Robotic Excavation

293
Hemami, A. (1993), A Study of the Bucket Motion Trajectory for Automatic Scooping in
LHD Loaders, Trans. of the Institution of Mining and Metallurgy, Vol. 102, A1-70, pp
A37-A42
Hemami, A. (1994), Study of Forces in the Scooping Operation of a Mechanical Loader,
Transactions of the CSME (Canadian Society of Mechanical Engineers), Vol. 18, No.
3, pp 191-205.
Hemami, A. (1995), A Fundamental Analysis of Robotic Excavation, J. of Aerospace
Engineering, Vol. 8, No. 4, pp 783-790.
Hemami A. , Seward D.W. and Quayle S. (1999), Some experimental Force Analysis for
Automation of Excavation by a Backhoe., Proc. 16
th
Int Conf. on Automation and
Robotics in Construction, Madrid, Spain, pp 503-508
Hemami, A and F. Hassani (2007), Simulation of the Resistance Forces Bulk Media to
Bucket in a Loading Process, Proc. 24
th
International Symposium on Automation and
Robotics in Construction, ISARC 24, India, pp 163-168
Lee, C.S.G. (1982), "Robot Arm Kinematics, Dynamics, and Control", IEEE Computer, Dec.
1982, pp. 62-80.
Michirev, P.A. (1986), Design of Automated Loading Buckets, Soviet Mining Science, Vol. 22,
No. 4, pp 292-297.
Ostoja-Starzewski, M. and Skibinievski, M. (1989), A Master-Slave Manipulator for
Excavation and Construction Tasks, Robotics and Autonomous Systems, Vol 4, No 4,
pp 333-337.
Sarata, S., Osumi, H., Hirai, Y and Matsushima, G. (2003), Trajectory Arrangement of Bucket
Motion of Wheel Loader, Proc. ISARC 2003, Eindhoven (Netherlands), pp 135-140.
Sarata, S., Weeramhaeng. Y. and T. Tsubouchi (2005), Planning of scooping position and

Controlled Excavation, Proc. 38th Conf. on Remote Syst., Washington D.C., pp 228-
232.
Zelenin, A. N., Balornev, V. I. and Kerov, I. P. (1985), Machines for Moving the Earth, Amerind
Publishing Co., New Delhi.
18
Development of a Semi-Automated Cost-
Effective Facade Cleaning System
Ernesto Gambao
1
, Miguel Hernando
1
and Dragoljub Surdilovic
2

1
Universidad Politécnica de Madrid,
2
Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik,
1
Spain
2
Germany
1. Introduction
Nowadays the number of buildings with large glass or flat façades is increasing all over the
World. These façades must be periodically cleaned with manual procedures that supposed
high cost and risk for the workers that have to develop their work under heavy conditions.
Although the cleaning cost depends a lot on several factors as the façade characteristics, the
cleaning periodicity or the total surface to be cleaned, the average cost is € 8-9 per square
meter. A typical building of 12.000 m2 supposes a total façade cleaning cost of € 100.000 and
this task is usually done every year. The use of an automatic or semi-automatic cleaning


Fig. 1. Automatic Facade Cleaning System for the Vaulted Glass Hall of the Leipzig Trade
Fair ( Fraunhofer FhG ) Fig. 2. SIRIUSC – Automatic Facade Cleaning System (Fraunhofer FhG, Dornier
Technologie)
Table 1 shows the different known robotic façade cleaning systems.
Development of a Semi-Automated Cost-Effective Facade Cleaning System

297
Manufacturer Robot Country Application Kinematics Overcoming
of obstacles
Facade
type
Taisei Exterior Wall Painting
Robot
Japan Coating rail guided No Vertical
Taisei Tile Separation Detection
Robot
Japan Tile inspection Tensed up with cables from roof
to floor
No Vertical
Kumagai Gumi Co. Ltd. KFR-2 Japan Coating Cables, vacuum cups No

Vertical
Shimizu Corporation SB- Multi Coater Japan Coating rail guided No Vertical
Kajima Corporation Tile Separation Detection
Robot
Japan Tile inspection Tensed up with cables from roof

Germany Fa¨ade cleaning wheels, secured by cables No Convex
Comatec - France Fa¨ade cleaning Vacuum cups No Inclined
Robosoft - France Fa¨ade cleaning Rail guided No Horizontal
Robosoft Autonomous Window
Cleaner Robot for High
Buildings (EC:
AUTOWIND)
France Fa¨ade cleaning Rail guided No Vertical
Fraunhofer-Institut IFF,
Dornier Technologie
SIRIUSc Germany Fa¨ade cleaning Rail guided Yes Vertical
Newcastle University;
OCS Group; Cradle
Runways
Arcow UK Fa¨ade cleaning Rail guided No Vertical
CSIC Tito Spain Fa¨ade cleaning Air suction No Vertical Table 1. Façade cleaning robots
In the frame of an European founded project, a consortium formed by several enterprises
and research centres has develop a low cost semi-automated system for the cleaning of
building façades, addressing an innovative concept of system that is able to work in
different types of homogeneous building façades, increasing the productivity, reducing the
risk for workers nearly to zero and contributing to preserve the environment. This system is
with minor changes adaptable to the largest possible number of buildings with
homogeneously-designed façades. Additional constructions to the façade such as guide rails
or scaffoldings are avoided or made unnecessary. The requirements for the control and
sensor concepts are very specific, because the proposed robotic system is able to operate
under adverse conditions such as changing weather conditions.
In this chapter, we present the description of the robotic façade cleaning system

• Cleaning Module (CLM)
• Kinematics Module (KM)
• Carrier Module (CaM)
• Control Module Fig. 4. Arrangement and interconnections of the CAFE hardware modules
The Cleaning Module is in charge of the actual façade cleaning. It mainly consists in a
cleaning mechanism and a positioning system. The most important features of the cleaning
module include:
• Cleaning with brushes and water (environmentally-friendly)
• Water recycling system (low water use)
• All actuators pneumatic (compliant motion, simple control structures, robust)
• Passive degrees of freedom in kinematics to account for unevenness in façade surface
and to protect against hard collision with framework when moving up and down the
façade (braking distance)
• Sensors for detecting glass framework and overseeing the condition of the cleaning
module
The cleaning system is able to clean up to between 3-10mm away from a window pane. The
cleaning Module is shown in Figure 7.
Robotics and Automation in Construction

300

Fig. 5. CAFE Cleaning Module
The carrier is the part of the façade cleaning system that safely holds and provides
horizontal, vertical and transversal motion to the kinematics and cleaning modules. It is
installed on the building rooftop and moves over rails or on a concrete path (guided along
the parapet), holding and providing motion to the cleaning and kinematics modules by
means of cables. While the cleaning robot might be moved from one building to another, the

located in the Common Platform. The Carrier Module controller is located attached to the
carrier on the top of the building. Finally, the operator, located on the ground, uses an
interface device (PC or PDA). So, all the three parts include their own microprocessor-based
computer. The main controller and the carrier controller are based on an embedded PC
equipped with TwinCat-PLC core and Windows CE, allowing the combination of Windows
based programming and PLC programming (IEC 61131-3) reliability. This configuration
reduces the total cost of the system and simplifies the integration.
A wireless connection (Ethernet WIFI 802.11b) is used for the connection between the
Control Module and the operator interface, and between the Control Module and the
carrier. The safety of this communication is critical and it has been guaranteed by a
watchdog system. In case of failure of the wireless communication, all the system adopts a
safety position and can be recovered manually from the Carrier Module Control. The
communication scheme is also shown in Figure 6. Fig. 6. Control System Architecture
Cleaning
module
Kinematics
module
Carrier module
Control module
HMI
(Symbol PDA,
Laptop PC)

CX
-
1000
-

WiFi-Ethernet 802.11b
K-Bus
Ethernet
Access Point
Access Point
HMI
(Carrier Manual
Control)

Robotics and Automation in Construction

302
The cleaning task has been decomposed into different actions that must be performed
simultaneously by the different robot modules. The control module is in charge of the
synchronization of all this tasks.
3.2 Software components
Although from the hardware point of view the robotic systems has three different
microprocessor-based parts, there are five agents working in parallel, corresponding to the
modules described in Figure 4 plus the operator interface:
• Control Module
• Kinematics Module
• Cleaning Module
• Carrier Module
• Operator Interface (HMI)
Additionally, each physical element requires a specific process in charge of establish the
communications between the different elements. The communication virtual bus generation
process is located in the main controller.
The distributed software architecture is shown in Figure 7.
The Server-side is included in the Control Module whereas the Client-side is in the Operator
Interface. Both the Client and the Server are programmed with C# and compiled for the
Microsoft .NET platform (so .NET Compact Framework is required).
The communication is established after accepting the Server a request from the Client. No
hand-shake protocol is implemented. The Server is able to detect both when the connection
is fortuitously cut and when it has not been recently used and reinitiates its state to a new
connection. The Client will receive the data and will only send Operator orders when
produced.
3.3 Human-machine interface operation
There are two possible modes: manual and automatic. Additionally, there are three other
modes: disconnection, emergency stop and error, that depend on the system status and
where the normal cleaning operation is not possible.
In the automatic mode the cleaning task is performed with no need of further information
after the system has been initialized. In the manual mode the operator must indicate the
action to perform that can be accepted of not by the robotic systems depending on the
command availability. The operator can select automatic or manual mode, but the mode
does not effectively change until the main controller confirms it.
When the communication between the interface and the main controller is not properly
established, the disconnection mode is set. In this mode the system is located in a safe
position until the communication is re-established.
When the robot is not able to operate in the normal modes (manual or automatic) it is
immediately set to the error mode and must be recovered manually.
Robotics and Automation in Construction

304
The operator, using the interface, can activate the emergency stop and the robot is stopped
in the next safe position. There are additional emergency buttons at the carrier.
The graphical user interface always shows the emergency stop option to the operator, as
well as the battery status of the PDA device, the manual/automatic mode change and the
WIFI connection status. In the automatic mode it shows the status of the current performed

8. References
Elkmann N., Felsch. T., Sack M., Böhme T. (1999). Modular climbing robot for outdoor
operations, Proceedings of CLAWAR 1999, Second International Conference on Climbing
and Walking Robots, Page 413-419, Portsmouth, U.K.
Elkman, N., Felsch, T., Sack, M., Saez, J. and Horting, J. (2002). Innovative Service Robot
Systems for Façade Cleaning of Difficult-to-Access Areas, Proceedings of the 2002
IEEE/RSJ Intl. Conference on Intelligent Robots and Systems, Lausanne, Switzerland.
Gambao E., Hernando M., Hernández F. and Pinilla, F. (2004). Cost-Effective Robots for
Façade Cleaning, Proceedings of the 2004 Inernational Symposium of Automation and
Robotics in Construction. Jeju, Korea.
Gambao E. and Balaguer C. (2002). Robotics and Automation in Construction, IEEE Robotics
and Automation Magazine. Vol. 9. No 1. (March 2002), ISSN 1070-9932 .
Robotics and Automation in Construction

306
Gambao E. and Hernando M. (2006). Control System for a Semi-automatic Façade Cleaning
Robot , Proceedings of the 2006 Inernational Symposium of Automation and Robotics in
Construction. Tokyo, Japan.
Schraft, R. D., Bräuning, U., Orlowski, T. and Hornemann, M. (2000). Automated Cleaning
of Windows on Standard Façades, Automation in Construction Vol. 9, Issues 5-6,
(September 2000) 489-501, Elsevier, ISSN: 0926-5805.
19
Design and Feasibility Verification of a Knee
Assistive Exoskeleton System
for Construction Workers
SeungNam Yu, SeungHoon Lee, HeeDon Lee and ChangSoo Han
Hanyang University
South Korea
1. Introduction
Robotic-powered exoskeletons and body joint-adapted assistive units are currently under


308
HAL utilizes the EMG signal for its command signal [3]. Moreover it shares external loads
with humans, that is partly assists the human’s loads but it is still requires much patience to
wear and difficult to maintain the quality of EMG signal for every wearing. (a) The Exoskeleton Design Concept Introduced in the Movie Clip (‘Alien II’ and ‘Matrix
Revolution’)

(b) Exoskeleton System for Whole Body Support ('XOS' of SARCOS and 'HAL' of Cyberdyne Inc.)
Fig.1. Developed Exoskeleton Systems to Support the Whole Body
As taken into account in the earlier three cases, the research target for the development of
exoskeletons can fall under the fourth type: the partly assistive muscle power system,
especially the leg assistive system. Many institutions around the world have carried out
research and development on exoskeletons and assistive devices in order to empower or aid
human lower limbs. A well-known system, BLEEX, can partly alternate with the human
muscle power system. This system provides a versatile load transport platform for mission-
critical equipment, so it has several applications without the strain associated with
demanding labor such as that of soldiers, disaster relief workers, fire-fighters, and so on [4].
Northeastern University’s Active Knee Rehabilitation Device (AKROD), Yobotics
Incorporation's RoboKnee, and the NTU-LEE rehabilitation prototype are some of the state-
of-the-art developments in the area of assistive devices to aid the human limb [5, 6, 7].
Design and Feasibility Verification of a Knee Assistive Exoskeleton System
for Construction Workers

309

(a) BLEEX (U.C. Berkeley)


Design and Feasibility Verification of a Knee Assistive Exoskeleton System
for Construction Workers

311
convenience and ease to adapt to humans. These sensors, however, are too complicated to
manufacture or are only verified to perform on a certain part of the human body. The EMG
sensor is one of the most accurate measurement tools to determine human motion intensity.
The approach using this sensor, however, is not considered in this study because of its
inconvenient preparation to assess the signals and its inappropriateness for the working
conditions at a construction site.

In this study, a feasible modular-type exoskeleton system and corresponding sensor systems
are newly proposed to assist construction workers with their lower limb movements. First,
for the purpose of adapting the modular-type exoskeleton system for lower limb assistance
at construction sites, several construction work groups were defined based on specific
boundaries. Second, the design process for the modular-type lower extremity focused on the
knee joint movement will be presented based on the confined boundary. Third, intent signal
processing methods for actuating a proposed system were introduced, and the feasibility of
the command signal was estimated. There were then several measures to quantify the
characteristics of human performance and the exoskeleton platform through an EMG signal
(This sensor is used as a measurement tool of muscle activity only to verify the feasibility of
the proposed system).
2. Analysis for designing the system
2.1 Occupational analysis
In the next step, the research target was brought into the part it would assist. For the sake of
embodiment, we first defined the target task at a usual construction site through a work
pattern analysis, which is strongly related to occupational disorders. Arndt et al. (2008)
conducted a 10-year follow-up research on 14,474 male construction workers. He reported
that musculoskeletal diseases led to an increased proportion of occupational disability [10].


laborers-were classified into four major groups. As shown in step 2 under Fig.6, sheet metal
workers, electricians, laborers, and cement masons were put in charge of each group.
Finally, in the third step, based on the occupational common task of upper groups, target
Design and Feasibility Verification of a Knee Assistive Exoskeleton System
for Construction Workers

313
tasks were selected which included heavy material handling using knee, loaded level
walking, loaded ascent walking, and loaded descent walking.
As earlier mentioned, we developed a modular-type exoskeleton system to assist the lower
limb, and we applied this mechanism in a real construction site. Thus, the target mission to
handle heavy materials and loads at ground level and on a stairway, which is described in
the following images, is critically considered.

Fig. 6. Work group analysis for construction workers
3. Mechanics of muscle activity at the knee
3.1 Extensors of the knee
Rectus femoris functions as an extensor of knee extension, hip flexion, lateral rotation of the
hip, and abduction of the hip [12,13]. Regarding the effect of its weaknesses, direct
measurements of the contribution made by the rectus femoris to knee extension strength are
not available. However, the physiological cross-sectional area of the rectus femoris is
approximately 15% of the total quadriceps femoris muscle mass. Therefore, its negative
Robotics and Automation in Construction

314
effect on a knee is as much as this contribution [14]. Vastus intermedius functions as an
extensor of knee extension and prevents impingement of the pouch in the patellofemoral
joint. It is based on the physiological cross-sectional area range from approximately 15~40%
of the total muscle bulk [14]. Vastus lateralis is a large pinnate muscle, and its uncontested
action is knee extension. The amount of its recruitment is proportional to the amount of

joints that the muscle crosses. To isolate movement at a single joint, the two-joint muscles
cross or they must contract with other muscles. The iliopsoas and the hamstrings, as an
example, together produce isolated knee flexion by canceling each other’s effect at the hip.
Similarly, simultaneous contraction of the gluteus maximus and quadriceps femoris
produces knee extension without hip flexion. However, the knee more frequently displays
simultaneous contraction of the quadriceps and hamstrings. This unusual pattern of
simultaneous contraction of two-joint muscles appears to increase the ability of the knee and
hip to generate the large moments needed during many activities [14].
4. System operation method – trial (1)
4.1 Angular displacement of the knee joint
Following the steps shown in the previous chapter, the final target task was defined more
specifically. We decided to devise a modular-type exoskeleton system for lower limb
assistance, that is, for handling heavy materials during level walks and on stairways. To
gather adequate motivation signals when the construction workers do their jobs at the site,
first, an analysis of knee joint movements was needed. Fundamentally, the muscle activation
status is completely different during level walks and on stairways. Figure.8 and Figure.9
show which parts of the muscle groups are mainly related to knee joint movement during
level walks. Thus, a different type of gait pattern is created for a dissimilar muscle activation
phase. In the case of the knee joint movement, three DOFs with angular rotations are
possible during the level walk.
The primary motion is knee flexion-extension with respect to a mediolateral axis. Knee
internal-external rotation and adduction-abduction (varus-valgus) also occur among healthy
individuals, but with less consistency and amplitude due to their soft tissue and bony
constraints to these motions. The information presented in this chapter was gathered from
the work of Spivak and Zuckerman (1998). The following table shows the range of normal
values of normal adult gaits at a free walking velocity. These values were used as reference
values while we performed the experiments.

Contents Values
Stride or cycle time 1.0 to 1.2 m/sec


(a) Muscle Activity Pattern of Anterior Side of the Leg during Walking and Proposed Sensor
Position ‘1’

(b) Muscle Activity Pattern of Posterior Side of the Leg during Walking and Proposed
Sensor Position '2' and '3' (Position '2' is discarded finally)
Fig. 9. Muscle Activity Pattern of Leg and Proposed Sensor Position for Exoskeleton
The gracilis and sartorius muscles may also contribute to swing-phase knee flexion when
they are activated during late pre-swing, initial swing, and early mid-swing. These muscles,
however, may very well be acting as primary hip flexors during this period [19]. Based on
Fig.8, we analogize that to explain or measure the gait pattern using the muscle activity
pattern, we must consider three positions of the muscle groups.
In this study, however, we propose a method that uses only two muscle sensing groups.
Although this approach is not perfect, it reduced the MSS module in the proposed system
and minimized the loads in the processing system. We decided to disregard the sensor
position (2) because we could explain the muscle activity pattern during the entire cycle
using only (1) and (3). Fig.9 describes the sensor position of the anterior side (1) and the
posterior side (3) of the sensor position we chose. The gray areas represent activation below
20% of the maximum voluntary contraction, and black areas represent activation above 20%
of the maximum voluntary contraction. Muscle activation means Knee Assistive System
(KAS) is inflated at the moment when the foot of the user touches the ground; the