The Giant Machine That Constructs Whole

Bridges
Development of Robotic Bridge

 Point on Robotic Bridge

1. Design of structure (Machine)
2. Purpose of Robotic Bridge
3. Advantages and Disadvantages of Robotic Bridge
4. Type of Robotic Bridge
5. Repair and maintenance of Robotic Bridge
6. Structure of Bridge

The SLJ900/32 is the kind of giant machine China builds when it needs to construct all
new bridges. Built by the Beijing Wowjoint Machinery Company, the machine is 300
feet long, 24 feet wide, 30 feet tall, and weights 580 tons. In the video below, you can get
an idea of its sheer scale when you watch workers scale down it. This thing is just
gigantic.

But it has a big job ahead. Essentially, it's laying the "track" for new bridges, piece by
piece, moving slowly across support girders. in the end, you have a hunk of machinery
that builds megaprojects as China's economy booms and its infrastructure needs
accelerate.
Anatomy

The SLJ900/32, manufactured by the Beijing Wow joint Machinery Company and
designed by the Shijiazhuang Railway Design Institute, weighs in at a mighty 580
tonnes, is 91 meters long and 7 meters wide. The colossal machine offers a totally new
way to complete large bridge projects without the need for expensive and complex
scaffolding and framework structures.

If seen away from the building environment the casual onlooker would be forgiven for
thinking the machine is simply something being used to transport structures via
roadways, such is its seeming simplicity and complete difference from more traditional
bridge construction devices.

How it works

After picking up the beam needed from its point of origin, the SLJ900/32 will drive to the
pillars of the bridge and lower a pneumatic support structure which essentially anchors
the machine to the first pillar, allowing it to extend itself out to the second pillar. From
here it moves on to a third, depositing the beam as it does so. To watch it in action is
perhaps the easiest way to fully understand it’s efficient, fluid and fuss-free operation.

The machine transports itself, via its set of 64 wheels, which are split into four sections
consisting of 16 wheels each. As each section is capable of rotating 90 degrees, the
SLJ900/32 can move sideways in order to facilitate easy pick-ups of beams.
Whereas traditional crane-based methods of bridge construction have necessitated that
beams and other segments come in sections usually just a few meters long, which are
then fixed together, the SLJ900/32 is able to utilize beams that are specially prepared to
be full length. Expansion joints are used to connect these prefabricated parts

Lifespan and Projects

At present the machine is said to be capable of up to 730 spans with about 40 per cent
of the products being able to push this total up to 1000 before reaching the end of its
lifespan which averages at four years.

The SLJ900/32 is currently in active use in China for the construction of large and high
rail viaducts, most notably between Chongquing and Wanzhou in the province of
Sichuan. Due to the heavy weight of the machine, these bridges are made to be far
bigger than necessary and hence capable of taking far heavier traffic and weight loads
than the rail networks for which they are intended.

The fact that this concession seems to be being made across China speaks volumes
about the way in which the SLJ900/32 is perceived within the industry, as well to the
time and cost savings that are inherent in the move away from structural scaffolding that
the machine facilitates.

 Weighing 580 tonnes, measuring 91.8 meters (301 ft.) long, 7.4 meters wide and
nine meters high, this mean machine is building bridges in a way we’ve never
seen before.
  Viewers watching the video above can see that the SLJ900/32 operates without
conventional crane technology.
 Instead, the machine travels to the edge of the bridge and reaches out to towers
yet to be connected to the bridge with a temporary track. Once the track is stable,
the machine pulls itself out with the new bridge segment in tow.
 Once fully extended with the track and at this point reaching another bridge
tower, the SLJ900/32 lowers the new bridge segment in place for the
construction crew to begin their work.
 Once this new segment is secure, the process can be repeated.
 For another demonstration of how the machine works, check out this other video.

 The SLJ900/32 is of Chinese design by Beijing Wowjoint Machinery Co., a

supplier of customized heavy duty lifting and carrying machinery.

 From the distance in the second video, it is hard to tell what safety equipment
was used by the Chinese workers. However, if something like the SLJ900/32
were to gain popularity in the US, you can bet harness and height training will
need to be up to date for workers.

Launching gantry

Launching gantry used for Honolulu Rail Transit guide way construction (2015)

A launching gantry (described also by other terms including beam launcher, girder launcher,
bridge building crane, and bridge-building machine, locally nicknamed the Iron Monster is a
special-purpose mobile gantry crane used in bridge construction. It is used to install precast box
girders in highway and high-speed rail bridge construction projects. The SLJ908 is a machine
that can carry, lift, and place sections of track, connecting pillars by heavy stone blocks. After
pillars are in place and construction has been carried out up to a certain pillar, the machine
advances over the gap to the next pillar and drops another block of track into place, spanning the
gap. The machine then backs up to collect another block, moves forward, and repeats the
process.
An example of a large launching gantry is the SLJ900/32 designed in China by the Shijiazhuang
Railway Design Institute and manufactured by the Beijing Wow joint Machinery Company. This
launching gantry is 91 meters long, 7 meters wide, and weighs 580 tonnes. When driving, the
machine is supported by 64 wheels, in four sections of 16 wheels each (forming two trucks, one
at each end). When launching, the forward end of the machine is supported (on sliding rails) by a
strut lowered onto a bridge support column, while the truck for that end hangs off the gantry
backbone with no support from beneath. Once the gantry straddles the open span, the work piece
is lowered onto the bridge support, and the process reverses to retract the launching gantry. The
SJ900 moves at 8 km/h (5 mph) unloaded, and 5 km/h (3 mph) carrying a block of track.
Contents
 Girder construction
o Honeycomb girder
o Truss girder
o Box girder
 References
 External links

Girder construction

Launching gantries are distinguished by the construction of the main girder of the gantry itself.

Honeycomb girder

Honeycomb girder

The honeycomb girder launching gantry was created in 1999 by ZZHZ, and has proven suitable
to a lifting range of 5–300 tonnes.
The main girder of a honeycomb girder is fabricated from welded plates in isosceles triangle
section. Regular hexagonal honeycomb holes are cut in the inclined web plate for air currents to
pass through. This reduces wind impact. Welded plate by plate, the welding surface will not lose
stability easily due to small welding defects.
Truss girder

truss girder

The main girder of a truss girder is fabricated from welded steel poles. Point-to-point welding
(steel poles to pole welding) has a high demand on welding technology. With small self-weight
and large deflection, the truss girder launching gantry is suitable for highway bridge construction
projects.

Box girder

box girder

The main girder of a box girder is fabricated from welded plates. This type of girder is usually
used in high-speed railway bridge construction projects.

1. DEFINITION

Bridges are defined as structures, which provide a connection or passage over a gap without
blockin the opening or passageway beneath. They can be over streams, canals and rivers;
creeks and valleys or roads and railways passing beneath. Bridges are now being provided
across ocean bodies and for linking a number of islands as in Japan. Some typical such links! are
listed in Annexure 1.1. The bridge crossing .carrying a road or railway over another road or
railway is called a grade separator or fly over. The bridge structure can be for passage/carriage
of persons, cattle, vehicles, water or other materia! carried across in pipes or conveyors. When
they are used for carriage of water, they are called aqueducts. Even a jetty in the ports and
harbours can be classified as a bridge. No other creation of a civil engineer has such a , general
appeal and fascination to the people as a bridge.
1.2 HISTORY OF DEVELOPMENT OF BRIDGES

Bridge building is not a new science. With the growth of civilizations, the need for travel has
impelled ,mankind to find ways and means of bridging gaps over deep gorges and perennial
streams, for walking across. The simple form could have been by felling trees across gaps and
using them for walking across. Thus timber can be considered as the earliest material to be
used for bridging. This has been followed by bridges built with stone and then of brick, used by
themselves or in combination with timber. Such bridges have been possible only for short
spans. These materials appear to have been in vogue for many centuries till iron was developed
around the middle of the last millennium.The earliest reference available is of a bridge across
Nile built in about 2650 BC though according to Swedish Institute of Construction the oldest
bridge built was an 1100-m-long wooden bridge built in England in 3306 BC. The oldest
pedestrian bridge still standing is a stone slab bridge across River Meles in Smyrna, Asia Minor
(Turkey) said to be 2500 years old. Swiss were the pioneers of timber bridges, specially using
trestle form. In known history, the Chinese appear to be the earliest to build stone bridges.
Their earliest stone bridges still in existence to be seen are Zhaozhou bridge and Anji Bridge, an
open spandrel arch bridge built between 595 and 605 AD (vide Plate 1.1 and Plate 1.2). They
are also believed to have built the earliest C.I. chain bridge in approximately 960 AD. Romans
are believed to have built bridges and aqueducts for carriage of water

2 Bridge Engineering

before even the start of the first millennium. Queen Nitocrin builta bridge in stone in about 780
BC with a piers built with stone and wooden plank decking. Stones were bound together with
iron and lead. Roman are also credited to have used timber pile bents for foundation and piers,
the first such known being the one across Rhine built in 95 BC. Plate 1.3 shows a typical bridge
of the type 'Pont du Gard' built by Romans in the First century AD. Similarly, Gordon River
Bridge built in 13 B.C in France was a masonary aqueduct 49 m high, with three rows of
superposed arches. Roman dominance over other countnies declined in the twelfth century
B.C.

Etruscans are believed to have used vaults for bridge construction as early as 600 BC Europe Is

considered one of the birthplaces of bridge design and technology. Thus, they must be earliest
to develop bridge building as a technique. The earliest timber bridge built was, probably,
Trojan’s Bridge built in AD 104 (destroyed 6 years later). It was made up of 20 numbers 30 m
timber arches over stone piers.The Roman bridge building art spread to Middle East and as far
as India. Marco Polo is said to haveremarked, ‘Indian cultures adopted their own tools under
this influence for bridge building and further developed suspension bridges’. Indians have built
suspension bridges with use of ropes for Suspensionand bamboo and timber planks for decks in
the hilly regions from early days. They are also credited to have built cantilever type of bridges
laying stone slabs one over the other in a progressive manner to bridge gaps, but have kept no
records.

In medieval times, the church as part of their activities of building cathedrals influenced bridge
building also and there was a Jesuit brotherhood devoted to bridge building. All these bridges
and buildings had been built with stone, brick masonry and timber using emphirical methods
for design. A typical example is the first London Bridge built by Peter of Colechurch in | 176-
1209 A.D. This was a masonry bridge with 19 pointed masonry arches on piers, none of them
with same dimensions. This bridge lasted for about 600 years. Wittengen Bridge built in 1758 in
Germany was the longest timber bridge in Europe with a 119 m span.It was during the
Renaissance period that a start was made to do bridge construction on a scientific basis. The
truss system based on the principle of triangles, which cannot be deformed, was developed
Andrea Pallaido (1508-1580 AD), evolved several truss forms, including the king post type.
Verrazino(1615) had written about roads, machines, waterwheels, bridges including masonry
arches with use of prestressing rods, as well as suspension bridges and use of iron eye bars for
suspension bridges. First metal bridge to be built was Coalbrookdale bridge (designed by
Abraham Darby III). (vide Plate 1.4). It was built in cast iron in 1776. James Finlay patented
suspension bridge form and built some with steel chains. But they were found to be subject to
severe corrosion problems and needed frequent attention. French Engineer Vicat invented the
acrial spun cables for suspension. This type has become the major form for building longer and
longer span bridges of today. The earlier popular arch form adapted to cast iron and the truss
form adapted for different shape in W.I. and steel, revolutionized bridge building using iron and
steel as basic spanning materials instead of masonry and timber respectively. This dominated
the scene for many centuries till the arrival of prestressed concrete. Eads bridge at St. Louis was
the first bridge to be built with extensive use of steels as early as 1874. (vide Plate 1.5) Firth of
Forth Railway Bridge in Scotland followed suit, with use of tubular steels sections for main
girders and columns. Though, today, there are critical comments on the over design of the
bridge, disaster and doubts about its cost economics in the wake of the Tay Bridge disasters,
this appears justified. Forth bridge design had been appreciated for the bold attempt made to
span such lengths and shaping the structure so as to follow clearly the force lines and giving an
elegant look for a viewer from distance, as can be seen in Plate 1.6. Trend in 18th and
nineteenth centuries for longer span bridges especially in U.S.A tended towards cable
suspension bridges. One of the most elegant of such structures, the Golden Gate Bridge in the
San Francisco bay in USA, built in 1937 is shown in Plate 1.7. It was the longest of the kind at
that time. The first Portland cement concrete bridge to be built was the Grand Maitre Aqueduct
across River Vane in France built in 1867-74. France is also the birthplace of prestressed
concrete, which is the major form of bridge superstructures all over the world today either by
itself or in combination with steel.
Present Trend

Japanese also built iron bridges in the same period as others. They have the longest cable
suspension bridge, Akashi Kaikyo Bridge, with a record span of 1991 m as on date. (vide Plate
1.8). Germany was the first to introduce the concept of cantilever construction and incremental
launching of concrete decks, as well as the modern form of cable stayed bridges. Russians used
timber as main bridge building material up to the end of 15"" century, though some of the
former republics in USSR had built masonry bridges. One of the old bridges, built in 1234 A.D.
still in existence is the Sanainsky Bridge over River Debda-chai in Armenia. China has built some
notable bridges using tied arch form and cable stayed bridges. Two elegant examples arc, the
Dagu bridge at Tianjin (vide Plate 1.10) and a railway bridge at high altitude on their recently
opened rail link to Tibet. They have built the longest steel tied arch bridge (as of 2006) Lupu
bridge (vide Plate 1.9). Franklin D. Roosevelt once said ‘there can be little doubt that in many
ways the story of bridge building is the story of civilization. By it, we can readily measure a
progress in each particular country. Based on this saying, the Indian civilization being once of
the oldest, must have built bridges well before Christian era. Unfortunately, to the best of the
author's knowledge, there is very little readily reachable record on the subject in India, from
which one can discern when the earliest bridge was built in India, except for the mythology
mentioned in previous section. According to records of Chinese travellers on Indian history,
India appears to have had a good highway system in the days of Harshavardhana or even
earlier. Such highways must have had a number of bridges. Firoze Shah who ruled in Delhi in
mid-fourteenth century is said to have built canals and bridges along with schools and hospitals.
One can still see some old masonry arch bridges built by the Portuguese in 16" or 17" century in
Goa. One old bridge still in use is the stone slab bridge across River Cauvery at Srirangapatnam
built by Tipu Sultan who ruled in the eighteenth century. We have a number of old masonry
and stone arch bridges built in the middle of the nineteenth century on the Railways, which
bear testimony to the skill of the local people in bridge construction. The British who built the
railways had brought the stecl bridge girders and their designs from UK, but they depended on
the local skills and expertise to build the others. Sructural forms and designs for longer spans
also appear to have come from the British. The technical knowledge within the country has
since kept pace with the developments abroad and their application

4 Bridge Engineering

has however, been governed by the opportunities available within the country .Number of
cable stayed bridges has been built in India in the past two decades,the major one of being
vidyasagar Sethu across Hooghly at Kolkata and the Naini Bridge on River Jamuna at Allahabad.
the railways are building a number of major bridges including a large steel arch bridge in
Jammu and Kashmir, The Border Roads Organisation has erected a cable stayed bridge using
Bailey Bridge girders in early part of this millennium, which bridge is claimed to be only bridge
of the type at highest altitude in the world at the time of construction.

Abstract: This article presents a precise landing system that allows rotary-wing UAVs
to approach and land safely on moving platforms, without using GNSS at any stage of
the landing maneuver, and with a centimeter level accuracy and high level of
robustness. This system implements a novel concept where the relative position and
velocity between the aerial vehicle and the landing platform are calculated from the
angles of a cable that physically connects the UAV and the landing platform.
The use of a cable also incorporates a number of extra benefits, such as increasing the
precision in the control of the UAV altitude. It also facilitates centering the UAV right on
top of the expected landing position, and increases the stability of the UAV just after
contacting the landing platform. The system was implemented in an unmanned
helicopter and many tests were carried out under different conditions for measuring the
accuracy and the robustness of the proposed solution. Results show that the developed
system allowed landing with centimeter accuracy by using only local sensors and that
the helicopter could follow the landing platform in multiple trajectories at different
velocities.

1. Introduction and RelatedWork

In the next decade, it is expected that civil applications of Unmanned Aerial Vehicles
(UAVs) will increase exponentially up to a market of 11,000 million euros in 2035 only in
Europe [1]. Moreover,and due to the intrinsic low risk of maritime operations with UAVs,
an important increase in the use of UAVs from ships is foreseen for different
applications: environmental monitoring, fishing support, surveillance, etc. In operations
from ships and boats, the landing maneuver is the phase of the operation that involves
a higher risk and where a higher level of precision in the position and velocity
estimation, along with a high level of robustness in the operation, is required. Although
landing of rotary-wing UAVs has raised the attention of multiple researchers during the
last 15 years [2], it has not yet been completely solved in a robust and reliable manner.
In fact, one of the three challenges at the Mohamed Bin Zayed International Robotics
Challenge (MBZIRC) Competition in 2017 was the landing of an UAV on a moving
platform [3]. One of the first research works to tackle the autonomous landing problem
on mobile platforms was by Saripalli et al. [4] in 2003, who developed a real-time,
vision-based landing algorithm. Later, Saripalli and Sukhatme [5] and Merz et al. [6]
presented vision-based algorithms that can estimate the helicopter position from images
of a specially designed landing pad with the necessary level of precision and accuracy
for the landing maneuver. Most data were gathered from manual flights and simulations.
More recently, Falanga et al. [7] and Polvara et al. [8] also applied vision-based landing
techniques increasing the autonomy level by including real-time trajectory generation. In
addition, it is possible to find recent works that study the problem of landing in an
oscillating platform by using visual sensors. For instance, the authors of [8] used a
fiducial marker to obtain the pose of the platform and implemented an External Kalman
Filter (EKF) to estimate the ship position. Simulations provide
very accurate results by using only the odometry and the inertial measurements for the
estimation.
However, measurements from these sensors may not always be available and they can
suffer physical interferences and limitations in their fields of view. Additionally, these
approaches generally only work under good light or visibility conditions so their
performance has a strong dependency with the weather conditions, seriously limiting
their applicability under a wide range of realistic scenarios for the landing operation.
Moreover, some of these vision-based techniques assume the use of the Global
Navigation Satellite System (GNSS) in their navigation systems, since the vision system
is only used as an input that increases the precision of the relative estimation of position
and velocity with respect to the landing platform. Therefore, the robustness of these
solutions is compromised due to the known problems of current GNSS systems in
cluttered environments, such as the deck of the ships. The use of a cable or tether that
physically connects the UAV and the moving platform is not new, although as explained
in the following, most existing works are theoretical and only with simulation
results. Some works are focused on the study of the control and stabilization problems
of tethered rotorcrafts (e.g., [9]). In addition, in [10], a study can be found on the
longitudinal stability of a hovering tethered rotorcraft. The first use of a tether for helping
in the landing phase of an UAV can be found in [11] where a controller is developed to
use the tether tension to couple the translation of the helicopter to the rotation. In [12],
the authors described the design of a flight control system architecture for a tethered
quad-rotor aircraft where the tether is also used for transmitting power to the
UAV. Results from simulated waypoint navigation and hovering of the tethered vehicle
suggest that the designed system is fit for use in an automated landing missions. In
[13], the authors took advantage of the tensile force acting along the taut cable and
solved the nonlinear control of the tethered UAV by using a cascade control scheme
based on thrust vectoring and using a novel “Reference Governor scheme”. The work
presented in [14] is also focused on the control design problem of a tethered drone and
uses the tether as a position sensor. Simulations results show that a full autonomous
flight could be achieved indoors by using its approach. However, in all these works, the
tether is only used for control and stability purposes and all results and conclusions are
obtained from simulations. There are very few experimental results regarding the use of
a tether with an UAV for landing purposes. The first experimental results can be found
in [15], where a tethered helicopter lands autonomously over a static landing platform in
an outdoor scenario. This work is mainly focused in the control strategy and the
advantages that a tether can provide to the stability of an UAV. In [16], a power-feeding
tethered micro UAV is used and a position estimation method based on observing the
slack tether is proposed. Some indoor experiments are carried out to prove the
feasibility of this method. The authors of [17] proposed localizing an UAV in indoor
environments by using only a quasi-taut tether. The tether’s sensory feedback is fed into
a catenary-based mechanics model to localize the UAV in an indoor global frame
defined by the tether reel center. They tested their localization method on a physical
robot (Fotokite Pro). Although it is possible to find real experimental data in [16,17],
landing tests in these works are performed over static platforms and in indoor scenarios
where the weather conditions do not affect the navigation capabilities. The system
developed in this work was designed for landing a rotary-wing UAV (RUAV) on a
mobile platform with a high level of accuracy and robustness, and without using GNSS.
The design was inspired by the Recovery Assist, Secure and Transverse (RAST)
system used by manned helicopters to improve the stability using a tether during the
landing operation [18]. This study was a large extension of the preliminary results
presented in [19–21], where the main contributions are: the use of the tether to estimate
the relative position and velocity between the UAV and the mobile
platform without using GNSS; the design of a robust guidance approach to perform
landings in a safe and robust manner; and the validation of the designed algorithms in a
large number of flying tests, including landings with speeds of the mobile platform up to
40 km/h. It is important to pointout that the calculation of an accurate estimation of the
relative position and velocity of the UAV with respect to the landing platform for this
navigation solution is obtained by using an altimeter, the inertial data from an on-board
Inertial Measurement Unit (IMU) and the tether orientation. This relative information is
completely independent from the GNSS and represents an alternative, low cost and
reliable positioning system for tethered helicopter UAVs or multicopters. Therefore, the
main contributions of this paper are twofold: It presents a robust estimation method
based on a novel tether system designed, manufactured and integrated into a rotary
wing UAV that provides relative measurements at 100 Hz with centimeter accuracy,
and, to the best of our knowledge, it includes the first field experiments with a tethered
unmanned helicopter landing on a mobile platform by using the tether as its only
positioning source. The paper is organized as follows. The developed GNSS-free
landing system is described in Section 2. Section 3 details the rotary wing UAV and the
equipment used during the real tests, whereas Section 4 presents the tests that were
carried out to validate the system developed along with the experimental results.
Section 5 closes the paper with the conclusions and future developments.

2. GNSS-Free Landing System Description

A typical mission of a RUAV is split in several phases. The helicopter takes-off from its
base, it flies to an area for mission execution usually using waypoint navigation, and
once the task is completed the RUAV starts the approach to the landing area. Finally,
once the helicopter is over the landing location, it starts the descent until it lands. The
majority of the navigation strategies employed in autopilots are based on fusing the
GNSS information with the navigation solution calculated using the accelerometers and
gyros of the inertial measurement unit. In fact, this is the most common strategy for
taking-off, and the waypoint navigation phases, where the positioning in absolute
coordinates provides enough accuracy for performing the different maneuvers.
However, if the landing phase has to be performed on moving platforms, a more robust
strategy is required based on more accurate sensors. Moreover, the use of a GNSS-
free landing system increases the robustness of the system by providing more sources
of positioning, especially in non-GNSS friendly environments. Regarding the different
phases of a mission mentioned above, in this work, it is assumed that the helicopter has
reached the landing area by using GNSS based navigation and that the rope
preparation phase has been completed: the tether has been deployed from the
helicopter and locked into the device installed in the landing platform for controlling the
tether tension and velocity. This phase is shown in Figure 1.
Figure 1. Tether deployment during the rope preparation and detail of the device on-
board the landing platform for tether control.
The work and the experiments presented in this article focused on the landing
maneuver of the RUAV. The different steps that compound this phase are summarized
below:
Initial condition: The rotary wing UAV is flying autonomously based on GNSS
navigation with the tether already attached to the moving platform. The tension applied
in the tether is low, so the tether can freely slide.
Step 1: Once the UAV is over the platform, the altimeter is activated and the autopilot
maintains the relative altitude to the platform.
Step 2: The ground controller applies a predefined value of tension to the tether to
keep the tether straight. This tension is constant during the whole operation.
Step 3: At this status, the tether is tense and the helicopter is maintaining the relative
altitude to the platform by using the altimeter. At this moment, it is possible to switch the
navigation strategy from GNSS-based to the tether navigation system.
Step 4: The helicopter follows the movements of the landing platform maintaining the
relative horizontal position and altitude commanded by the guidance system.
Step 5: The helicopter moves in the horizontal plane towards the origin of the landing
platform (point where the tether is attached on ground). In this position, the rope keeps
vertical. Once the UAV reaches this point, it is possible to start the descent maintaining
a constant vertical velocity.
Step 6: Landing procedure is finished when the rotary wing UAV is standing on the
platform and the rotor stops. GNSS-Free Navigation Algorithm
One of the main contributions of this work is the relative navigation strategy by using a
tether for the landing phase of the rotary-wing UAV. To obtain useful information from
the tether, a specific device was developed. It consists of two-axis coupled cardan joints
that allow estimating the angles
between the tether and the helicopter frame in terms of the two successive rotations
and a load sensor to measure the tension level of the tether.
Three different reference frames were considered in this study (see Figure 2):
Body frame (B): The body frame is a non-inertial coordinate system associated with the
vehicle with the origin at its center of gravity. The x-axis points in the forward direction,
the z-axis down through the vehicle and the y-axis completes the right-hand coordinate
system. This frame is denoted by the superscript b.
Local Navigation Tangent Plane frame (N): This is an inertial coordinate system
determined by fitting a tangent plane to the geodetic reference ellipsoid at a fixed point.
This point is taken as the origin of the coordinate system. The x-axis points to the true
north, the y-axis points to the west and the z-axis points up. This frame is denoted by
the superscript n.
The tether frame (T): It is a non-inertial coordinate system associated with a cardan joint
mechanism. It has its origin in the point where the tether is connected to the helicopter.
The x and y-axes rotate with respect to the fuselage of the helicopter and the z-axis is
always pointing towards the landing point. This frame is denoted with superscript t.
Figure 2 shows a scheme with the different elements that play a role in the landing
procedure presented along this work.

Figure 2. Landing scenario frames and elements that play a relevant role in the landing
procedure presented in this paper. The center of gravity of the RUAV is designed as
CG, the Contact Point (CP) is the location where the tether system is installed in the
fuselage of the UAV, Hal is the altitude above
ground level (AGL) measured by the altimeter, hcp is the AGL in the CP, lt is the
longitude to the landing point O, leg is the lever arm between the CG and the CP and
last is the lever-arm between the CP and the altimeter. For the sake of clarity, Figure 3
shows a model of this device with the tether and body coordinate systems represented
over it.

Figure 3. Tether device in which the tether frame T is represented with respect to the
body frame B.

The precision obtained by usual algorithms based on fusing a GNSS sensor with an
Inertial Navigation System (INS) is not enough to perform a safe approach and landing,
especially in moving platforms. Therefore, a technique to estimate the position in real
time with high accuracy is needed in order to successfully accomplish the autonomous
landing safely. In our study, a relative estimator was developed and implemented in an
autonomous helicopter in order to be used during the landing phase on static or mobile
platforms. Figure 4 shows the architecture of the relative estimation module.
As can be seen, the inputs to this module are the data provided by the tether system
(angles h and r, and tension T), the altitude of the altimeter (halt), the accelerations and
angular velocities of the INS (a b and w) and the magnetic field measurements of the
magnetometer (mb). This scheme is composed by:
Attitude and Heading Reference System (AHRS) block is the module in charge of
calculating the attitude of the UAV (roll f, pitch q and yaw y). The attitude is defined as
the inclination of its body-axes reference frame to the navigation reference frame. In
addition, in this block, the accelerations are rotated to the navigation axes(an).
Tether Conversion block performs all the geometric and rotation operations needed to
translate the tether information to a relative position vector (prelims). Sensor Fusion
block fuses all the information obtained from the AHRS and the Tether conversion
block and estimates the relative state vector that is used by the controller of the RUAV
Figure 4. Architecture of the relative estimation system.

In our work, a crucial requirement for the estimation module is to obtain a precise
tracking of the relative position and velocity between the helicopter and the landing
platform. Most of the relative kinematics works are based on the fact that both vehicles
have an external positioning source (generally GNSS) and they can share their own
information through a communication link. This is very common for example in leader–
slave architectures for formation flights [22] where the system model uses information of
the state vector received from the other vehicles and the relative position
measurements are obtained by using a Differential GPS architecture. However, in this
work, we did not rely on communication links or external very accurate positioning
systems. In this way, as the true behavior of the vehicles was not known, the control
input of the relative state vector was modeled as a random process with certain
properties. To build the model for the relative estimation in the approach maneuver, we
had to take into account that the filter does not have any information about the dynamic
of the landing platform. In this case, we chose to use a stochastic dynamic model of the
relative vector between the vehicles, where a random variable represents an unknown
time-varying quantity. In particular, our system model is a modified version of a Singer
acceleration model. The Singer acceleration model [23] is a popular model [24,25] for
target maneuvers that characterizes the unknown target acceleration as a time-
correlated stochastic process. It is an a priori model since it does not use online
information about the target maneuver, although it can be made adaptive through an
adaptation of its parameters. In this case, the acceleration is modeled as a zero-mean
first-order stationary Markov process with an
autocorrelation function [26]:
Ra(t) = E[a(t + t)a(t)] = s2e�ajtj,
(1) where s2 is the variance process noise and a is the reciprocal of a maneuver time
constant t that depends on how long the maneuver lasts, for instance, in a slow turn of
an aircraft t, _60 s, and, in an evasive maneuver t, _10–20 s [23]. In a Markov process,
its value at a given time depends on values at other times only through its nearest
neighbors. To provide values for these parameters, some typical simplifying
assumptions for the ship model were taken [27,28]: the landing platform follows a
straight trajectory with (nearly) constant velocity. Regarding the helicopter, it was
assumed that the
autopilot is capable of following the ship in a soft way during the maneuver (this last
assumption was proved by the tests that are presented in Section 4). One of the
shortcomings of the Singer model is that the acceleration has a zero mean at any
moment [26]. However, we could use information from the inertial sensors on-board the
RUAV, so some modifications can be done in the model in order to overcome this
limitation. In the landing scenario, as the ship was assumed to have a slow dynamic,
most of the changes in the relative velocity between both vehicles are due to the
accelerations of the helicopter. These accelerations are not zero and can be measured
by the accelerometers on-board.
Robotic System for Inspection by Contact of
Bridge Beams
Abstract: This paper presents a robotic system using Unmanned Aerial Vehicles (UAVs) for bridge-
inspection tasks that require physical contact between the aerial platform and the bridge surfaces, such
as beam-deflection analysis or measuring crack depth with an ultrasonic sensor. The proposed system
takes advantage of the aerodynamic ceiling effect that arises when the multirotor gets close to the
bridge surface. Moreover, this paper describes how a UAV can be used as a sensor that is able to fly
and touch the bridge to take measurements during an inspection by contact. A practical application of
the system involving the measurement of a bridge’s beam deflection using a laser tracking station is
also presented. In order to validate our system, experiments on two different bridges involving the
measurement of the deflection of their beams are shown.

1. Introduction
In recent years, interest in Unmanned Aerial Vehicles (UAVs) has grown exponentially [1]. These
aerial robots have been used in a huge variety of applications, such as detection, exploration, monitoring,
precise localization, or assistance in natural catastrophes. Nevertheless, in most of these applications,
aerial robots have merely a perceptive role, carrying cameras and other optical sensors for environment
sensing. Thus, these robots do not have any physical interaction with the environment.
Recently, the development of autonomous aerial robots capable of physically interacting with
the environment is attracting the interest of researchers in the aerial robotics field [2,3]. For
example, some researchers have embedded small and lightweight serial manipulators [4–6],
offering a new innovative solution for applications such as the inspection and maintenance of
industrial facilities or aerial power lines. The authors of Reference [7] proposed the use of UAVs for
the assessment of post-disaster situations such as earthquakes.
The inspection of reinforced concrete bridges involves different tasks that are performed by
inspectors to obtain information for the structural assessment of bridges. Inspectors usually employ
ladders, scaffolding, or lifters to reach the parts of beams and piers of the bridge that are not easily
accessible. One of the tasks is visual inspection in order to detect cracks, rebar corrosion, delamination,
and other defects. Other tasks require direct contact of the sensor or device with the bridge surfaces. One
of these contact tasks is the measurement of beam deflection with and without load, which gives
information on the internal state of the bridge. This is usually done with an operator manually placing a
reflector prism attached to a pole in contact with different points at the beam and then
 2 of 17

measuring the position of the prism with a laser total station . Other tasks that require contact are
the measurement of crack depth with an ultrasonic sensor, and the measurement of crack width
with tactile sensors .
Previous studies proposed autonomous data-collection systems to overtake the less-efficient
and costlier current method of manual bridge inspection, including novel methods to automatically
assess cracks based on the combination of image processing and terrestrial laser scanning. The
authors of Reference described the use of a robotized system for autonomously acquiring data from
the top of a bridge to evaluate its status with nondestructive techniques. Some studies have
proposed solutions using Unmanned Ground Vehicles (UGVs) for robotic crack inspection and
mapping, delaminating and concrete quality assessment, and even a complete mechatronic system
for high-efficiency bridge inspection . The authors of Reference also presented a semi-autonomous
robotic system focused on road and railway bridges.
The use of UAVs for bridge inspection is not as extended as the use of ground vehicles, but
has increased significantly in previous years. UAVs have been mostly used for the visual inspection
of bridges, since standard UAVs have to fly at a safe distance from bridges.
Aerial manipulators are usually composed of an aerial platform and an articulated robotic arm.
The authors of References developed a system with an arm attached to the top of the multirotor,
which was used to conduct inspection tasks that require contact with the bridge. However, payload
requirements for this kind of manipulator are high, resulting in large and heavy platforms that are
slower and more complex to control. This paper proposes a small and lightweight multirotor design
that is capable of flying close to the bridge and sticking to it thanks to the aerodynamic ceiling effect
to perform contact inspection tasks (Figure 1). This aerial robot was designed under the framework
of the AEROBI European Project, which aims to develop an aerial robotic system that can perform
both visual and contact bridge inspection tasks.

Figure 1. Prototype of multirotor system operating in contact with the ceiling.

Airflow generated by rotary wings is sensitive to the physical obstacles close to it. This
modification of the airflow changes the resulting thrust and torque generated by the robot’s engines.
This effect can be observed when UAVs take off, land, or even fly at very low altitude [22,23]. The
presence of the ground perturbs the natural motion of the air, known as the ground effect. This
effect is characterized by the production of an increment in the thrust generated by the rotors
without increasing the power. A similar effect is caused by flat surfaces placed above the rotors,
known as the ceiling effect. This phenomenon induces an additional thrust on the robot, pushing it
towards the ceiling. This fact can be dangerous for standard multirotors because propellers and
engines may collide with the ceiling and break, or destabilize the platform.
This paper describes an integrated solution using a laser-tracking station and a UAV that are able to
carry out inspection tasks. Thanks to the ceiling effect, the UAV is able to remain in contact with the
bridge deck, and beams from below using a protective frame. This frame acts as armor that protects
 3 of 17

rotors when the robot is in contact with the beams, taking advantage of the ceiling effect.
Meanwhile, a laser-tracking station in the ground accurately measures the position of a reflector
prism mounted on the UAV. These measurements are later used to estimate bridge deflection.
The total station is an optical and electronic instrument used in building construction and surveying.
This device can measure distances and horizontal/vertical angles from it to a single far away point. This
point can be any surface or a reflector prism. Total stations are used by civil engineers to record features
in topographic surveying. Archaeologists also use them to record excavations. Another situation in which
it is essential to have environment reconstruction is in a crime scene or a traffic accident, where the
police or insurance companies need to take measurements of scenes.
A short version of this paper with preliminary results was presented at ICUAS 2017 [. Extensive
testing is presented in this paper, including the laser-prism tracking-system results, the effects on
UAV magnetometer sensors while flying close to reinforced concrete structures, and experimental
contact tests on different bridges.
The remainder of the article is organized as follows: Section 2 presents the general
architecture of the proposed robotic bridge-inspection system. This shows the concepts of the
system and the performance during the inspection. Section 3 presents the requirements of the
aerial platform and the solution proposed for carrying out the bridge-beam inspection system.
Section 4 focuses on the inspection sensors that are used in taking measurements of the bridge
beam deflections. Section 5 is about the bridge-inspection experiments and the validation of the
system. Lastly, conclusions and future works for this research are discussed.

2. Robotic Bridge-Inspection System

This section presents robotic bridge inspection by a contact system using a UAV that is
proposed in this paper. The system focuses on measuring the deflection of bridge beams using a
UAV as a flying sensor that is able to touch the bridge in order to get the best possible measure with
an aerial platform. Traditionally, deflection measurements of bridges have been carried out by a
human operator who manually places the reflector prism at several points under the beam. Thus,
this research proposes an autonomous system that uses a UAV with an inspection sensor onboard
to make this task easier, faster, and safer for people (see Figure 2).

Figure 2. Inspection by manual contact task vs. with an Unmanned Aerial Vehicle (UAV).

The inspection system is composed of an aerial platform, which has a reflector prism mounted
onboard that acts as an inspection sensor, a ground-control station to monitor the state of the aircraft,
and a laser-tracking station which follows the position of the reflector prism and takes the deflection
measures of the bridge beams. Figure 3 shows the hardware architecture of this system. During
inspection, the UAV is commanded to touch a specific point on the bottom surface of the bridge with the
fairing of the rotors; this is the target inspection point. Once the multirotor establishes the contact with the
bridge, it remains stuck to it during the inspection phase. Thus, the onboard reflector prism is also in
contact with the bridge while the inspection task is carried out. Meanwhile, the total station takes several
high-precision measurements using the position of the reflector prism mounted in the UAV. These
measurements can be used to estimate the deflection of the bridge and the changes with
 4 of 17

respect to the bridge design. This concept presents the aerial platform as a flying sensor that can
obtain high-precision measurements by putting the reflector prism in a target inspection point to
perform the bridge inspection by contact. Moreover, one of the advantages of including a robotic
total station and a reflector prism in this robotic inspection system is that it allows the establishment
of a fixed reference frame over time so that the results obtained at different inspection points for
each bridge can be compared. This is explained in detail in Section 5.

Figure 3. Robotics bridge-inspection system. (A) Aerial platform, (B) onboard reflector prism, (C)
total station, and (D) ground-control system for monitoring and collecting flight and inspection data.

In summary, this robotic system is composed of two main parts, the aerial platform (on the left)
and the total station (on the right), as well as the reflector prism. Each one is explained in its
corresponding section below.

3. Aerial Platform
Bridge inspection usually requires high precision in taking measurements during the inspection
task. This requirement imposes strong constraints from an aerial-platform design point of view. In
this paper, an aerial robot that is able to touch and maintain contact with the beams during
inspections under safe conditions, and prevents all movement, is proposed. Thus, there are no
errors derived from the motion of the aircraft during the inspection. Moreover, the maximum size of
the aerial platform is limited by the size of the bridge beams that are going to be inspected. The
multirotor should fit in the beam width in a full contact condition, in which the entire upper body is
touching the bridge. This section is about the design of the aerial platform and how this design can
deal with the size and the contact constraints.

3.1. Ceiling Effect

As was mentioned in the previous section, the kind of inspection proposed in this paper requires
that the multirotor flies very close to the bridge beams where the aerodynamic disturbance produced by
the presence of a ceiling-type surface begins to be significant. This aerodynamic disturbance is known as
the ceiling effect in the literature [25–27]. This ceiling effect has been studied by the authors and was
modeled for a single-rotor working close to a ceiling surface in Reference [24] as follows:

TICE(z) 1
Kc(z) = T
OCE
= 1 1( R )
2 (1)
K1 z+K2
where TICE(z)/TOCE is the ceiling factor Kc(z) that models the changes between the thrust of a rotor,
known as the ―in-ceiling effect‖ (TICE), and the thrust of a rotor, known as the ―out-ceiling effect‖ (TOCE).
R is the radius of the rotor, and z is the distance from the rotor to the ceiling. K 1 and K2 are
 5 of 17

experimental coefficients obtained by the least-squares method as K1 = 6.924 and K2 = 0.03782(m)

for this aerial robot. This ceiling factor was presented by the authors in Reference [18].

3.2. Multirotor Design

In applications that need the UAV to fly very close to a surface with a sensor touching it, as is
the case for concrete-bridge inspection with ultrasonic and laser sensors, it is challenging to
maintain the stability of the UAV with sufficient positioning accuracy for taking measurements. Thus,
this paper proposes the design of a custom multirotor that takes advantage of the ceiling effect to
stick firmly to the bridge’s surface. In this way, the stability of the multirotor when taking
measurements with the sensor is significantly improved. Another compelling advantage is that once
the multirotor is stuck to the bridge, the rotation speed of the rotors can be lowered, reducing energy
consumption and increasing flight time due to the ceiling effect. However, most aerial platforms are
not designed to maintain contact with a surface with their upper body. Some of them may touch the
surface with the propellers, leading to dangerous flight conditions, and others have antennas or
sensors that stand out without any protection. Therefore, it is necessary to design a new multirotor
configuration that can safely touch the bridge surface, maintain contact during the inspection, and
provide protection from impacts to the upper part and the propellers of the aircraft.
The multirotor proposed in this paper is based on a cross-layout quadrotor with a special fairing
surrounding the propellers. All electronic components (autopilot, batteries, sensors, and electronics)
are placed at the rotor’s plane or below it, as shown in Figure 4. This design allows the multirotor to
safely maintain contact with the lower part of the bridge. The fairings are designed so that while in
contact with bridge beams, the rotors spin a few centimeters from the surface without colliding with
it. The proposed design sets this value to 0.36R.

Figure 4. 3D Design of the platform.

A DJI E305 motor kit from the Chinese technology company SZ DJI Technology Co., Ltd. was
selected with a power supply consisting of a 4s LiPo battery. This set delivers a nominal thrust of
approximately 400–450 g per axis. The total weight of the platform is about 1.5 kg with a maximum flight
time of 14 min. The fixed distance between the rotors axis is 480 mm, and the total size of the platform is
about 600 mm, which fits most of the beams and complies with the inspection requirements.
The fairings surround each rotor with a cylindrical shape to ensure the safety of the propellers and
the multirotor itself. The diameter of each fairing is 280 mm, and height is 74 mm. The top of the fairing is
covered with a rubber material to prevent slippage on the ceiling surface and to dampen the contact with
the surface. As can be seen in Figure 4, these fairings present air intakes around the circumference to
ensure thrust is generated by the rotors when the multirotor is stuck to the ceiling. Moreover, this fairing
isolates the aerodynamic effect between different rotors and allows the aerodynamic effect of the rotor to
be modeled, assuming that it is working without any disturbance from the other rotors. This is because,
as was presented by the authors in Reference [24], it was expected that there would not be much
difference between the single-rotor and the complete-quadrotor cases. This was taken as a preliminary
design to validate the presented prototype in this paper. Nevertheless, the optimization of the fairing
design in order to improve the weight and the aerodynamic performance will be studied
 6 of 17

in future works. It is important to remark that the fairing makes the aircraft more sensitive to wind
and limits the operation of the aircraft to wind speeds below 20 km/h.
Figure 5 shows the final prototype of the quadrotor with the fairings surrounding the rotors. The
fairings were built in polylactide (PLA) with 3D-printing technology. Front rotors were printed in
orange and rear rotors in black.
The platform cannot carry out the bridge inspection by itself. So, this prototype has a custom socket
to mount the inspection sensor onboard that is used by the laser system to accurately measure the
position co-ordinates of each point in the bridge. The prism platform has four silent blocks that are
compressed in the contact condition. These have two functions. First, they guarantee the contact of the
prism because they overhang the fairings that, during contact, could be a bit compressed. Second, they
adapt the prism contact to the bridge surface in cases where the bridge has a soft slope or small defects.

Figure 5. Final prototype of the platform and prism platform detail.

3.3. Other Effects in Sensors

In addition to the aerodynamic effects, other problems arise when an aircraft flies in narrow
places or under structures like bridges. These problems are mainly related to the specific autopilot
system that controls the aircraft and the constraints suffered by the integrated sensors onboard for
position and attitude estimation. The main problems arise in the Global Positioning System (GPS)
signal and the magnetometer measurements. GPS signals are partially or fully blocked under
bridges. Hence, these signals cannot be used to feed the position estimator. Additionally, reinforced
concrete-bridge armatures significantly change the magnetic field measured by the onboard
magnetometers. Consequently, this measurement of the magnetometer cannot be used reliably for
attitude estimation. This effect is better explained through telemetry data during flights in Section 4.
The experiments of this paper were carried out using the PX4 that is an open source flight
stack obtained from . However, some modifications had to be included in the estimation and the
control algorithm. The main changes in the estimation module were due to mentioned constraints in
the GPS and magnetometer and are extended below. Relative to the control part, it is necessary to
remark that the position controller cannot use the GPS signal because it might not be available
under the bridge. Instead, optical flow or visual odometry can be used for relative position
estimation. Besides, during the contact, it is necessary to control the integral errors of the attitude
controller because the multirotor is never in contact with a perfectly null tilt. Thus, if the integral
attitude controller is on during contact, the separation maneuver can be very aggressive and
dangerous due to the high number of integral errors.
The attitude estimator must be adjusted because of the errors in the magnetometer. This
disturbance is caused by the possible metallic structure of the bridge or the presence of cables close to or
inside it, among other reasons. Thus, the measurements of this sensor are very distorted while flying
close to the bridge surface and this is tough to model, so it cannot be used to estimate the attitude of the
multirotor. This fact mainly affects the accuracy of the yaw orientation angle estimation. Concerning the
attitude control of the autopilot, only the yaw controller was modified to prevent the controller saturating
due to error accumulation when the multirotor is stuck to the ceiling. This error accumulation arises in this
condition because the multirotor cannot correct this error and starts to
 7 of 17

saturate the mixer. To guarantee that contact occurs under safe conditions, a yaw rate controller
without an integral term is used during the contact.

4. Total Station and Reflector Prism

A total station is a robotized sensor used to measure distances with laser technology. It can
measure the distance to any surface or a reflector prism using a highly accurate laser system. This
tool is often used in topography and building construction [31]. The tool has a laser emitter and
sensor that measures the distance to a point that can be oriented in azimuth and elevation angles.
Using these angles and the laser measure, the total station triangulates the positions of the desired
points in local axis, which are defined by the user, with high accuracy.
Even if the total station is able to measure the distance to any object or surface, when it works
with the reflector prism, it generates more accurate measurements. There are several different types
of prism, such as flat prisms, which are the most frequently used in topography (thought to work at
larger distances), and the 360-degree reflector prisms. The latter are of high interest for UAV
applications. The main advantage of this kind of prism is that it is not necessary to orient it in a
specific direction, i.e., it reflects the laser beam in any direction in order to measure its position,
although the accuracy is a bit lower.
The model used for this project was the MS50 from the Swiss company Leica Geosystem [31]
(Figure 6). It has a theoretical accuracy of 1 mm for the prism and 2 mm for any surface. The
maximum distance measurement is 3000 m for 360 prisms and 10,000 m for flat prisms. The prism
used in this project was a 360 miniprism from Leica (Figure 6), which does not need to be oriented
to obtain its position. To be more precise, the prism works at 360 in the vertical axis of the prism
and, starting from the horizontal plane, it covers 60 .

Figure 6. Total station and 360 mini-prism. (Left): actual laser system and the reflector mini-prism
360. (Right): scheme of the utilization of the laser tracker to obtain the UAV position.

The total station is placed on a tripod that allows it to level and produce reliable measurements.
Then, the laser system can track the prism and continuously monitor it. This is possible thanks to a
pair of high-accuracy motors with encoders that orientate the device towards the prism. The UAV
trajectory can be obtained through the prism trajectory by a constant affine transformation that is
designed with the overall design of the platform.
Additionally, as the laser emitter can be used to measure the distance to any surface, the Leica
Total Station can be used to acquire 3D laser scans of the bridges. However, these scans usually take
long periods of time, as each point is captured individually. Additionally, the origin of the point cloud is
located in the original position of the TS. This fact can be a problem for the inspection, because
measurements need to be consistent over time. For this reason, each of the maps is always recorded
with a set of landmarks that are used to reallocate the total station before each experiment. This means
that it is not necessary to place the tool in the same place, but instead, each time the tool is placed, the
relative transformation between the original location and the current one is computed by manually
 8 of 17

identifying the landmarks defined within the map. Thus, the 3D dense map can be reused, saving time in
all the future experiments, and all the locations and measurements are consistent over time. Figure 7
shows an example of the 3D point cloud captured by the total station. Additionally, landmarks have been
highlighted with large red spheres. The actual landmarks are single points in space.

Figure 7. Scan of bridge and landmarks to determine consistency in the experiments.

In order to give a quantitative analysis of the performance of the total station when the prism is
built in the UAV during a flight, a set of measurements was carried out to compare the manual and
aerial inspection, as shown in Figure 2. This experiment resulted in two scattered sets of points. To
compare them, the mean of each of these sets was subtracted and these were plotted aside each
other, as shown in Figure 8. It can be seen that the static measurement has less dispersion.
Nevertheless, the variance of the experiment stuck while flying was 0.1 mm, with the maximum
distance between any pair of points being lower than 1 mm.

Figure 8. Comparison of the prism of the total station during flying inspection and manual inspection.

These values are valid regardless of wind conditions, as long as they are moderate and allow
the flight to be carried out under safe conditions, because during the contact phase, the platform
can develop enough force to stay in contact with the ceiling and to prevent movement of the aircraft.
 9 of 17

5. Experimental Validation
This section presents an experimental validation of this robotic system for bridge inspection by
contact using UAVs. The experiments presented in this paper were carried out on two bridges with
different features. These were the Cartuja and the Grazalema bridges.
The Cartuja bridge (see Figure 9) has a reinforced concrete voided slab as its superstructure,
which is totally flat on the bottom surface. The height of this bridge is approximately 4 m, and the
substructure is made up of reinforced concrete piers with a rectangular section. This kind of bridge
is the simplest type for carrying out inspection by contact due to its flat surface.

Figure 9. Cartuja bridge.

The Grazalema bridge (see Figure 10) has a reinforced concrete superstructure that is simply
supported by I-beams. Because of this, the bottom surface of the bridge shows beams and spans
(between the beams) at different height levels, and the inspection task is more complicated. The
height of the bridge is around 14 m, and the substructure is made up of reinforced concrete walls.

Figure 10. Grazalema bridge.

Figure 11 shows three sample sequences of the experiments with the proposed system for bridge
inspection. The first sequence was carried out in the Cartuja bridge that has an entirely flat bottom
Sensors 2019, 19, 305 10 of 17

surface (Figure 9). In the second and third sequences, the multirotor operated in the Grazalema
bridge (Figure 10). In the second sequence, the multirotor inspected a bridge beam, and the third
sequence shows that the aerial platform could also inspect the span between the beams.

Figure 11. Sequences of the inspection by contact applications in three different kind of bridge. (a)
Application for the Cartuja bridge (flat surface), (b) application for the Grazalema bridge (beam
inspection), (c) application for the Grazalema bridge (gap inspection).

These experiments validate the design and confirm the benefits of using the ceiling effect in the
applications discussed in Section 2. The results of the experiments are shown in Sections 5.1 and
5.2. A video compilation of the experiments can be found in Reference .

5.1. Cartuja Bridge Results

The results of Cartuja’s bridge-inspection task are presented in Figures 12–17. These figures show
an experiment in which the UAV first approaches the bottom surface of the bridge superstructure and
then sticks to it. Later, it releases from the bridge and flies to a second point, coming in contact with the
bridge and maintaining contact for several seconds. Finally, the UAV flies away from the bridge and
lands. The first period of time in which the UAV is in contact with the bridge is colored in green in the
figures, and the second contact period is colored in cyan. In both cases, the results are similar because
the contact surface is flat and horizontal (see Reference ).
The presented results are a combination of the flight data and the laser-tracker data collected
through telemetry and the laser tracker, respectively. The abscissa axes of these figures represent
flight time in seconds.
The results presented in Figure 12 show that the z-accelerometer is triggered at the start of the
contact condition due to the initial impact when contacting the bridge. Thus, this sensor can be used
to detect the beginning of the contact condition. The design presented in this work guarantees that
this impact is not a risk for UAV safety.
 )
11 of 17

2
-
m
s
(
Z-Accelerometer

A
c
c
20
Contact1 Contact2 AccZ
10

-10

-20
15 20 25 30 35 40
Time (s)

Figure 12. Z-accelerometer—Cartuja bridge experiments.

The next results are related to multirotor’s attitude during contact. In Figure 13, it can be
observed that the angles remain steady while the multirotor is stuck to the bridge. During contact,
the multirotor stays still, allowing the inspection task to occur without requiring attention to the
control of the aerial robot. This is one of the most important results, because the fact that it stays in
contact decreases the motion blur that UAVs can suffer during this type of operation.

Pitch and Roll

5.7

2.9

0
º
(

-2.9

-5.7
Contact1 Contact2 Pitch Roll
-8.6
15 20 25 30 35 40
Time (s)
Figure 13. Pitch and roll—Cartuja bridge experiments.

Figure 14 shows the ceiling effect interference during contact. These experiments show that
the thrust necessary to remain in contact is approximately 38%, which is significantly lower than the
thrust needed in the hover condition (approximately 48%). This result is similar for the two contacts
because the aerodynamic problem is similar in both cases.

% Throttle
60

50

40

30

20

10 Contact1 Contact2 Thrust

0
15 20 25 30 35 40

Figure 14. % Throttle—Cartuja bridge experiments.

 12 of 17

The relation between the hover-thrust ―in-ceiling effect‖ and ―out-ceiling effect‖ during contact
with the bridge is presented in Figure 15, which clearly shows that the aerodynamic ceiling effect is
well-modeled with the expression (1).

Figure 15. Ceiling-factor results. The blue error bar shows the experimental results obtained in a
test bench, the black curve is the numerical approximation, and the red marker represents the
ceiling factor obtained with the flight data.

Figure 16 shows the measurements of the magnetometer during the flight. It can be observed
that the magnetic-flux density is greatly influenced by the metallic structure of the bridge, making
the magnetic sensor unusable for attitude estimation. This provides evidence that the changes
introduced in the autopilot (Section 3.3) are necessary to safely carry out these experiments.

Effect of the contact in magnetometer

0.5
Mag Flux Density (G)

0.4

0.3

0.2

0.1

0
Contact1 Contact2 MagX MagY MagZ
-0.1
15 20 25 30 35 40
Time (s)
Figure 16. Magnetic-flux density—Cartuja bridge experiments.

The last figure presents the results of a complete inspection experiment. Figure 17 shows the
trajectory of the prism measured by the laser tracker during the flight during the inspection task
overlapped with a point cloud of the bridge. In this case, the UAV touches the bridge five times and
stays in contact for five seconds each time. The measurements of the contact point can be used to
estimate the deformation of the bridge and complete the inspection task. This result confirms that
inspection is possible with the prototype presented in this work.
 13 of 17

Figure 17. Prism trajectory measured with the laser tracker—Cartuja bridge experiments.

5.2. Grazalema Bridge Experiments

The results presented in this section are the telemetry data collected during the bridge-
inspection task at the Grazalema bridge (Figure 10). These results are presented in Figures 18–22.
The two contact conditions were different in this case. The first contact (green background)
was carried out between two beams, and the second contact occurred at the bottom part of the
beam (cyan background). This experiment provides evidence related to two different problems. In
the first contact, the aerodynamic problem is more constrained because the UAV needs to fly in a
more confined space (second experiment of Reference [32]). However, in the second contact, the
available area was narrower, making it difficult to perform the contact with the whole fairing (third
sequence of Reference [32]); the cylindrical design of the fairings solves this issue.
The results presented in Figures 18 and 19 show that the effects in these measurements are similar
to those shown in previous results presented in Figures 12 and 13. The z-accelerometer is triggered at
the start of the contact condition, and the attitude angles remain almost constant during it.

Z-ACCELEROMETER
20
CONTACT1 CONTACT2 ACCZ
10
(MS -2 )

-10

-20
25 30 35 40 45
TIME (S)

Figure 18. Z-accelerometer—Grazalema bridge experiments.

Pitch and Roll

Contact1 Contact2 Pitch Roll
22.9
11.5
º
(

-11.5

25 30 35 40 45
Time (s)

Figure 19. Pitch and roll—Grazalema bridge experiments.

 14 of 17

Figure 20 shows that the effect is different in the two contact conditions. It can be clearly seen
that the ceiling effect is more significant in the first contact (green background). This is a
consequence of carrying out the contact in a span between two beams where airflow is more
constrained by the lateral beams. Therefore, the suction force generated by the ceiling effect is
stronger and the thrust of the rotors has to be lowered significantly more than in the second case for
the UAV to un-stick it from the bridge, as can be seen in the figure.

% Throttle
80

60

40

20 Contact1 Contact2 Thrust

0
25 30 35 40 45

Figure 20. % Throttle—Grazalema bridge experiments.

The measurements of the magnetometer in Figure 21 show that the magnetic-field perturbation
effect close to the structure again appears in a different bridge, and it reinforces the need for the
changes in the autopilot estimator to occur to avoid dangerous crashes.

Effect of the contact in magnetometer

1.5
Mag Flux Density (G)

0.5

0
Contact1 Contact2 MagX MagY MagZ
-0.5
25 30 35 40 45
Time (s)
Figure 21. Magnetometer—Grazalema bridge experiments.

As the last result, Figure 22 presents the prism trajectory measured with the total station during
a real experiment. The UAV contacts the bridge at four points in this experiment, two points in the
span and two in the beam. In the case of an inspection, the number and positions of the points can
be determined by the bridge-maintenance engineer.

Figure 22. Prism trajectory measured with the laser tracker—Grazalema bridge experiments.
 15 of 17

6. Conclusions
This paper proposed an autonomous methodology to carry out bridge-beam inspection by contact
using UAVs. The robotic inspection system presented in this paper takes advantage of the aerodynamic
ceiling effect that was previously studied by the authors. Moreover, possible drawbacks were analyzed.
The design of an aerial platform that meets the inspection requirements and exploits the benefits of this
ceiling effect was designed, built, and tested in real bridge-beam inspections.
The presented design allows the aerial platform to safely stick to the bridge. The stability of the
platform during contact is improved by customizing the fairings. Furthermore, the maximum flight
time is increased using the ceiling effect in favor of the aerial platform. Additionally, the inspection
method and the platform design ensure that the movements of the platform are conservative in
order to increase the precision of the inspection measurements. Moreover, the presented system
focuses on monitoring the state of bridges over time because, thanks to the robotic total station
used, it is possible to directly compare the measures of two different inspections due to them both
being in the same reference frame.
Future work related to this research will focus on the implementation of a dedicated control
system during the contact phase, increasing the system technology readiness level (TRL) and
improving the performance of the aircraft in order to autonomously carry out bridge-beam
inspections. Moreover, the authors aim to validate the inspection system presented in this paper
and to assess the limits of operation by comparing the inspection results with the results obtained
by a classical experimental campaign.

Author Contributions: conceptualization, P.J.S.-C. and G.H.; methodology, P.J.S.-C. and P.R.-S.; software,
P.R.-S.; validation, P.J.S.-C. and P.R.-S.; investigation, P.J.S.-C. and P.R.-S.; resources, G.H. and B.A.;
writing—original draft preparation, P.J.S.-C. and P.R.-S.; writing—review and editing, G.H., B.A., and A.O.;
supervision, G.H., B.A., and A.O.; project administration, G.H.; funding acquisition, G.H. and A.O.
Funding: This work has been supported by the AEROCROS Project, funded by the Spanish Ministerio de
Economia, Industria, y Competitividad (DPI2015-71524-R), the H2020 AEROBI Project, funded by the
European Commission (H2020-ICT-2015-687384), the H2020 RESIST Project, funded by the European
Commission (H2020-MG-2017-769066), and by the Spanish Ministerio de Educacion, Cultura, y Deporte, FPU
Program. The authors acknowledge the collaboration of J. Prada in the experiments.

References
1. Vachtsevanos, G.; Valavanis, K. Handbook of Unmanned Aerial Vehicles; Springer: Berlin, Germany, 2015;
pp. 93–103.
2. Orsag, M.; Korpela, C.; Oh, P. Modeling and Control of MM-UAV: Mobile Manipulating Unmanned Aerial
Vehicle. J. Intell. Robot. Syst. 2013, 69, 227–240. [CrossRef]
3. Fumagalli, M.; Naldi, R.; Macchelli, A.; Forte, F.; Keemink, A.Q.L.; Stramigioli, S.; Carloni, R.; Marconi, L.
Developing an Aerial Manipulator Prototype: Physical Interaction with the Environment. IEEE Robot.
Autom. Mag. 2014, 21, 41–50. [CrossRef]
4. Kondak, K.; Ollero, A.; Maza, I.; Krieger, K.; Albu-Schaeffer, A.; Schwarzbach, M.; Laiacker, M.
Unmanned Aerial Systems Physically Interacting with the Environment: Load Transportation, Deployment,
and Aerial Manipulation. In Handbook of Unmanned Aerial Vehicles; Valavanis, K.P., Vachtsevanos, G.J.,
Eds.; Springer: Dordrecht, The Netherlands, 2015; pp. 2755–2785.
5. Kim, S.; Seo, H.; Kim, H.J. Operating an unknown drawer using an aerial manipulator. In Proceedings of the
2015 IEEE International Conference on Robotics and Automation (ICRA), Seattle, WA, USA, 26–30 May 2015;
pp. 5503–5508.
6. Heredia, G.; Jimenez-Cano, A.E.; Sanchez, I.; Llorente, D.; Vega, V.; Braga, J.; Acosta, J.A.; Ollero, A. Control
of a multirotor outdoor aerial manipulator. In Proceedings of the 2014 IEEE/RSJ International Conference on
Intelligent Robots and Systems, Chicago, IL, USA, 14–18 September 2014; pp. 3417–3422.
7. Recchiuto, C.T.; Sgorbissa, A. Post-disaster assessment with unmanned aerial vehicles: A survey on
practical implementations and research approaches. J. Field Robot. 2017, 35, 459–490. [CrossRef]
 16 of 17

8. Psimoulis, P.A.; Stiros, S.C. Measuring Deflections of a Short-Span Railway Bridge Using a Robotic Total
Station. J. Bridge Eng. 2013, 18, 182–185. [CrossRef]
9. Loupos, K.; Doulamis, A.D.; Stentoumis, C.; Protopapadakis, E.; Makantasis, K.; Doulamis, N.D.; Amditis,
A.; Chrobocinski, P.; Victores, J.; Montero, R.; et al. Autonomous robotic system for tunnel structural
inspection and assessment. Int. J. Intell. Robot. Appl. 2018, 2, 43–66. [CrossRef]
10. Marini, D.; Belsito, L.; Mancarella, F.; Bonafè, F.; Roncaglia, A. Micro-Opto-Mechanical sensors for tactile
width measurements of surface opening cracks in concrete. In Proceedings of the 2017 IEEE SENSORS,
Glasgow, UK, 29 October–1 November 2017; pp. 1–3.
11. Valença, J.; Puente, I.; Júlio, E.; González-Jorge, H.; Arias-Sánchez, P. Assessment of cracks on
concrete bridges using image processing supported by laser scanning survey. Construct. Build. Mater.
2017, 146, 668–678. [CrossRef]
12. La, H.M.; Gucunski, N.; Kee, S.H.; Nguyen, L.V. Data analysis and visualization for the bridge deck
inspection and evaluation robotic system. Vis. Eng. 2015, 3, 6. [CrossRef]
13. Lim, R.S.; La, H.M.; Sheng, W. A Robotic Crack Inspection and Mapping System for Bridge Deck
Maintenance. IEEE Trans. Autom. Sci. Eng. 2014, 11, 367–378. [CrossRef]
14. Gucunski, N.; Kee, S.H.; La, H.; Basily, B.; Maher, A. Delamination and concrete quality assessment of
concrete bridge decks using a fully autonomous RABIT platform. Struct. Monit. Maint. 2015, 2, 19–34.
[CrossRef]
15. La, H.M.; Lim, R.S.; Basily, B.B.; Gucunski, N.; Yi, J.; Maher, A.; Romero, F.A.; Parvardeh, H.
Mechatronic Systems Design for an Autonomous Robotic System for High-Efficiency Bridge Deck
Inspection and Evaluation. IEEE/ASME Trans. Mechatron. 2013, 18, 1655–1664. [CrossRef]
16. Sutter, B.; Lelevé, A.; Pham, M.T.; Gouin, O.; Jupille, N.; Kuhn, M.; Lulé, P.; Michaud, P.; Rémy, P. A
semi-autonomous mobile robot for bridge inspection. Autom. Construct. 2018, 91, 111–119. [CrossRef]
17. Chan, B.; Guan, H.; Jo, J.; Blumenstein, M. Towards UAV-based bridge inspection systems: A review and
an application perspective. Struct. Monit. Maint. 2015, 2, 283–300. [CrossRef]
18. Seo, J.; Duque, L.; Wacker, J. Drone-enabled bridge inspection methodology and application. Autom.
Construct. 2018, 94, 112–126. [CrossRef]
19. Jimenez-Cano, A.E.; Braga, J.; Heredia, G.; Ollero, A. Aerial manipulator for structure inspection by
contact from the underside. In Proceedings of the 2015 IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS), Hamburg, Germany, 28 September–2 October 2015; pp. 1879–1884.
20. Jimenez-Cano, A.E.; Heredia, G.; Ollero, A. Aerial manipulator with a compliant arm for bridge inspection.
In Proceedings of the 2017 International Conference on Unmanned Aircraft Systems (ICUAS), Miami, FL,
USA, 13–16 June 2017; pp. 1217–1222.
21. AEROBI. Available online: http://aerobi.eu/ (accessed on 11 January 2018).
22. Hayden, J.S. The effect of the ground on helicopter hovering power required. In Proceedings of the AHS
32nd Annual Forum, Washington, DC, USA, 10 May 1976.
23. Sanchez-Cuevas, P.; Heredia, G.; Ollero, A. Characterization of the Aerodynamic Ground Effect and Its
Influence in Multirotor Control. Int. J. Aerosp. Eng. 2017, 2017, 1823056. [CrossRef]
24. Sanchez-Cuevas, P.J.; Heredia, G.; Ollero, A. Multirotor UAS for bridge inspection by contact using the
ceiling effect. In Proceedings of the 2017 International Conference on Unmanned Aircraft Systems
(ICUAS), Miami, FL, USA, 13–16 June 2017; pp. 767–774.
25. Sanchez-Cuevas, P.J.; Heredia, G.; Ollero, A. Experimental Approach to the Aerodynamic Effects
Produced in Multirotors Flying Close to Obstacles. In Proceedings of the ROBOT 2017: Third Iberian
Robotics Conference, Sevilla, Spain, 22–24 November 2017; Ollero, A., Sanfeliu, A., Montano, L., Lau,
N., Cardeira, C., Eds.; Springer International Publishing: Cham, Switzerland, 2018; pp. 742–752.
26. Powers, C.; Mellinger, D.; Kushleyev, A.; Kothmann, B.; Kumar, V. Influence of Aerodynamics and
Proximity Effects in Quadrotor Flight. In Experimental Robotics: The 13th International Symposium on
Experimental Robotics; Desai, J.P., Dudek, G., Khatib, O., Kumar, V., Eds.; Springer International
Publishing: Heidelberg/Berlin, Germany, 2013; pp. 289–302.
27. Rossow, V.J. Effect of Ground and/or Ceiling Planes on Thrust of Rotors in Hover; NASA Ames Research
Center: Moffett Field, CA, USA, 1985.
28. PX4 Pro Open Source Autopilot. Available online: http://px4.io/ (accessed on 11 January 2018).
 17 of 17

29. Chao, H.; Gu, Y.; Napolitano, M. A survey of optical flow techniques for UAV navigation applications. In
Proceedings of the 2013 International Conference on Unmanned Aircraft Systems (ICUAS), Atlanta, GA,
USA, 28–31 May 2013; pp. 710–716.
30. Fuentes-Pacheco, J.; Ruiz-Ascencio, J.; Rendón-Mancha, J.M. Visual simultaneous localization and
mapping: A survey. Artif. Intell. Rev. 2015, 43, 55–81. [CrossRef]
31. Leica Geosystems. Available online: http://leica-geosystems.com/products/total-stations (accessed on 11
January 2018).
32. Video of the Experiments in This Paper. Available online: https://youtu.be/MaQDqkek4FY (accessed on
11 January 2018).