing, the system will be comprised of an innovative and affordable multi-modal sensing array that uses data fusion concepts. Currently, a test vehicle has been developed that is outfitted with an acoustic sensor, dynamic tire pressure sensor, camera, mm-wave radar, and laser height sensor. This paper will introduce this system and discuss its recent application and results in the City of Brockton Massachusetts. A data fusion machine learning approach for mobile pavement condition assessment will be developed and shown to have accurate and repeatable results when implemented in the field.

Full Text: