Review of modelling and remote
control for excavators
Hongnian Yu*, Yang Liu, and Mohammad Shahidul Hasan
Faculty of Computing, Engineering and Technology,
Staffordshire University, Stafford, ST18 0AD UK
E-mail: h.yu@staffs.ac.uk
*Corresponding author
Abstract: An excavator is a typical hydraulic heavy-duty human-operated machine
used in general versatile construction operations, such as digging, ground levelling,
carrying loads, dumping loads, and straight traction. However, there are many
tasks, such as hazard environment (nuclear decomposition, earthquake, etc) which
is not suitable human working on site. The remotely controllable excavators are
required to work in such environment. In this paper, we report the current progress
of the on-going project. We investigate modelling and remote control issues of
industry excavators. After reviewing the literature on the related work, architecture
for remotely controllable excavators is proposed. The architecture covers actuators,
modelling, sensors, image signal processing, communication networks, controllers,
task & path planning, human computer interaction, optimal design, co-simulation,
and virtual training environment. The details of modelling, communication and
control of a remotely controllable excavator are provided.
Keywords: Excavator, remote control, mechatronics, modelling.
Reference to this paper should be made as follows: Yu, H., Liu, Y. and Hasan, M. S. (2009)
‘Review of modelling and remote control for excavators’, Int. J. Advanced Mechatronic
Systems, Vol. x, Nos. x/x/x, pp.xx–xx.
Biographical notes: Professor Hongnian Yu has held academic positions at the Universities
of Yanshan, Sussex, Liverpool John Moor, Exeter, Bradford and Staffordshire. He is
currently Professor of Computer Science and the head of the MCDS Research Group at
Staffordshire University. He has extensive research experience in modelling and control of
robots and mechatronics devices, and neural networks, mobile computing, modelling,
scheduling, planning, and simulations of large discrete event dynamic systems, RFID with
applications to manufacturing systems, supply chains, transportation networks and
computer networks. He has published over 140 journal and conference research papers. He
has held several research grants from EPSRC, the Royal Society, and the EU, AWM, as well
as from industry. He is a member of EPSRC Peer Review College. He is a Program Chair of
IEEE Conference on Networking, Sensing and Control in 2007, a General Chair of
International conference on Software Knowledge Information Management and
Applications in 2006, and is serving on various other conferences and academic societies.
He was awarded the F.C. William Premium for his paper on adaptive and robust control of
robot manipulators by the IEE Council in 1997.
Yang Liu received his B.Eng. degree in automation from Hunan University, China, in 2003
and M.Sc. degree in control systems from University of Sheffield, UK, in 2005. He is
currently a Ph.D. student with the Faculty of Computing, Engineering, and Technology,
Staffordshire University, UK. His research interests include control of underactuated
mechanical systems, ground mobile robots, and robotics for medical applications.
Mohammad Shahidul Hasan received his BSc and first MSc in Computer Science from the
University of Dhaka, Bangladesh. He obtained his second MSc in Computer and Network
Engineering from Sheffield Hallam University, UK. He is currently pursuing his PhD at
Staffordshire University, UK in Networked Control Systems over Mobile Ad-hoc Network
(MANET). He has worked as full time lecturer of computer science and engineering in
Bangladesh and is engaged in part time teaching at Staffordshire University. His research
interests include computer networks, networked control systems, remotely controllable
mobile robot systems etc.
1 INTRODUCTION
Earthmoving machines, such as bulldozers, wheel loaders,
excavators, scrapers, and graders are common in construction.
An excavator is a typical hydraulic heavy-duty
human-operated machine used in general versatile
construction operations, such as digging, ground levelling,
carrying loads, dumping loads, and straight traction.
However, there are many tasks, such as hazard environment
(nuclear decomposition, earthquake, etc) which is not suitable
for human to work on site. The remotely controllable
excavators are required to work in such environment. We will
make a brief review from the two aspects: 1) modelling of
excavators and 2) remote control of excavators.
On the modelling aspect, early research work on the dynamic
model of excavators has been done by Vaha et al. (1993).
Based on Vaha et al. (1993), Koivo et al. (1996) did further
studies on the modelling of excavator dynamics during
digging operations. Later on, a number of researchers
investigated the feasibility of autonomous excavation. Many
of these studies have addressed the possible use of an
autonomous excavator (Le et al., 1998; Bradley and Seward,
1998).
Based on the earlier research work, implementation of an
autonomous teleoperated excavator mainly focused on three
parts: modelling, parameter identification, and control
strategy. The key reason for modelling and parameter
identification during the digging operation is to provide online
parameters for the development of an autonomous strategy. In
Tafazoli et al. (1999), an experimental determination
approach of the link parameters and friction coefficients was
developed on the excavator arm. Zweiri et al. (2004)
presented another robust, fast, and simple technique for the
experimental identification of the link parameters and friction
coefficients of a full-scale excavator arm. Furthermore, in
order to carry out autonomous excavation, an online soil
parameter estimation scheme was proposed by Tan et al.
(2005). At the earlier stage of study on excavation, impedance
control was considered as a popular robust control approach
to achieve compliant motion in contact tasks. Details of robust
impedance control for a hydraulic excavator have been
presented by Lu et al. (1995) and Ha et al. (2000). In Tafazoli
et al. (2002), a position-based impedance controller was
presented on various contact experiments by using an
instrumented mini-excavator. Rather than excavation control
strategy, motion and path planning for autonomous excavation
have also been studied in a number of research papers by
Bernold (1993) and Singh (1995). In Saeedi et al. (2005), a
vision-based control system for a tracked excavator was
presented. The system includes several controllers that
collaborate to move the excavator from a starting position to a
goal position. In the paper, both path-tracking accuracy and
slippage control problems have been addressed.
The idea of teleoperated excavator was studied by Parker et al.
(1993), Lawrence et al. (1995), and Kim et al. (2008) based on
the force-feedback control. In a teleoperated excavator
system, if the operator cannot sense the condition of contact,
the work efficiency will decrease compared to a direct control
by the human operator. So, design of the joystick with proper
force feedback can make skilful operators adapt their
operation to the excavating environment based on their
empirical knowledge, and can realise efficient excavation. In
Lawrence et al. (1995), it has proposed the single joystick
endpoint velocity control, which is controlling joystick
stiffness as a function of endpoint force. It was found to be
both a stable and effective form of feedback for a system
where joystick position maps to endpoint velocity. Different
from controlling a real hydraulic excavator, there are many
studies which implement their work on the virtual excavator
including development and evaluation of the controller
(Dimaio et al., 1998), operator training (Tao et al., 2008), and
investigation of remote control issues (Yang et al., 2008).
Apparently, the virtual excavator system is a low-cost, safe,
and reliable system that can both test the system and the
control strategy in virtual environment.
As discussed above, many research studies have focused on
modelling and controller development stages, but few
literature studies the remote operation from a network
communication point of view. Furthermore, it is found that
efficiency of excavation by human operator (Sakaida et al.,
2008) is a notable issue that has potential commercial value.
On the other hand, a teleoperated excavator has always been
desired by industry and manufacturing during the past two
decades. Much of the work on terrestrial excavation has
focused on teleoperation, rather than on the system
requirements for autonomous operation (Ha et al., 2002).
However, although remarkable and valuable progress has
been made on automated excavation, teleoperation of a
full-scale excavator has not been commercially demonstrated.
This paper identifies the issues on designing a remotely
controllable excavator. Section 2 identifies the requirements
of remotely controllable excavators and proposes remote
control architecture of excavators. Section 3 provides the
forward kinematics, inverse kinematics and dynamics of
excavators. Those models will provide the basis for the system
design, development of the controllers, task/path planning,
simulation, validation etc. Section 4 presents several control
schemes for controlling excavators. Some of those control
schemes are based on the authors’ previous work conducted in
robotics context. Section 5 proposes a wireless networked
control scheme for excavators. Finally the conclusions and
future works are given in section 6.
2 REQUIREMENTS OF REMOTELY CONTROLLABLE
EXCAVATORS
Remotely controllable robots or excavators using wired
networks restrict the coverage area and offer very limited
flexibility. On the other hand, wirelessly controlled mobile
robots or excavators provide the freedom from wired
networks and support a higher degree of movement and hence
are preferable to wired versions. Researchers and many
industries are concentrating more and more on such systems
as they are suitable for various applications e.g. nuclear plant
decommissioning, disaster rescue, military operation etc. The
proposed overall system is shown in Figure 1. The excavator
is equipped with the necessary sensors and camera for
gathering data (signal), actuators for moving it and a wireless
communication module to transmit the signals, etc. The sensor
and camera data are transmitted to the control (decision
making) centre through the wireless network which is
composed of multiple mobile robots. The primary
responsibility of these robots is to relay the data to and from
the excavator. These robot nodes can also have a camera
mounted on them to provide additional visual feedback of the
excavator to the control centre. The sensor and camera data
are monitored and analysed at the control and decision making
centre to make the right decision and to send the necessary
action or command to the excavator over the wireless
network.
Staffordshire University with a UK based excavator developer
presented a physical demonstration in an exhibition hosted by
a UK based nuclear decommissioning company, where ten
universities and fourteen companies in robotics made the
presentations. The demonstration system made by the
Staffordshire University team involved a dummy excavator,
an observer robot being controlled over a Mobile Ad-hoc
Network (MANET). This paper presents the system design for
a remotely controllable excavator based on the experience and
the requirements for such applications.
Figure 1: The overall system design.
In order to develop the remotely controllable autonomous
excavators, the following issues and requirements should be
investigated.
2.1 Modeling of Excavators
During the digging operation, it will require not only the
bucket trajectory but also the forces exerted by the bucket on
the soil. Therefore, the modelling of the excavator will involve
(Koivo et al., 1994 and 1996):
1) the kinematics which give the trajectory of the
excavator bucket based on the trajectory of the
excavator arm joints,
2) the inverse kinematics which give the desired joint
variables corresponding to the desired bucket
trajectory,
3) the dynamics which describe the behaviour of the
excavator system,
4) modelling of the interaction between the excavator
bucket and the environment which is necessary for the
remote control during the digging task.
2.2 Sensors and Camera
Remote or autonomous controls for the excavators can
potentially improve the operational safety and efficiency.
Sensors are crucial to this requirement, since feedback signals
are necessary to carry out an unmanned or indirect controlled
task. The sensors used in remote control will include position /
velocity sensors that monitor the joint angles/velocities, force
sensors that detect the interactive force between the excavator
bucket and the environment, and the vibration sensors that
measure the vibration status of excavators. In addition, the
camera is another key sensor which can be used for the
vision-based control system (Saeedi et al., 2000 and 2005).
From the vision information, the operator can better operate
the excavator remotely.
2.3 Actuators
There are a number of nonlinearities affecting the dynamics of
hydraulic actuators, such as the basic flow equation through an
orifice, flow forces on valve spools, and friction (Tafazoli,
1997). To overcome these nonlinear effects, investigation of
the hydraulic actuator is necessary (Tafazoli, et al., 2002).
2.4 Communication systems between excavators and
remote controllers (decision-making)
With the development of high-speed networks capable of
carrying real-time traffic and a network interface with built-in
sensor/actuator, control systems over network have become
an interesting area of research (Cervin, 2003), (Cervin et al.,
2002). Nowadays a low cost and easily deployable remotely
controllable excavator system can be implemented using
IEEE 802.11 standards as shown in Figure 2.
Figure 2: Block diagram of remotely controlled excavator systems.
2.5 Signal and Image signal processing
The data (signal) measured from the sensors and cameras will
be unavoidably contaminated by all sort of noise. To extract
the required valid signal and data for the purposes of control
and 2D/3D virtual view, the certain type of filters and data
processing meads are needed (Fua, 1993; Schmid and
Bauckhage, 1998). The data captured by multiple cameras
mounted on the excavator and the observer robots shown in
Figure 3 can be processed and combined to produce a
complete 3D virtual view of the excavator surroundings
(Shapiro, et al., 1995). However, this process will consume
valuable wireless network bandwidth to transfer video stream
and involves heavy computation of image processing.
2.6 Intelligent control
To achieve the goal of remote control, adaptive and robust
control law is required to compensate for the nonlinear
dynamics of the excavator system. For example, Vossoughi
and Salcudean (2000) used the feedback linearisation
technique, and in Heinrichs et al. (1997), a nonlinear
proportional-integral controller was used. An impedance
controller was used by Tafazoli, et al. (2002) on a teleoperated
excavator. In addition, excavators often conduct respective
tasks; therefore iterative control approaches can be applied.
2.7 Path planning and task planning
Remote control of the excavator in natural environments
requires planning every movement in order to avoid any
obstacle and to locate the machine at each time with respect to
a global coordinate system (Saeedi, et al., 2005). With the
application of an effective path planning, human steering of
the excavator can be removed. Task panning (Singh, 1995) is
to design an operation sequence based on the tasks to be done.
Human operative error can be minimised or completely
removed, and more consistent operation of the machine can be
achieved to increase efficiency.
2.8 Human computer interaction
The system can have two modes of operation: manual and
autonomous. In manual mode, an operator can observe
different views i.e. excavator and observer robot views on
screen and move the excavator manually using a joystick (Kim
et al., 2008; Yang et al., 2008) attached to the control centre
computer. In the autonomous mode, the intelligent control
centre can move the excavator autonomously based on the
sensor and camera data.
2.9 Optimum overall design
The overall design of the excavator is shown in Figure 3. It is
equipped with an adjustable overhead camera, IEEE 802.11
wireless communication module and several electromagnetic
and ultrasonic sensors around it. The overhead camera
produces the operator’s view (Saeedi, et al., 2005). Two
mobile observer robots carrying a remotely adjustable camera
on both sides of the excavator will provide the left and right
views at the controller end.
Figure 3: The overall excavator design.
2.10 Simulation environment: Co-simulation
This paper adopts the co-simulation framework developed in
(Hasan et al., 2009) utilising MATLAB-SIMULINK to model
the plant-controller and OPNET to simulate the network to
accelerate the remotely controllable excavator system
research by producing more realistic simulation results.
2.11 Virtual training environment
Providing training for new operators on actual systems can be
expensive in terms of time and money. A virtual training
environment (Dimaio et al., 1998) shown in Figure 4 can
reduce the cost dramatically. The trainee operator interacts
with the system through the joysticks. The excavator’s
dynamics are simulated on a computer (Makkonen, et al.,
2006).
Figure 4: Virtual training environment.
By summarising the concepts discussed above Figure 5 can be
obtained. This paper will focus on the modelling,
communication and control of a remotely controllable
excavator.
2.1
Excavator
2.2 Sensors
2.2 Camera
2.6
Controller
2.8 Human
Computer Interface
Operator
2.3 Actuators
2.7 Path
Planning
Requirements
2.4 Wireless
Network
2.4 Wireless
Network
2.5 Signal
and Image
Signal
Processing
2.9 Optimum Overall Design
2.10 Simulation Environment: Co-simulation
2.11 Virtual Training Environment
Figure 5: Remote control architecture of excavators.
3 MODELLING OF EXCAVATORS
3.1 Kinematics
The excavator schematic diagram is shown in Figure 6. The
coordinate systems are assigned systematically by applying
the Denavit-Hartenberg convention in Koivo (1994). To
describe the positions of the points on the excavator, the
Cartesian coordinate systems are defined to attach to the links,
which include a fixed Cartesian coordinate system with the
origin on the body of the excavator. It is noticed that the
rotational axis for the first link (i.e. the base) is vertical,
whereas the rotational axes for the other links are horizontal.
The forward kinematics is used to describe the positions and
orientations of the points on the excavator in the Cartesian
coordinate for the given joint positions during the digging
operation. The problem can be summarised as below:
For the given Θ=[θ
2
θ
3
θ
4
]
T
, find the coordinate P=[X Y
Z]
T
=[f
x
(Θ) f
y
(Θ) f
z
(Θ)]
T
.
