3D Mapping of Pavement Distresses Using An Unmanned Aerial Vehicle (UAV) System
3D Mapping of Pavement Distresses Using An Unmanned Aerial Vehicle (UAV) System
3D Mapping of Pavement Distresses Using An Unmanned Aerial Vehicle (UAV) System
Abstract. The aim of road surface monitoring is to detect the distress on paved
or unpaved road surfaces. Depending on the types of surface rupture, required
parameters are measured on-site to determine the severity level of the road
damage. Local infrastructure engineers and supervisors must therefore optimize
their resources when monitoring road conditions and scheduling maintenance
activities. Automation of road surface monitoring process may result in great
monetary savings and can lead to more frequent inspection cycles, for this
reasons departments of road maintenance, repair and transportations have
become more interested in using automatic systems for pavement assessment.
The scope of the presented work is the performance evaluation of a UAV system
that was built to rapidly and autonomously acquire mobile three-dimensional
(3D) mapping data to identify pavement distresses.
1 Introduction
Monitoring and maintenance of paved road such is an expensive but essential task in
maintaining a safe operation. The road surface condition assessment is an essential
piece for recent pavement management systems methods as rehabilitation strategies are
planned based upon its outcomes. Traditional road surface monitoring methods use
visual inspection for proper evaluation of existing pavements that must be continuously
and effectively monitored using practical means [1]. Conventional visual inspection of
road distress assessment requires manual visual inspection by personnel in-charge. The
truck-based pavement monitoring systems represent the common methods have been
in-use in assessing the remaining life of in-service pavements and the detection of road
distress. This system produces accurate results, but their use can be expensive and data
processing can be time consuming, which make them difficult to use considering the
demand for quick pavement evaluation. Because of the small funding base of local
government, the human and financial resources available for maintaining roads are
often inadequate. In the last decades the rate of making and utilization of computer
vision methods for pavement engineering applications have been exponentially
increased [2] with the help of developments high-level processing techniques to extract
information from the images and sensing technologies to capture images efficiently and
accurately under various lighting conditions.
Satellite images can be able to monitoring road distress. However, due to the cost
and the limited spatial resolution quality of the image, it is not preferred. So Images
from different platforms such as Unmanned Aerial Vehicles (UAVs) can be cost
effective compared to satellite’s imagery or traditional aerial photography.
In this paper a road pavement condition was monitored in Reggio Calabria sing an
Unmanned Aerial Vehicle (UAV) to inspection of road distresses, such as road surface
cracks and potholes, like through image processing algorithms for the detection of
cracks in building facades.
The flexible pavement surface due to repeated vehicle loading, environmental inter-
action and insufficient structural strength can realize pavement failures as rutting,
fatigue cracking, and migration of fine particles (pumping) [3, 4]. Rehabilitation and
road maintenance work is not only time-consuming, but it also creates traffic distur-
bances and is also very costly. Pavement condition inspection can be realized in various
ways. The First and the oldest way, yet in-use, consist to walk on the roads and
carrying out visual analyses and measuring the required parameters through manual
inspection (Fig. 1). In this procedure, a pavement expert travels along the road and
collects both visual and quantitative data from the pavement surface by investigating
the cracks with bare eye. Visually inspecting the infrastructure and evaluating them by
subjective human experts is the simplest method.
described in the first method. Using this method, continuous and high-resolution data
can be collected at high speeds, so they are widely used in pavement inspection systems
but the initial cost of developing such truck-based systems is relatively expensive.
Pavement inspection systems can be categorized into two parts as data collection,
and data processing. In data collection part, the information from the pavement surface
is gathered as 2D images or 3D surface data by using moving vehicles or aerial
platforms. After performing field experiments and collecting data, in general, the
gathered data are processed using sophisticated image processing algorithms and
machine learning techniques. The integration of more automated and semi-automated
remote sensing methods is a relative new concept and may provide valuable results
when generated and interpreted in a systematic manner. The application of these
methods is likely to be less expensive, more consistent, and conducive to faster
inspections that cover wider areas [5]. Non-destructive evaluation techniques, such as
Digital Image Processing (DIP), Ground Penetration Radar (GPR), fiber optic sensors,
laser systems (LS) or Hybrid systems (HS) are emerging procedures for health mon-
itoring. Furthermore, in the last decade Unmanned Aerial Vehicles (UAVs) have
become extremely popular especially in civilian applications because of its low cost
and practicality. In fact, they are able to offer high resolution and almost real-time
images often cheaper than other space born platforms. A UAV, commonly known as a
drone (Fig. 3), is an air vehicle which can fly without a human pilot and with the
advancements of its technology, newer commercial UAVs come to the market more
frequently. Compared to traditional evaluation methods just described, remote sensing
techniques offer non-destructive methods for road condition assessment and
image-processing techniques to assess road conditions are considered a good nonde-
structive method to quantify pavement distress like cracking. Autonomous inspections
of paved road infrastructure remain a relatively unexplored area of application, how-
ever, their quick response times, maneuverability, and resolution, make them important
alternatives for pavement assessment [6].
In this paper, the authors propose a UAV based pavement crack identification
system for monitoring flexible pavements’ existing conditions based on the combina-
tion of image processing techniques and machine learning. In the following section, the
method developed is presented.
3D Mapping of Pavement Distresses 167
A UAV, so called a drone, is an air vehicle which can fly without a human pilot. UAVs
started to be used by military reconnaissance purposes in the 1950s.
With the advancements of its technology, newer commercial UAVs come to the
market more frequently. Thanks to potential of high flexibility and efficiency of UAV
system, pavement managers can assess large, often inaccessible areas, in little time,
providing valuable up-to-date information on pavement deterioration.
The first road condition assessment using UAV was introduced by Zhang [7]. The
flights were done by a radio controlled low cost airframe helicopter equipped with
position and velocity detection sensors, then was using a number of image processing
algorithms are under development to calculate some features such as length and size of
corrugation, geometry of cross section, rutting, potholes etc. In addition, the acquired
imagery and developed algorithms may also be useful for the extraction and mea-
surement of other road properties, such as road way width, curvatures, etc. which are
also important components in road way management. Zhang and Elaksher [8] pub-
lished a study where an improved form of UAV-based road condition assessment was
performed. The new system included a low-priced model helicopter armed with a
10.1-megapixel resolution digital camera, a GPS receiver and an Inertial Navigation
System (INS) and Ground Control Station (GCS). The ground resolution up to 5 mm
was obtained from a UAV flight at 45 m altitude. To obtain more accurate measure-
ment for the road surface distress, new set of algorithms were developed for generation
of 3D surface model and orthoimages. The difference between the 3D model and field
survey data that author has individuated indicated that maximum accuracy could be
achieved up to 0.5 cm ground accuracy using the UAVs. The results show that the
proposed system is practical for local transportation agencies to gather surface con-
ditions of unpaved roads.
Shenzhen, China) and Pix4d software (Pix4d SA, Lausanne, Switzerland) for automatic
mapping area and image acquisition through waypoint. Mavic pro camera specs are
reported in Table 1.
To obtain a high quality ordered dataset and produce a model with defined char-
acteristics, an image acquisition plan was created, defining also camera settings and geo
information acquisition type. The image acquisition plan was divided into three steps:
(1) definition of image acquisition plan type, (2) definition of Ground Sampling Dis-
tance (GSD), and (3) definition of image overlap. The Image acquisition type was set to
automatic waypoint flight mission to create 3d object, so the UAV perform an auto-
matic flight as shown in Fig. 5. The GSD calculation [10] define the height of flight
according to the required definition (cm/pixel) of the model. To obtain centimetre
precision (GSD < 1) at comfortable flight height, considering Mavic Pro camera specs
(Sw = real sensor width = 6,17 mm, FR = real focal length = 5 mm) the height flight
was set to 30 meters for vertical flight. The overlap between two consecutive images,
that determined the image acquisition rate at fixed UAV speed, was set to 80%. The
image rate acquisition calculation for front overlap, is made automatically by Pix4d
software.
Moreover, the software calculates the image acquisition plan and mission settings
automatically, defined the following parameters: flight eight (and consequentially
GSD), overlap (%), and area to be mapped as shown in Fig. 6.
The overall survey to map 1 km2 was performed in almost 2 min. Mavic pro
camera setting were set to automatic and geo tag information were captured using the
UAV built in GPS and store in the images EXIF data.
Fig. 5. Waypoint flight mission and overlap scheme for automatic mapping area.
three-dimensional point cloud of the object. Then, starting from point cloud, the
software reconstructs a 3d mesh. The output DEM (Digital Elevation Model), recon-
structed using WGS84 geotag information, is represented in the following figure
(Fig. 7):
4 3
1 2
Starting from DEM the crack on the pavement were highlighted applying a filter on
3d point cloud under the pavement level using 3dReshaper software (Fig. 9 and
Table 2). Four different potholes were individuated and measured directly on DEM as
shown in Fig. 8:
5 Conclusion
The pavement road degradation analysis is fundamental for the planning of mainte-
nance interventions. Innovative methods, such as the use of drones, allow savings both
in terms of time and maintenance costs. The analysis carried out verified the drone use
for the identification of the potholes on the pavement surface and their geometric
characteristics, in particular of the depth. Through the comparison between the mea-
surements made in site and those obtained from the DEM model, it is clear the validity
of drone use. In order to reduce the images process time, a real time procedure will be
developed to analyse the images obtained by drone and identify the distresses with its
geometric characteristics.
References
1. Liq Yee Tiong, P., Mustaffar, M., Rosli Haini, M.: Road surface assessment of pothole
severity by close range digital photogrammetry method. World Appl. Sci. J. 19(6), 867–873
(2012)
2. Cet, K., Georgieva, K., Kasireddy, V., Akinci, B., Fieguthd, P.: A review on computer vision
based defect detection and condition assessment of concrete and asphalt civil infrastructure.
Adv. Eng. Inform. 29(2), 196–210 (2015)
3. Buonsanti, M., Cirianni, F., Leonardi, G., Scopelliti, F.: Dynamic behavior of granular
mixture solids. Key Eng. Mater. 488–489, 541–544 (2012)
4. Leonardi, G.: Finite element analysis for airfield asphalt pavements rutting prediction. Bull.
Pol. Acad. Sci. Tech. Sci. 63(2), 397–403 (2015)
5. Schnebele, E., Tanyu, B.F., Cervone, G., Waters, N.: Review of remote sensing
methodologies for pavement management and assessment. Eur. Transp. Res. (2015)
6. Leachtenauer, J.C., Driggers, R.G.: Surveillance and Reconnaissance Imaging Systems:
Modeling and Performance Prediction. Artech House, Boston (2011)
7. Zhang, C.: An UAV-based photogrammetric mapping system for road condition assessment,
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information
Sciences, ISPRS Congress, XXXVII. Part B (2008)
8. Zhang, C., Elaksher, A.: An unmanned aerial vehicle-based imaging system for 3D
measurement of unpaved road surface distresses. Comput. Aided Civ. Infrastruct. Eng. 27
(2), 118–129 (2012)
9. Lowe, D.G.: University of British Columbia, Canada - Object Recognition from Local
Scale-Invariant Features (1999)
10. Wu, C.: University of Washington - Towards Linear-time Incremental Structure from
Motion (2012)