U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This summary report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-14-059    Date:  August 2014
Publication Number: FHWA-HRT-14-059
Date: August 2014


The Exploratory Advanced Research Program

Use of Vehicle Noise for Roadways, Bridge, and Infrastructure Health Monitoring Workshop Summary Report August 20-21, 2013

Day 2 Presentations

Presentation 1: Structural Health Monitoring Using New Technologies

Ming L. Wang,
Northeastern University

Ming L. Wang’s presentation focused on the project, “Versatile Onboard Traffic-Embedded Roaming Sensors.” The goal of the project is to create a system that can be outfitted on cars that would lead to less congestion during infrastructure monitoring and maintenance.

Wang began by explaining that at present, infrastructure asset owners generally visually inspect a small percentage of the road every 2–5 years. Most monitoring is performed on the surface of the material, although monitoring the subsurface condition would be more effective in preventative maintenance.

The VOTERS project provides a framework and sensor systems that would be carried by cars to shift from periodic, localized inspections to continuous network-wide health monitoring of roadways and bridge decks at traffic speed. VOO, such as buses and post office vehicles, could also be used to collect surface and subsurface roadway and bridge deck condition information at traffic speed. The collected data would then be transferred to a control and visualization center for further analysis, visualization, and decisionmaking.

Wang explained that the VOTERS system uses millimeter-wave radar (MWR; 24 GHz) to detect surface conditions, such as water, ice, potholes, surface roughness, and rutting. Subsurface radar (2 GHz) is used to detect subsurface conditions, such as rebar location, delamination, rebar corrosion, pavement thickness, moisture, and voids. In addition, acoustic sensors are used to detect debonding, voids, stripping concrete, potholes, MTD, and international roughness index (IRI). Optical sensors are used to view the surface profile, rutting, shoving, cracking, and potholes.

The acoustic system includes tire impact as the source and two different types of sensors. One sensor inside the tire measures the variation in tire pressure at high sampling rates. The other sensor is a microphone behind the tire, as shown in figure 8; however, Wang noted this system is complex, so the researchers also designed a simplified system to apply to a passenger or service vehicle.

Wang explained that, by using the physics-based system as proposed in the VOTERS acoustic system with the rear-mounted microphone, asset owners can calculate the condition assessment with Weibull probability density function, variance of PCA signal, IRI, and MTD. Using the DTPS, the system can interpolate the longitudinal profile, IRI, and subsurface condition. In addition, the video inspection allows for a visual inspection of crack density, crack type, crack severity, bleeding and raveling, and condition assessment.

Photo of a microphone mounted on the underside of a vehicle and pointing at a tire.
Figure 8. Microphone placement.

Wang noted that using the tire as a sensor alone is not sufficient for pothole detection, because most people try to avoid potholes when driving; however, the MWR addresses this concern. As shown in figure 9, it is an array that can detect rutting by evaluating the transverse profile, identify the surface material, and allow for condition assessment with radar longitudinal profile.

Photo of the rear underside of a car equipped with a five-sensor millimeter-wave radar (MWR) array.
Figure 9. 24-GHz millimeter-wave radar array.

The VOTERS system can output a pavement condition index (PCI) equivalent rating by fusing collected data from the different sensors in near-real time. Moreover, the video can be used as a secondary opinion to confirm what is seen in the data. Wang highlighted that the VOTERS PCI has been compared and validated against the traditional PCI in the field. Although the traditional PCI reports one value between intersections, VOTERS can report one value for smaller stretches of road at the order of 3–20 m (10–65 ft), depending on the vehicle speed.

The DTPS measures the longitudinal road profile, and materials can be identified based on the response. In addition, the IRI can be derived by the DTPS road profile and compared with the American Society for Testing and Materials’ standard for IRI. Wang noted that to evaluate subsurface, researchers tested 25 different pavement designs at the National Center for Asphalt test track and detected pavement debonding as a change in frequency, as shown in figure 10.

Dot plot showing levels of pavement debonding in a normal lane and a debonding lane. The Y axis measures frequency in kilohertz; the X axis measures time in seconds.
Figure 10. Pavement debonding detection.

Wang proceeded to explain to workshop participants how, by using the millimeter array, the transverse and longitudinal directions are measured and types of issues and materials can be identified, as shown in figure 11.

Multicolored representation of metal detection by a millimeter-wave radar (MWR) array.
Figure 11. Metal characterization example.

Wang also noted that the sensors used in the VOTERS system confirm and complement each other to identify types of issues and materials, as shown in figure 12.

Graph showing various sound waves measured during normal vehicle acceleration and by a dynamic tire pressure sensor (DTPS), a laser height sensor (LHS), and a millimeter-wave radar (MWR).
Figure 12. Multiple sensor confirmation.
(Accl = acceleration, DTPS = dynamic tire pressure sensor,
LHS = laser height sensor, MWR = millimeter-wave radar.)

Detailed Inspection

During the presentation, Wang explained that it is necessary to conduct a detailed inspection for areas with a lot of degradation. In events like this, a service vehicle with more robust equipment can inspect these locations, for example, a mobile acoustic subsurface sensing (MASS) vehicle could interrogate every meter of the roadway at walking speed. The impact source is also important, for example, if using a high, wide-band frequency, the material can be evaluated around 1-m (3-ft) deep; however, if the impact is short and a high frequency is used, the material can only be evaluated very close to the surface.

Wang explained that the system described in this presentation uses a frequency below 100,000 Hz and a 113 kg (250 lb) impact and  that the depths and modal elasticity of the materials can be extracted from the collected data. The MASS vehicle system is a noncontact sensing system that uses microphones, mechanical and software noise filtering, and fast-processing algorithms that processes data in roughly 2–3 seconds during collection.

Gen-3 Ground Penetrating Radar System

According to Wang, the GPR system can inspect at 80–96 km/h (50–60 mi/h) and consists of an array of 8 to 10 channels that are small enough to fit under a car chassis or can be hand-held. Wang explained that the development of the antennae was a large focus of this effort, and after many iterations, the research team selected an antennae called Pacman Bowtie. By using two Pacman antennae together as a transmitter–receiver pair, the researchers can evaluate depth, height, and property of the subsurface conditions. In addition, the GPR can detect areas of corroded rebar shown by comparing the reflection amplitudes to half-cell potential measurements. Wang noted that statistical analysis of GPR rebar amplitudes allows for immediate deck condition results.

Pavement Monitoring System

The VOTERS project also focuses on system integration to allow for full automation. Wang explained how, after fusing the data and integrating the data into the system, it could be visualized with the GIS-based pavement monitoring (PAVEMON) system. The system also uses colors to represent road condition and tracks the condition every 3 m (10 ft). As shown in figure 13, the data is overlaid on a map provided by Google Maps™. Wang noted that the PAVEMON visualizations can be compared and validated against historic PCI ratings of surveyed streets.

Satellite view map overlaid with pavement monitoring (PAVEMON) system data. The PAVEMON data is represented on the map by a colored line (mostly yellow, with some red and green areas) overlaid on the area of the map encircling the research area; the different colors represent different road conditions.
Figure 13. Pavement monitoring system visualization.

Lifecycle Analysis

According to Wang, asset owners typically conduct a lifecycle analysis by developing a pavement performance curve during the design phase to approximate lifetime costs and a repair schedule. The VOTERS system can monitor conditions more frequently to determine position on the performance curve during the lifetime, allowing the repair schedule to be updated and optimized. Wang noted that pavement deterioration modeling was used to estimate repairs and lifecycle costs. By holding the PCI at 77, the research team found that using the VOTERS system and sealing cracks resulted in substantial cost savings. Wang summarized the overall VOTERS vision as being able to inspect and repair in the right time, right place, and right way.

VOTERS Deployment Options

Wang outlined several deployment options during the presentation, as follows:

Wang also defined the following sensor levels during the presentation:

  1. DTPS, microphone, accelerometer, MWR.
  2. Level one, plus GPR array and video.


At the end of the presentation, Wang responded to the following questions from workshop participants:

Wang confirmed a microphone can be used behind each of the tires. Moreover, there is also the potential and existing capability to outfit a regular passenger car with a DTPS system or with a concealer microphone array underneath the car. Wang explained that the MWR is designed to be a very inexpensive system that identifies other properties, such as potholes, routing, and surface materials. Therefore, one cannot replace the other—they are complementary.

Wang confirmed that this sensor is specifically designed to measure only the variation of tire pressure (dynamic) and can be modified for use on a regular car.

Presentation 2: Decision Support System for Remote-Sensing and Geographic Information System Technologies

Colin Brooks,
Michigan Tech Research Institute

Colin Brooks’ presentation focused on the integration and display for DSS. Workshop participants were given an overview of the project, “Bridge Condition Assessment with Remote Sensors.” Brooks explained that remote sensing is the collection of data about an object, area, or phenomenon from a distance with a device that is not in contact with the object. This can be thought of as another form of NDE. Technologies used in this project include three-dimensional (3D) optics, thermal IR, digital image correlation, and ultra-wide band radar.

Three-Dimensional Optical Bridge Evaluation System

Brooks explained that the idea was to take high-resolution and overlapping photographs to create a high-resolution 3D model of the bridge deck to evaluate spalling, cracking, and other issues. In accordance, the research team mounted a camera to a frame 3 m (10 ft) above the bridge deck. As shown in figure 14, this camera was used to collect images of the bridge deck as the vehicle it was attached to maintained a speed of 3 km/h (2 mi/h) to collect images that overlapped by 60 percent. The 3D data was accurate to 2 mm (0.08 inches) in the X, Y, and Z directions.

Photo of a camera mounted on the flatbed of a pickup truck driving on a bridge deck.
Figure 14. Three-dimensional optical bridge evaluation system.

Brooks informed workshop participants about a new version of this project called “NDE Bridge Decks at Near-Highway Speed.” The investigators are currently developing a system that operates at a minimum of 72 km/h (45 mi/h) and will run simultaneously with the thermal IR camera from Talon Research’s BridgeGuard system. The system will process images using the program, AgiSoft Photoscan, to create a digital 3D model of the bridge deck and integrate images to produce a 1-mm (0.039-inches) bridge deck photo (shown in figure 15).

Three images of a bridge deck produced using the AgiSoft Photoscan program. The images depict the same section of bridge deck as a textured surface, a 3D model, and a digital elevation model (DEM).
Figure 15. Bridge deck three-dimensional model.

Figure 16 shows a selection of images produced from the automatic spall analysis, compared with the actual spall location on the pavement. Brooks explained that the algorithm used in the spall detection analysis is programmed in Python and uses ArcPy to utilize ArcGIS geospatial tools. The focal statistics tool is used to find potential spalls, shapefiles (a geospatial vector data format) are generated for detected spalls, and a table is generated with spall sizes and volumes. Brooks noted that the computation time for this analysis can be reduced if a spall-size threshold is included.

Three images of spall on a bridge deck: a photo of spall with overlaid green lines; a thermal image; and a digital elevation model (DEM) with the spall outlined in red.
Figure 16. Spall analysis.

Brooks explained that the research team also used thermal IR technology to find delaminations based on local temperature deviations and additional measurement methods, including chain drag. The thermal imaging and chain-drag method both found defects, but they did not always find the same defects. In figure 17, the red areas are the defects that were detected by the thermal IR method, and the green areas are the defects that were found by using the chain-drag method.

Satellite view of a section of bridge deck tested for delaminations using both thermal infrared imaging and chain-drag methods. The figure shows that different defects were detected by the two methods.
Figure 17. Bridge deck delamination by thermal infrared and chain drag.

The research team also used other remote-sensing technologies in this study, including remote-camera photo inventory with high-resolution location-tagged photos, LIDAR, satellite and aerial photos, interferometric synthetic aperture radar, and speckle image processing.

Integrating Remote-Sensing

Brooks informed workshop participants that the goals for integration include a comparison of remote-sensing observations to established measures, a comparison of remote-sensing observations to one another, and a way to derive established measures from remote-sensing observations.

Existing Decision Support Systems

Brooks proceeded to outline a selection of existing decision support systems, as follows:

Pontis AASHTO BRIDGEWare System
Over 45 agencies throughout the United States and abroad have adopted this system. It represents bridges as sets of structural elements and supports optimization and asset management workflows. One gap in this system is that agencies cannot integrate remote-sensing NDE data directly in a geographic environment at this time.

Michigan Bridge Reporting System
This system is a prototype for MDOT that integrates remote-sensing NDE data into MDOT’s existing data.

Online Long-Term Bridge Performance National Bridge Inventory
The LTBP program is researching sensor data integration.

Asset Management

During the presentation, Brooks explained that components of the asset management decision process include previous bridge inspection reports, visual inspection, remote sensors, the National Bridge Inventory (NBI) rating and Pontis condition, decisions on maintenance, and bridge inspections. A bridge deck surface rating could provide a comparison to the NBI rating to assist with decisions on maintenance. Moreover, IRI can also be calculated from the data.


Brooks explained that the decision support software is open source, can be viewed in any browser, and can access all inventory data. In addition, through the interface, users can access typical features, as shown in figure 18. These include existing inspection data, sorted by region or county, and additional sorting features, such as driving directions and NBI deck ratings. Remote-sensing data is integrated, and users can click on a bridge and view directions, remote-sensing data, IRI ratings for each lane, the percent spalled, the percent delaminated, crack density, and images. Brooks noted that the main challenge with this system is the way bridge data is not stored with a focus on integrating the data rapidly into an online mapping tool.

Screenshot of open source bridge condition decision support software.
Figure 18. The decision support software interface.


At the end of the presentation, Brooks responded to the following questions from workshop participants:

Brooks explained that the vehicle speed relative to frame rate must be considered to minimize motion blur.

Brooks confirmed that the key is to have a base layer that all other images will use as a reference.

Brooks suggested that this could perhaps work, but the investigators had not considered it.