Automation in Construction 94 (2018) 39–46

Contents lists available at ScienceDirect

Automation in Construction
journal homepage: www.elsevier.com/locate/autcon

The study on the integrated control system for curtain wall building façade T
cleaning robot
Yong-Seok Leea, Sang-Ho Kima, Myeong-Su Gilb, Seung-Hoon Leec, Min-Sung Kangd,
⁎
Sung-Hoon Jange, Bo-Hyun Yue,f, Byung-Gab Ryua, Daehie Hongg, Chang-Soo Hanh,
a
Graduate School of Hanyang University, 222, Wangsimni-ro, Seongdong-gu, Seoul 04763, Republic of Korea
b
ROHAU Corp., 55 Hanyangdeahak-ro, Sangnok-gu, Ansan, Gyeonggi-do 15588, Republic of Korea
c
Dept. of Advanced Mechanical Eng., Korea Polytechnic University, TIP #325, 237, Sangidaehak-ro, Siheung-si, Gyeonggi-do, Republic of Korea
d
Dept. of Interdisciplinary Engineering Systems, Hanyang University, 55 Hanyangdeahak-ro, Sangnok-gu, Ansan, Gyeonggi-do 15588, Republic of Korea
e
Dept. of Mechatronics. Eng., Hanyang University, 55 Hanyangdeahak-ro, Sangnok-gu, Ansan, Gyeonggi-do 15588, Republic of Korea
f
Dept. of Robot Technology, Institute for Advanced Engineering, Young-in, Gyeonggi-do, Republic of Korea
g
Dept. of Mechanical Eng., Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea
h
Dept. of Robot Eng., Hanyang University, 55 Hanyangdeahak-ro, Sangnok-gu, Ansan, Gyeonggi-do 15588, Republic of Korea

A R T I C LE I N FO A B S T R A C T

Keywords: Recently, with a growing number of high-rise buildings in cities, interest in building facade maintenance is
Curtain wall building cleaning robot increasing. The existing method of cleaning the exterior walls of existing high-rise buildings depended on the
Integrated control system methods by workers who used ropes, gondolas, and winch systems. Recently, however, BMU (building main-
Built-in guide rail tenance unit) has been developed and applied to resolve safety problems and boost work efficiency. In Germany,
Building façade cleaning robot
USA, France and other countries, various types of robot systems for building façade maintenance are being
applied. In South Korea, façade cleaning robots attached with curtain walls are also being developed. In this
paper, we propose an integrated control system for the stable control of robots with the building façade cleaning
technology. The proposed control system can be divided into three stages such as preparation stage, cleaning
stage, and return stage. Each independent robot system performs tasks such as cleaning, moving, and obstacle
detection according to each stage. A wireless communication system for stable communication between robots
was proposed and applied for controlling the robot system. The proposed integrated control system was applied
to building façade cleaning robots and its efficiency was verified compared with existing high-rise building
cleaning methods.

1. Introduction maintenance work for high-rise buildings performed in a dangerous

As the population density has increased and the function of the city
has become more complex, the number of high-rise buildings has been
increasing. As the façade of such high-rise buildings can be used for
advertising and PR purposes, they are usually designed to embody di-
verse and unique characteristics in terms of their structure. Moreover,
the interest in building façade maintenance as opposed to building re-
modeling is growing as an effective means to extend the life cycle of
buildings. The conventional building facade maintenance methods for
the cleaning, painting, and crack inspection of high-rise buildings
mainly depend on ropes, gondolas, and the winch system. As these
methods are highly dependent on manpower, in contrast to the other
construction processes that are being automated, their use also incurs
high socioeconomic losses due to accidents. The demand for automated
environment is gradually increasing [1,2,3].
To address the aforementioned demand, many studies on the au-
tomation of the building maintenance unit (BMU) and the related ro-
bots have been conducted in Europe, Japan, and the U.S [4,5].
For example, many research institutes and companies, including
Fraunhofer IPA and Manntech in Germany and Nihon Bisoh in Japan,
introduced robot platforms in the test operation stage [6]. In particular,
Nihon Bisoh has already successfully applied robots to the cleaning
work of high-rise building landmarks in Tokyo, Osaka, and Yokohama
[7,8,9,10]. SIRIUS-c, a BMU developed by Fraunhofer in Germany, is
equipped with two sets of linear translation module in addition to the
wire-based vertical mobility function along the building façade [11,12].
The BMU is designed to perform uninterrupted cleaning while main-
taining a constant distance from the building façade with the help of

⁎
Corresponding author.
E-mail address: cshan@hanyang.ac.kr (C.-S. Han).

https://doi.org/10.1016/j.autcon.2017.12.030
Received 1 May 2017; Received in revised form 15 November 2017; Accepted 19 December 2017
Available online 09 June 2018
0926-5805/ © 2017 Published by Elsevier B.V.
Y.-S. Lee et al. Automation in Construction 94 (2018) 39–46

Table 1
Cleaning façade robot systems and features.
Name Figure Features

Horizontal cleaning ▪Consists of horizontal and

system vertical cleaning systems
[6] ▪Cleaning system moves along the
guide rail

SIRIUSc ▪Rooftop gantry type

[11,12] ▪Power is supplied from the
gantry

SKY Cleaner 3 ▪Supported by a cable from the

[13,14,15] units installed at the top of the
building
▪Movement in X and Y directions
Fig. 1. Work environment.
via pendulum cylinder and
vacuum suckers
TITO 500 [16] ▪Vertical and horizontal The façade of the building to which the built-in guided robot was ap-
movements by using a gondola
plied was covered with two sets of rail, one running vertically and the
▪Contact to the glass by using the
central fan other running horizontally, while the cleaning sections were divided
Robuglass [17] ▪Moves freely by attaching a robot into straight and curved sections depending on the shape of the
onto the glass surface utilizing building.
several adsorption devices
▪Wiper type cleaning tool

GEKKO Façade ▪Moves freely by attaching a robot 2.2. System configuration and roles
Robot onto the glass surface utilizing
[18] several adsorption devices The built-in guided robot that performs cleaning in the work en-
▪Rotary brush type cleaning tool vironment shown in Fig. 1 is composed of an HMR (horizontal-moving
robot), a VMR (vertical-moving robot), a winch module, and a mon-
itoring module as shown in Figs. 2, 3 and 4. The HMR performs
this module. Likewise, Skycleaner 3 in Germany is hooked up with the
cleaning while moving along the rail installed horizontally across the
following unit mounted on top of the building via wires while cleaning
façade whereas the VMR plays the role of transporting the HMR to each
the building, with the cleaning supplies transported from the ground
floor while moving along the rail installed vertically along the façade,
level by a supporting vehicle [13,14,15].
as well as of cleaning the façade. The winch module moves the HMR
Skycleaner 3 moves along the building façade using pneumatic cy-
and VMR to their target floors, respectively, while the monitoring
linders as well as multiple suctions pads installed along the sides of the
module sends the cleaning command, monitors the current position of
cylinders. Tito 500 developed by RATIOFOREM in the United States
the cleaning robot, and transmits the robot status information to the
moves up and down the building façade along the vertical wires while
control staff and various other systems.
cleaning the façade, without bumping onto the glass curtain walls,
The VMR and HMR, which play the key roles in the built-in guided
through the use of a wheel mechanism installed inside the robot, which
robot, are configured as shown in Figs. 2 and 3. The VMR contains four
functions as a damper [16]. RobuGlass, developed by Robosoft in
brake units and brake check sensors designed to prevent safety acci-
France [17] and Gekko Façade, developed by Serbot in Switzerland
dents due to a shift in its weight, as well as a docking rail, which allows
[18] were designed to perform cleaning independently by sucking onto
the HMR to dock in the rail running horizontally along the building.
the building façade only with suction cups, without using wires. Table 1
The two rail extension units connect the building and the docking rail.
shows the characteristics of the various cleaning robots introduced
In Figs. 2 and 3, the part marked ① indicates 24 V and 48 V DC
above.
batteries that supply electricity to the motors and various electronic
In this paper, a built-in guided robot system mounted with cleaning
devices of the VMR and HMR, whereas the part marked ② indicates a
supplies and that cleans the building façade by gliding along the rails
barrel, in which the cleaning agent that is used to clean the outer glass
installed around the building will be introduced, along with the in-
tegrated control system required to carry out uninterrupted and stable
control by considering the robot's cleaning method and cleaning en-
vironment. This study on the integrated control system is aimed at
proposing and verifying an automated control strategy, which has never
been presented in the previous studies conducted on high-rise cleaning
robots, by considering the efficiency and safety of the robot-guided
cleaning works when an automated system was constructed by re-
flecting the inherent mechanical characteristics of the proposed system
and the requirements for cleaning high-rise buildings.

2. System configuration and work environment

2.1. Work environment

The work environment applied to the built-in guided robot system

described in this paper is shown in Fig. 1. The building on which the
robot was tested was a three-story building, with each floor surrounded
by curtain walls, each measuring 1800 × 1320 and 900 × 1320 mm. Fig. 2. Configuration of VMR.

40
Y.-S. Lee et al.
Fig. 3. Configuration of HMR.
Fig. 4. Communication network for robot system.
walls is stored. After spraying the cleaning agent on the glass walls
through their installed nozzles, the VMR and HMR clean the glass walls
with a cleaning tool while moving along the walls.
Contaminants that are produced in buildings or structures can be
roughly divided into two cases. Plant piping and tank surfaces may be
contaminated with oil, various chemicals, soil, or dust. In comparison
to walls, windows are more often.
contaminated with mud or dust, and occasionally with bird excre-
ment. Therefore, when the cleaners clean the outer glass walls, a liquid
cleaning agent is applied to the glass surfaces before using a squeegee.
Built-in guided robots also employ this method when performing the
cleaning task. The vertical cleaning tool mounted on the VMR is used to
clean the curtail walls along rails installed vertically. The HMR is made
up of two units connected with each other via a curb link unit so that it
would be able to move along regardless of whether the section to be
cleaned is a linear or curved section. When the HMR moves along the
horizontal rail of the building, the suspensions and obstacle detecting
sensors is installed to prevent damage by vibration and to detect col-
lisions with obstacles such as opened. A rail cleaning unit, a small
compressor designed to remove the contaminants (fallen leaves, dust,
bird nests) inside the horizontal rails, is also included.
41
Automation in Construction 94 (2018) 39–46
2.3. Configuration of communication network
The built-in guided robot system should be able to return to its
original location safely in the event of system failure due to an external
factor or a system error while performing cleaning work on the building
façade. This is where the communication network functioning smoothly
between the modules kicks in to improve the cleaning performance and
to prevent a safety accident. The built-in guided robot system has a
network configured as shown in Fig. 4. In the built-in guided robot
system, which performs the algorithm at regular intervals while sending
and receiving various commands, the VMR transmits the control com-
mands that it received from the monitoring system to each module as
well as forwards the status information it received from each module
and unit back to the monitoring system. Such network configuration
can prevent the occurrence of a communication error during the car-
rying out of the sequential commands. In the event of communication
failure between the VMR and the monitoring system during cleaning,
the operator shall communicate with each module independently via
RF communication, using the handheld operator system.
3. Integrated control system architecture for the building Façade
cleaning robot
3.1. Integrated control system strategy
To clean a building, the monitoring system, winch system, VMR,
and HMR perform cleaning by either transmitting or receiving the
control command sequentially via communication between the mod-
ules, or in certain cleaning sections, each module performs the given
task independently according to the specific control command pre-
scribed for the section. Fig. 5 shows the structure of the control system
for the built-in guided robot system, including the control flowchart
among the modules in accordance with the cleaning work sequence.
All the modules of the built-in guided robot remain in the standby
mode after the power is turned on, until the communication check is
completed. Upon receiving from the monitoring system (an interface for
the initial operator) the command to start the work on the floor, the
VMR then decides the target speed and direction of the winch by cal-
culating them with the rail alignment function. Once the VMR and HMR
arrive at the target floor, the VMR locks the brake and expands its rails.
When the VMR executes the rail expansion command, the HMR moves
to the horizontal rail along the building. The HMR's arrival at the initial
position located on the horizontal rail of the building marks the com-
pletion of the preparation stage by both the VMR and HMR.
Unlike in the preparation stage, where the tasks are performed se-
quentially via inter-module communication, in the cleaning stage, the
HMR and VMR perform their respective tasks independently. In the case
of the VMR, the rails contract and the brakes are unlocked once the
cleaning work begins. Then the VMR performs cleaning starting from
the highest floor and moving all the way down to the lowest floor, by
spraying a small amount of water onto the curtain walls and then
wiping them dry. In the case of the HMR, it decides its target speed and
direction by calculating them with the motion planning function by
receiving feedback on the robot's current location based on the locali-
zation algorithm. As the HMR performs cleaning along the horizontal
rail surrounding the building, it includes a fault detection function,
which protects the robot from the opened windows or other elements.
Both the VMR and HMR embark on the return maneuver to their
original locations once the cleaning is over. As in the preparation stage,
the VMR locks up the brakes and expands the rails while the HMR
moves back to its initial position and remains in the standby mode.
Once the signal indicating that the VMR has completed the expansion of
the rails is confirmed, the HMR embarks on the dock-in maneuver. Once
the HMR completes the dock-in maneuver, the VMR contracts the rails
and unlocks the brakes before returning to the standby mode. The built-
in guided robot system allows the operator to check the status of each
Y.-S. Lee et al.
Fig. 5. Integrated control system for built-in guided robot.
module by sending the status information to the monitoring system
once every second throughout the three stages. In the operator mode,
the aforementioned sequential commands can be transmitted by the
operator in real time to ensure the smooth return of the robot to the
initial location in the event of work failure.
3.2. Rail alignment control
For the VMR to be able to clean the areas between the floors, it
should first move to the inter-floor rails that connect the floors. The rail
alignment algorithm controls the precise location and speed of the VMR
so that the VMR could dock into the horizontal rail surrounding the
building. First, the two laser sensors (Sharp EX-L211) are mounted on
the upper/lower half of the VMR's rail extension module, as shown in
Fig. 6, to perform the algorithm and to detect the space and building
wall inside the horizontal rail with ON/OFF signals. Assuming the DIO
signals from the laser sensor mounted on the upper section as S1 and
DIO signals from the sensor mounted on the lower section as S2, the rail
alignment algorithm attempts to align the VMR rails with the building's
horizontal rails while moving slowly from the lower floors to the upper
floors to prevent failure due to the weights of the HMR and VMR. As
Fig. 6. Case study between VMR and horizontal rail at test building.
42
Automation in Construction 94 (2018) 39–46
mentioned earlier, the VMR moves from the lower floors to the upper
floors, and if S1 and S2 detect the inner cavities of the horizontal rails of
the building one by one while doing so, the HMR defines the rails as
aligned. The rail alignment algorithm performs the aforementioned
definition in accordance with the following five modes, depending on
the locations of the building's horizontal rails and the VMR's rails. Fig. 7
shows the flow chart for rail alignment control. Fig. 8 present the flow
chart in Mode 1 case. Mode 1 signifies an occasion where S2 is ON and
S1 is OFF, and the VMR glides down slowly until S2 is turned OFF. If S1
is turned ON, the VMR starts to glide up again slowly while the algo-
rithm function stops when S2 is turned OFF. Mode 2 signifies an oc-
casion where S1 is ON and S2 is OFF, and as in Mode 1, the VMR
performs its algorithm while slowly gliding up the building.
In Mode 3, 4, and 5, both S1 and S2 are OFF, and the process of
detecting the rails is included in the algorithm. The search process in
the rail alignment algorithm signifies the process where S1 or S2 detects
the inner cavities of the rails while the VMR is gliding up to the upper
floors. Mode 3 signifies an algorithm where S1 detects the inner cavities
of the building rails, which is followed by the performance of the task in
Mode 2. In Mode 4, S2 detects the inner cavities of the building rails,
which is followed by the performance of the task in Mode 1. In Mode 5,
the robot glides up the building for 15 s considering its inter-floor
distance and VMR's speed before it glides down again when the S1 and
S2 sensors are OFF, and then performs the task in Mode 1.
3.3. Fault detection control
An ultrasonic sensor was used to perceive an obstacle like the
opening of the external wall ventilation window. To prevent the
building external wall from being perceived as an obstacle, a vibration
resistant and waterproof ultrasonic wave sensor was used one with a
narrow ultrasonic-wave directional angle and for industrial use (Fig. 9).
The control system transmits a trigger signal and sends an ultrasonic
wave from the sensor behind 20 msec to transmit an analog signal by
ultrasonic-wave strength. The control system was designed to judge a
signal larger than a specific size as an obstacle, and to determine the
location of an obstacle by the reflection time of the ultrasonic wave.
Y.-S. Lee et al. Automation in Construction 94 (2018) 39–46

Start

Get Sensor
Data

Sensor2 Check No
the empty
space?

Yes Sensor1 Check No

the empty
Mode 1 space?
Yes VMR UP
(high speed)
Mode 2
Get Sensor Fig. 9. Ultrasonic sensor module for obstacle detecting.
Data
3.4. Experiment of proposed system

Sensor1 Check No In order to verify the proposed integrated control system for
the empty building façade cleaning robots, it was applied to robots in a test bed
space?
equipped with a curtain wall with horizontal and vertical rails.
Yes No
Sensor2 Check Experiments were carried out according to preparing step, cleaning
the empty step, and returning step, and each step was run continuously to verify
Mode 3
space?
the performance of the integrated control system.
Yes No Fig. 10 shows the results of applying the integrated control to a
Over 15sec?
building façade cleaning robot. According to the integrated control
Mode 4
Yes system procedure, the preparing step is first performed in advance. The
VMR moves to the target floor after receiving the movement command
VMR Stop
from the integrated control monitoring system. When the VMR ap-
proaches the target floor, the rail alignment function is performed for
Mode 5 horizontal movement of the HMR. After the rail alignment function is
Fig. 7. Flow chart for rail alignment control. performed, the VMR operates the brake module to fix the robot in the
vertical rail and extracts the horizontal rail for the HMR to move on the
horizontal rail. During the cleaning process, the HMR places the blade
Mode 1 of the wiper module on the glass façade to clean the glass walls. The
Start water jet module attached to the HMR injects a tiny amount of water
onto the glass surface. The HMR then moves along the horizontal rails
while the blades, which are attached to the glass, remove the water on
VMR DOWN VMR UP
the glass by squeezing it. In addition, the HMR sensor detects obstacles
(slow speed) (slow speed)
(e.g., windshields) and stops the robot. The cleaning process is also
performed along the straight and curved areas.
The VMR contracts the horizontal rail and releases the brake
Get Sensor Get Sensor
module. Meanwhile, the robot climbs along the vertical rail in order to
Data Data clean the window façade. The wiper module in the VMR brings the
blade into close contact with the glass wall, and the water jet module
injects a tiny amount of water onto the glass surface. The system cleans
the vertical window façade while descending along the vertical rail.
Sensor1 Check No Sensor2 Check No In the returning step, the VMR climbs the vertical rail up to the floor
the empty space? the empty space? where the HMR is stationed in order to perform the rail alignment
function. When the robot and the horizontal rail are accurately aligned
with each other, the brake module for locking the robot onto the ver-
Yes Yes tical rail is activated while the horizontal rail is extended in order to
facilitate the movement of the HMR robot. When the VMR is ready for
VMR Stop docking, the HMR docks with the VMR in order to form a single con-
VMR Stop
joined robot. When the docking procedure of the robots has been
completed, the VMR contracts the rails and releases the brakes while
waiting for the command to clean the other side of the building's glass
Alignment façade. From the results shown in Fig. 10, it can be confirmed that the
Complete glass façade cleaning robots, which were applied with the proposed
integrated control system, have successfully performed stable control
maneuvers that are pre-requisites for robot cleaning. In the cleaning
Alignment
step, the wiper module operation processing is performed to conduct
Control End
façade cleaning. The waterjet module sprays the water on the curtain
Fig. 8. Flow chart for Mode 1 case study. wall glass and simultaneously, the water is squeezed by wiper module
according to move the HMR. As shown in Fig. 10, if the ultrasonic

43
Y.-S. Lee et al. Automation in Construction 94 (2018) 39–46

Fig. 10. Experiment results of the integrated control system.

sensor detects obstacles while moving, the fault detection will be per-
formed, alerting the stationary user. The cleaning step is also performed
in the straight line area and the curved line area. In the returning step,
the HMR returns to its initial position to board the VMR. If the HMR
arrives at its initial position, it will conduct the dock-in process, and the
VMR and HMR will become one robot. Also, the VMR will clear its
brake, contract the rail and wait for the command to move to a different
floor. From the results of Fig. 10, it was confirmed that the proposed
integrated control system was applied in controlling the curtain wall
building façade cleaning robot.

Fig. 11. Cleaning process by labor. 4. Efficiency analysis of the building Façade cleaning robot

4.1. Construction of test environment

To prove the effectiveness of the built-in guide robot, the existing

44
Y.-S. Lee et al.
cleaning method which depends on the worker and the new cleaning
method by the proposed robot system were compared and analyzed. To
perform the experiment, a 3-story mockup building made of curtain
walls equipped with the built-in guide rail was used, as shown in
Fig. 11.
Using the existing method which uses workers, the cleaning time
and water consumption amount were measured. The cleaning time in-
cluded the workers' preparation time for cleaning equipment, for
binding ropes tightly, and for installing work chairs, as shown in
Fig. 11, as well as the net glass cleaning time, and the workers' resting
time during work. The water consumption amount was measured by
comparing it between before and after work.
A total of five workers were assigned to the equally divided sections
of the building and conducted cleaning simultaneously as they climbed
down from a top the building.
The cleaning robot system is designed for the external wall glass of
10-story or higher high-rise buildings. Thus, with this assumption,
water replacement time during cleaning was regarded as 1 cycle to
measure the water consumption amount and the required time for
cleaning was also measured.
4.2. Comparative analysis of efficiency between the building Façade
cleaning robot and worker
The graph in Fig. 12 shows the required cleaning time and an
average one-cycle time of exchanging water when five workers si-
multaneously cleaned the exterior wall glass of a 10-story building in
each equally divided section. Generally, the 1st cycle time was the
longest, but the 2nd ~6th cycle time had no big deviation. In cleaning
the 10-story building, the water is changed about six times about once
every 20 min by worker.
Fig. 13 shows the water consumption amount by worker vs. the
water consumption amount by the developed robot system in each
cycle. In the cleaning of the 10-story building, the water consumption
amount of the robot was compared with that of the worker, based on
the worker's cleaning area per cycle. The test building had linear and
curved sections, so the robot movement speed varied accordingly. The
robot's water spray amount was controlled to change according to its
movement speed, and thus its water consumption amount for each cycle
was measured as varying.
Table 2 showed the cleaning efficiency comparison between the
workers and the robot. In covering a total cleaning area of 626.4 m2 in
the 10-story building, 5 workers took 2 h, while the robot took 1.3 h,
Fig. 12. Cleaning time by five workers and average cleaning time.
45
Automation in Construction 94 (2018) 39–46
Fig. 13. Water consumption comparison between human labor and façade
cleaning robot system.
Table 2
Cleaning performance by human and façade cleaning robot system with in-
tegrated control.
Cleaning efficiency Man power Robot system Remarks
Execution Time 2h 1.3 h 35% less time
(Façade cleaning, 10-story (5 workers)
building)
Consumption 27.5 L/h 10 L/h 64% less utility
(water, detergent, etc)
Cleaning performance 2
156.6 m /h 2
234.9 m /h 50% more area
per hour
and the water consumption amount for workers was measured as 27.5 L
and that for the robot was measured as 10 L, per hour. The robot also
covered the cleaning area more than 1.5 times than the workers.
5. Conclusion
The study introduced the façade cleaning system for buildings
equipped with the built-in guide rail-applied curtain walls, and pro-
posed the integrated control system for cleaning motion. It further in-
troduced the façade cleaning robot system which was composed of VMR
responsible for vertical movement and cleaning, HMR responsible for
horizontal movement and cleaning, the winch system responsible for
VMR movement, and the monitoring system responsible for the robot's
cleaning and its operation status. To ensure the integrated control of
this system, the system was connected via two wireless communication
systems, and each robot system was incorporated into single integrated
façade cleaning robot system by sending diverse commands. To conduct
the robot's vertical and horizontal movement, we proposed the in-
tegrated control system architecture, applied it to the robot system, and
examined it on the test bed, proving the integrated control technology
for the robot system. To prove the efficiency of the façade cleaning
robot, the manual cleaning by human labor and the automatic cleaning
by the façade cleaning robot system that is applied integration control
system were compared in terms of work time and water consumption
amount. The analysis revealed: compared to the human labor, the robot
cut the work time by about 35% and the water consumption by 64%,
and it improved the cleaning performance by about 50%. This confirms
that the façade cleaning robot's cleaning performance is superior to
human labor in terms of consumption time, water consumption, and
Y.-S. Lee et al.
safety.
Acknowledgement
This research was supported by the Industrial Strategic technology
development program (No.10052965) funded by the Ministry of Trade,
Industry and Energy (MOTIE), KOREA and was supported by a grant
(14SCIP-B079344-01) from Smart Civil Infrastructure Research
Program funded by Ministry of Land, Infrastructure and Transport
(MOLIT) of Korea government and Korea Agency for Infrastructure
Technology Advancement(KAIA).
References
[1] R.M.W. Horner, M.A. El-Haram, A.K. Munns, Building maintenance strategy: a new
management approach, J. Qual. Maint. Eng. 3 (4) (1997) 273–280.
[2] A. Spedding, Building Maintenance Economics and Management, Routledge, 5
Howick Place, London, (1987).
[3] A.H. Christer, Modeling inspection policies for building maintenance, J. Oper. Res.
Soc. 33 (8) (1982) 723–732.
[4] S.M. Moon, J.M. Huh, S.H. Lee, S.P. Kang, C.S. Han, D.H. Hong, A survey on robot
systems for high-rise building wall maintenance, J. Korean Soc. Precis. Eng. 30 (4)
(2013) 359–367.
[5] R.D. Schraft, U. Bräuning, T. Orlowski, M. Hornemann, Automated cleaning of
windows on standard facades, Autom. Constr. 9 (5) (2000) 489–501.
[6] Manntech, Façade Cleaning Systems, On-line: http://www.manntech.de/facade-
46
Automation in Construction 94 (2018) 39–46
cleaning.html , Accessed date: 30 April 2017.
[7] Sugiyama, Y. Device for Horizontal Moving an Apparatus Working at a High
Location. Nihon Biso Co., Ltd., US Patent, No. 4811819, 1989.
[8] Yokota, T., Nishiyama, Y., and Takasaki, T. Horizontal Moving Automatic Outer
Surface Cleaning. Nihon Biso Co., Ltd., US Patent, No. 4809384, 1989.
[9] Mimura, N. Window-Pane Cleaning Device. Nihon Biso Co., Ltd., US Patent, No.
5020180, 1991.
[10] Kitahara, G. Device for Cleaning a Window Glass. Nihon Biso Co., Ltd., US Patent,
No. 5086533, 1992.
[11] N. Elkmann, D. Kunst, T. Krueger, M. Lucke, T. Bohme, T. Felsch, T. Stürze,
SIRIUSc—Façade cleaning robot for a high-rise building in Munich, Germany,
Proceedings of the 7th International Conference Climbing and Walking Robots,
2004, pp. 1033–1040 (Madrid, Spain).
[12] N. Elkmann, M. Lucke, T. Krueger, D. Kunst, T. Stürze, J. Hortig, Kinematics, sen-
sors and control of the fully automated facade-cleaning robot SIRIUSc for the
Fraunhofer headquarters building, Munich, Ind. Robot. 35 (3) (2008) 224–227.
[13] H. Zhang, J. Zhang, R. Liu, W. Wang, G. Zong, Pneumatic climbing robots for glass
wall cleaning, Proceedings of the 7th International Conference on Climbing and
Walking Robots, 2004, pp. 1061–1069 (Madrid, Spain).
[14] H. Zhang, J. Zhang, G. Zong, Realization of a service climbing robot for glass-wall
cleaning, Proceedings of the IEEE International Conference on Robotics and
Biomimetics, 2004, pp. 395–400.
[15] H. Zhang, J. Zhang, W. Wang, R. Liu, G.A. Zong, Series of pneumatic glass-wall
cleaning robots for high-rise buildings, Ind. Robot. 34 (2) (2007) 150–160.
[16] T. Akinfiev, M. Armada, S. Nabulsi, Climbing cleaning robot for vertical surfaces,
Ind. Robot. 36 (4) (2009) 352–357.
[17] Robosoft robuGLASS, On-line: http://www.robosoft.com/robotic-solutions/
cleanliness/glass-roof.html , Accessed date: 1 May 2016.
[18] A.G. Serbot, Gekko Façade, On-line: http://www.serbot.ch/en/products/facade-
cleaning/gekko-facade , Accessed date: 1 July 2016.