Original Article

Proc IMechE Part C:

J Mechanical Engineering Science
Development and experimental 0(0) 1–16
! IMechE 2020

investigation of an automatic Article reuse guidelines:

sagepub.com/journals-permissions

control system for an excavator DOI: 10.1177/0954406220936308

journals.sagepub.com/home/pic

Guangjun Liu, Lei Yang , Di Wu, Gaoyang Wu and

Hongyuan Chen

Abstract
Excavators are widely used at various construction sites, and they can significantly reduce difficulties during construction
and improve construction efficiency. However, conventional excavators still exhibit many shortcomings. In certain con-
struction environments that require precision, the construction accuracy cannot be guaranteed because excavators
usually operate in harsh environments. Hence, they can cause potential life hazards to the drivers in certain situations. To
address these issues, an excavator automatic control system is developed via retrofitting, modelling, and designing a
control program for an excavator. This system can realise three working modes: manual operation, remote control
operation, and automatic operation. A test prototype based on the aforementioned concept is developed and relevant
experiments are performed. The results under automatic operation mode show that the automatic control system can
move the excavator smoothly along the expected trajectory. The fluctuation in the operation velocity is low, and the
7cm positioning error to the target position is within 7 cm. Thus, the system can realise the desired control accuracy and
satisfy other task design requirements.

Keywords
Excavator, automatic control, retrofit, accuracy test, remote control

Date received: 3 January 2020; accepted: 30 May 2020

Choi1
TVSMC

an intelligent excavator and obtained better experi-

Introduction Gu 2
mental results than that obtained via general PI and PIP
PI PID
With the rapid development of the infrastructure PID control. However, this method was based on a
industry in recent years, excavators are increasingly specific excavator and had no universal applicability. Danko3
more widely used. A conventional excavator exhibits Danko3 proposed an automatic control that could
high construction efficiency. Nevertheless, the oper- potentially aid in improving the trajectory of excava- Tanimoto 4

1 ation process of an excavator mainly involves a syn- tors and buckets in loading excavators. Tanimoto
ergistic effect of the swing, boom, arm, and bucket, et al.4 also proposed a semi-automatic system that
which in turn leads to high requirements from oper- integrates manual and automatic operations to main- 5 PaziewskiOdolinski

ators. Additionally, the hazards involved in a tain the same sense of control as that of a manual Sieradzki6
GPS

construction environment are also an important operation. With respect to the positioning, sensing, RTK
factor that cannot be ignored. During the construction and communication technology of excavators,
of a mountain road, potential safety hazards are posed Odolinski et al.5 and Paziewski and Sieradzki6 exam-
due to bending mountain roads and falling debris, as ined real-time kinematic (RTK) positioning technol-
shown in Figure 1. Thus, the excavators are required to ogy based on GPS, Beidou, and other satellite
accurately place the soil and stones on the designated navigation systems. We examined their research to
position of the rear conveyor belt. However, in a aid us in monitoring real-time location of excavators.
manually operated excavator, it is difficult to ensure
the accuracy of the excavator bucket posture.
Several studies examined the automation of exca- School of Mechanical Engineering, Tongji University, Shanghai, China
vators. Choi1 designed a hydraulic excavator system
Corresponding author:
based on a time-varying sliding mode controller Guangjun Liu, School of Mechanical Engineering, Tongji University,
(TVSMC) fuzzy algorithm, which improved the tran- Shanghai 201804, China.
sient performance. Gu et al.2 applied PIP control to Email: [email protected]
 2 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Target bucket position

Current bucket position

Figure 1. Sketch of an excavator in operation.

Beck 7
In Beck et al.,7 the reliability of an unmanned con-
Mohammed Athraa8 trolling excavator robot system was verified.
Mohammed and Athraa8 examined nonlinear model-
Immonen 9
ling of a fuzzy control design for a 4-DOF (degree of
Mohammad Shahidul
freedom) robotic backhoe excavator. Furthermore,
Hasan10
Immonen et al.9 studied the applicability of a three-
axis tiltmeter in an automatic control system of an
excavator. Mohammad Shahidul Hasan10 investi-
gated modelling and issues with the remote control
of an industry excavator. Hence, these studies pro-
moted the development and maturity of excavator
automation and intelligent technology.
Hence, in this paper, we present an automatic con- Figure 2. System architecture of the electronic control
retrofit.
trol system of an excavator. We developed this system
11
12 by using the computer technology11 and intelligent
CAN
13 sensing technology12 based on previous research.
Furthermore, we used a CAN bus for communica- based on the hydraulic control system of the original
tion.13 By retrofitting the original excavator, we excavator, wherein an electromagnetic proportional
designed a control program with three working control valve group was added to realise electro-
modes of an excavator, namely manual control, hydraulic control of the excavator swing, boom,
remote control, and automatic operation. arm, and bucket. The second part involved installing
Subsequently, we tested the positioning control accur- the control centre and sensors for the system. This
acy of the excavator to ensure that it meets the accur- mainly included the controller, host computer, wire-
acy requirements of the system in automatic less remote control, and angle sensor. The data were
operation. communicated via the CAN bus, and the control
system controlled the proportional solenoid valve
group through pulse width modulation (PWM).
Hardware retrofit
Complete scheme for retrofitting the control system Hydraulic system retrofit
The control system is based on the automation of an The original hydraulic system of type ZX200 excava-
actual hydraulic excavator. Hence, the system should tor adopts a pilot control mode and mainly includes
be installed in the test prototype and verified by rele- two hydraulic control circuits, namely the main con-
vant tests. Therefore, the primary design requirement trol oil circuit and pilot control oil circuit, as shown in
was to determine the entire control scheme of the Figure 3.
system and to build a suitable operating platform Figure 3 shows that the main oil circuit is com-
for the system, which ensures that the automatic con- posed of two pumps and two oil circuits to ensure
trol system can operate entirely on the hardware plat- sufficient power is available for the hydraulic actu-
form. Furthermore, another design requirement ator. The main pump 1 mainly drives the arm cylinder
involved developing an efficient automated control and swing motor, and the main pump 2 mainly drives
system based on the hardware platform. Thus, the the bucket cylinder and boom cylinder. In the case of
automatic control system was expected to ensure a compound action or quick action, the oil circuits of
automatic operation of the excavator as per the task pump 1 and pump 2 pass through the valve group to
requirements and monitor the operating state of the jointly drive the target cylinder. The main oil circuit
excavator in real-time. control valve of the excavator is a three-position six-
In the study, the Hitachi ZAXIS200 hydraulic way valve with a neutral return. The pilot oil circuit
excavator is used as the prototype, and the entire acts on the control valve group through the oil pres-
scheme for the electrical control retrofit is designed, sure, and thereby drives the valve core to move and
as shown in Figure 2. The entire scheme was mainly generate an oil circuit for the hydraulic cylinder or the
divided into two parts, namely the hydraulic system hydraulic motor. Thus, it realises control of the exca-
retrofit and control system retrofit. The first part was vator on each working mechanism.
Liu et al. 3

valve controls the oil circuit in terms of turning on

and off, and the proportional pressure reducing valve
can adjust the oil pressure in the oil circuit. Therefore,
with the combined control, the magnitude and direc-
tion of the movements of the swing, boom, arm, and
bucket are realised.

Control system retrofit

The diagram of the control system retrofit is designed
5
based on the completed hydraulic system retrofit, as
shown in Figure 5. The control system consists of a USB-CAN PC
host computer and controller. The host computer part
consists of an industrial PC, a USB-CAN, and a slave
HFX48m
controller. Furthermore, another part includes a main HFX20m
MURPHY
6
controller, wireless remote controller, and two types CAN

of sensors. The master and slave controllers adopt 433M

HFX48m and HFX20m on-board controllers
(MURPHY Company), as shown in Figure 6. The AI/DI AO/DO

entire control system uses the CAN bus for commu- 12 /24

nication. Additionally, the main controller and wire-

less remote control communicate wirelessly via the
433 M transceiver. The main controller receives the
AI/DI signal of the wireless signal receiver while con-
trolling the solenoid valve group in the hydraulic
Figure 3. Schematic of the original hydraulic system. system through the AO/DO. The power for the
system is supplied by a 12 V/24 V battery.
The core of the automatic control system is com-
Based on the aforementioned analysis, the hydrau- prised of the master controller, whose main function
lic main oil circuit is controlled by the pilot oil circuit. is to process the control command from the host com-
To realise electronic control of the excavator hydrau- puter (sent by the slave controller) and remote con-
lic system, it is necessary to retrofit the pilot oil circuit. troller and to transmit the real-time status data of the
Hence, a modified hydraulic system is developed after excavator to the host computer. Hence, accurate con-
retrofitting the pilot oil circuit, as shown in Figure 4. trol of the excavator’s swing, boom, arm, and bucket
Figure 4 shows that a set of solenoid valves are is realised by acquiring sensor data and sending con-
connected in series between the original pilot valve trol commands to drive the solenoid valve group. The
group and signal control valve to ensure electronic slave controller is responsible for connecting the main
control of the pilot oil can be achieved based on controller and host computer and provides a trajec-
retaining the original operating oil circuit. Hence, tory planning function.14,15 Wireless remote controller
the two-position two-way solenoid valves E21 and realises remote control function of the excavator and
E22 were used to switch the inlet oil circuit. The performs the switching between the three modes. Two
manual pilot valve is connected to the inlet oil circuit types of sensors are used on the excavator and they
by default to ensure that the excavator operates in the include the inclination sensor and electronic compass.
manual mode. When the solenoid valve is energised, Three inclination sensors are installed at the axis of
the manual valve is closed and the automatic control the bucket, arm, and boom (Figure 7(a)) to measure
circuit oil is simultaneously opened. Thus, the exca- the real-time attitude position of the operating device.
vator enters the automatic or remote control mode. The electronic compass is mounted on the top of the
Subsequently, in the automatic or remote control cab. Furthermore, its level is maintained parallel to
mode, the pilot hydraulic oil passes through multiple the axis of the working plane (Figure 7(b)) to measure
sets of proportional pressure reducing valves, two- the real-time slewing angle (a) of the excavator.
position three-way solenoid valves, signal control
valves, and finally reaches the control valve group to
realise electromagnetic control of the hydraulic actu- Modelling and simulation of the
ator. The electric control oil circuit is divided into excavator system
eight branches and each of the branches has a propor-
Coordinate system of the excavator
tional pressure reducing valve and a two-position
three-way solenoid valve in series. Furthermore, two Regardless of the movement of excavator track, the
branches form a group, and thereby leading to a total swing and working (boom, arm, bucket) mechanism
of four groups. The two-position three-way solenoid (jointly known as motion mechanism) of excavator
4 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Figure 4. Diagram of the reformed hydraulic system.

Figure 5. Control system retrofit.

can be simplified to a four-bar linkage with 4-DOF. kinematics modelling. The excavator can be simplified
For a multi-degree-of-freedom open-chain manipula- to a 4-DOF manipulator structure. Thus, the working
tor structure, the coordinate system is usually estab- coordinate system satisfying the D-H method can be
lished via Denavit–Hartenberg (D-H) method to solve established, as shown in Figure 8.
Liu et al. 5

Figure 6. Master and slave controllers.

Figure 7. Schematic showing the sensor installation positions. (a) Inclination sensor. (b) Electronic compass.

The base coordinate system (O0, X0, Y0, Z0) is reference coordinate system. Based on the parameters
located at the centre of the swing mechanism and and variables of the excavator working coordinates
the origin (O0) is located at the chassis of the crawler. system as well as the general transformation relation
In Figure 8, ðOi , Xi , Yi Þði ¼ 1, 2, 3, 4Þ denote the joint of the joint coordinates as per the D-H method, the
coordinates system established at the position of the mapping of the joint space and attitude space can be
swing, boom, arm, and bucket. Furthermore, d1 and l1 obtained in equation (1) as follows
denote the height and width of the boom hinge point
relative to the base coordinates origin (O0), respect- 2 3
x
ively. Additionally, li ði ¼ 2, 3, 4Þ denotes the length of 6y7
the boom, arm, and half of bucket (bucket centre). 6 7
6 7
Moreover, i ði ¼ 1, 2, 3, 4Þ denotes the angle value of 4z5
each mechanism relative to the X-axis of the respect- ’
ive joint coordinate systems, which are set as positive 2 3
cos 1 ðl4 cos ’ þ l3 cosð2 þ 3 Þ þ l2 cos 2 þ l1 Þ
for counterclockwise operations. ’ denotes the angle 6 sin  ðl cos ’ þ l cosð þ  Þ þ l cos  þ l Þ 7
between the bucket and horizontal plane (X0, Y0), 6 1 4 3 2 3 2 2 1 7
¼6 7
which denotes the attitude of the bucket in the base 4 l4 sin ’ þ l3 sinð2 þ 3 Þ þ l2 sin 2 þ d1 5
coordinate system. 2 þ 3 þ 4
ð1Þ
Excavator positive motion equation
The positive motion equation of the excavator is By using the positive motion equation of the exca-
based on the angle of each working joint of the exca- vator, the input bucket centre target coordinates can
vator, including the angle between the boom and hori- be converted into the angle relation for the controlling
zontal plane, to calculate the three-dimensional excavator movement, thus the positioning function of
coordinates and attitude of the bucket relative to the the excavator can be realised.
6 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Figure 8. Working coordinates system of the excavator.

As per the aforementioned set coordinates and par-

Inverse motion equation
ameters, the inverse kinematics equations of the exca-
By establishing the inverse kinematics model of the vator mechanism are shown in equation (6) as follows
excavator system, the precise control of the mechan-
8
ical movement via electrical signals can be realised >
> 1 ¼ tan1 ðy=xÞ
>
> !
based on the model. Firstly, after the three-dimen- >
>  y0  2 2
> x02 02
o þyo þl2 l3
sional attitude ½x, y, z, ’T of the bucket centre relative < 2 ¼ tan1 o0 3 þ cos1
x
3 p ffiffiffiffiffiffiffiffiffiffiffiffi
3
2 02 02
o3 2l2 xo þyo
to the base coordinates system O0 is known, the swing 3 3 ð6Þ
>
>
angle 1 can be calculated in equation (2) as follows >
>
02 02 2 2
1 xo3 þyo3 l2 l3
>
>  3 ¼  cos
>
: 2l l
2 3

1 ¼ tan1 ðy=xÞ ð2Þ 4 ¼ ’  2  3

Next, the base coordinates system of the excavator

is transferred from O0 to O1, and the three-dimen-
sional coordinates of the bucket relative to O1 is
PID control of the excavator system
½x0 , y0 , z0 , ’0 T , and the transformation relation is The excavator system in this study mainly includes
shown in equation (3), where z0 and ’ are constant four subsystems, namely swing, boom, arm, and
bucket, and the control mode of each subsystem is
2 3 2p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
x0 2
x2 þ y 2  l 1 similar. The control mode involves comparing the
6 y0 7 6 z  d1 7 real-time data obtained by the sensor with the presup-
6 7 6 7
6 07¼6 7 ð3Þ posed value and then controlling the hydraulic valve
4z 5 4 0 5
to realise the movement of the mechanism. Therefore,
’0 ’ the control model of each subsystem is first
established, and the transfer functions of hydraulic
The two-dimensional coordinates of the slewing cylinder/swing motor, multi-channel valve, propor-
hinge point of the bucket relative to the new base tional reducing valve, and other components are
coordinates system (O3) are given by equation (4) obtained. Given that each component is connected
( in series to form the entire subsystem, the transfer
x0o3 ¼ x0  l4 cos ’ ¼ l2 cos 2 þ l3 cosð2 þ 3 Þ function of the subsystem is the product of the trans-
y0o3 ¼ y0  l4 sin ’ ¼ l2 sin 2 þ l3 sinð2 þ 3 Þ fer function of each component. Hence, the time-
domain simulation analysis of each subsystem is
ð4Þ
conducted in MATLAB.
As shown in Figure 8, a triangle is formed by the After calculation, the closed-loop transfer functions
boom l2, arm l3, and links of hinge O1O3. By using the of the four subsystems can be expressed as follows
cosine theorem, equation (5) can be obtained as follows 1:894
GBoom ðsÞ ¼ ( )
x02 02 2 2 1:265  10 s þ 9:765  105 s3
6 4
o3 þ yo3  l2  l3
cos 3 ¼  ð5Þ
2l2 l3 þ 0:159s2 þ s þ 1:894
Liu et al. 7

2:054 to control the dynamic response speed of the system,

GArm ðsÞ ¼ ( )
7:262  10 s þ 7:252  105 s3
7 4 I is used to eliminate the static error of the system,
and D affects the dynamic characteristics of the
þ 0:159s2 þ s þ 2:054 system.16 The transfer function for PID control is as
3:78 follows (equation (8))
GBucket ðsÞ ¼ ( )
5:055  10 s þ 5:989  105 s3
7 4
KI
þ 0:159s2 þ s þ 3:78 GPID ðsÞ ¼ KP þ þ KD s ð8Þ
s
0:985
GSwing ðsÞ ¼ ( )
4:689  10 s þ 2:021  104 s3
6 4 To obtain appropriate control parameters and
ensure that the system reaches a stable state quickly,
þ 0:159s2 þ s þ 0:985 it is necessary to conduct a PID control simulation
ð7Þ (Figure 9) of each subsystem to obtain the parameters
under the simulation conditions. Thus, this provides a
The key of a PID control system involves setting reference for the control program.
the three control parameters. Each parameter has dif- The PID control model, shown in Figure 9, is
ferent control effects on the system. Where P is used built in Simulink, as shown in Figure 10. The step

Figure 9. PID control system block diagram.

Figure 10. PID control model. (a) Boom mechanism. (b) Arm mechanism. (c) Bucket mechanism. (d) Swing mechanism.
8 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

response of the excavator mechanism before and 0:0297s2 þ 4:2355s þ 0:00985

GSwingpid ðsÞ ¼ ( )
after adding the PID control algorithm is shown in 4:689  106 s5 þ 2:021  104 s4
Figure 11.
Figure 11 shows that the response speed of each þ 0:159s3 þ s2 þ 0:985s
mechanism is improved, the overshoot is small, and ð9Þ
the system can reach a stable state quickly. By model-
ling the components of the excavator (motion mech-
anism and valve group), the transfer function of the Software design
corresponding components can be obtained. Given
that each subsystem is composed of several parts in
Software framework
series, the transfer function of the subsystem is a Based on the hardware retrofit test prototype,
product of the transfer function of each part. Hence, simulation data, and task demands, the software
by adding the PID control parameters, the transfer platform was designed. The structure of the software
functions of each mechanism after the PID control consists of a three-layer framework involving human–
are as follows computer interaction, automatic control, and
sensing acquisition. The overall framework is shown
0:05682s2 þ 5:3s þ 0:0947 in Figure 12.
GBoompid ðsÞ ¼ ( ) As shown in Figure 12, the excavator software
1:265  106 s5 þ 9:765  105 s4
system is mainly composed of three parts. The
þ 0:159s3 þ s2 þ 1:894s human–computer interaction part is responsible for
0:02054s2 þ 5:135s þ 0:06162 the input and display of information and is located
GArmpid ðsÞ ¼ ( ) at the top of the software system. The automatic con-
7:262  107 s5 þ 7:252  105 s4
trol part is located in the core layer of the system.
þ 0:159s3 þ s2 þ 2:054s Furthermore, the master–slave controller constitutes
0:189s2 þ 6:048s þ 0:1134 a distributed system, which is responsible for the real-
GBucketpid ðsÞ ¼ ( ) isation of core functions such as running automatic
5:055  107 s5 þ 5:989  105 s4 control programs. The sensor acquisition part is
þ 0:159s3 þ s2 þ 3:780s located at the bottom of the software framework
and directly interacts with the excavator mechanical

Figure 11. Step response. (a) Boom mechanism. (b) Arm mechanism. (c) Bucket mechanism. (d) Swing mechanism.
Liu et al. 9

Figure 12. Software framework.

Figure 13. Flow chart of the system main control program.

10 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

mechanism to obtain the most primitive data on exca- The master controller program is the running pro-
vator status. gram of the entire system. The slave controller pro-
gram mainly solves the previous track point and
executes the CAN communication program.
Controller programming
The running process of the main program in a cer-
The controller program is based on the embedded tain task time cycle (Figure 13) mainly includes the
real-time operating system and is divided into pre-processing part and task execution part. The pre-
two parts, namely the master and slave controllers. processing part performs updates on the relevant data

Figure 14. Experimental setup. (a) Test prototype. (b) Industrial PC. (c) Signal cabinet. (d) Control cabinet & wireless receiver.
(e) Solenoid valve. (f) Inclination sensor.

Figure 15. Remote controller.

Liu et al. 11

before the task is executed. The main controller has corresponding control programs. Finally, the execu-
two CAN communication networks, which are pro- tion of the control program is completed and thus a
cessed sequentially. The main controller judges the control cycle is ended.
content of the data based on the received data
and sets the corresponding quantity of the working
state or completes the related data update. Then, Experimental study
the main program enters the second part. It first
Experimental setup
determines whether the task requires an emergency
stop. If everything is normal, then the system is The excavator automatic control test platform inte-
started and the status parameters are evaluated. grated with software and hardware is shown in
Furthermore, it selects from the different working Figure 14.
modes, namely remote control mode, automatic The excavator prototype and various components
mode, or manual mode and executes the are shown in Figure 14. The slave controller is
installed inside the signal cabinet, while the master
controller is installed in the control cabinet. To
Table 1. PID control parameters of the excavator. avoid signal interference, the wireless receiver is
installed outside the control cabinet. The main
Control
power supply for the system is provided by the ori-
parameter Swing Boom Arm Bucket
ginal battery of the excavator, which is installed inside
P 4.3 2.8 2.5 1.6 the signal cabinet. The battery leads outwards to the
I 0.01 0.05 0.03 0.03 control cabinet, sensor, industrial PC, and other
D 0.02 0.03 0.05 0.05 equipment. Thus, the control system can be easily
accessed and stopped inside the cab.

(a) (b)

20 40
Angle(e)
e

15 35

10 30

5 25
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.
Swing(TV) Swing(AV) Boom(TV) Boom(AV)
(c) (d)

-80 -60
-90 1 2 3 4 5 6 7 -70 1 2 3 4 5 6 7
-100 -80
Angle(e)

Angle(e)

-110 -90
-120 -100
-130 -110
-140 -120
-150 -130
Experiment No. Experiment No.
Arm(TV) Arm(AV) Bucket(TV) Bucket(AV)

Figure 16. Comparison of theoretical and actual values of source coordinates (TV: theoretical value; AV: actual value). (a) Swing.
(b) Boom. (c) Arm. (d) Bucket.
12 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

than 50 m. Additionally, the wireless receiver is

Experiment results installed outside the cab (as shown in Figure 14(d))
Remote control test. The remote controller is shown in and is connected to the main controller via the AI/DI
Figure 15. Through switching the operation mode the interface and is powered by the control system.
dial control lever to confirm whether the driving
system and the working system of the excavator can Test for setting PID parameters. In the test for the entire
respond correctly according to the control machine, the PID parameters are first set for the exca-
requirements. vator automatic control system. In the experiment,
As shown in the Figure 15, the remote controller different PID parameters are set to observe the oper-
controls the operation direction and size of four ation of each mechanism of the excavator in the same
mechanisms of the excavator via two handles. task coordinates. With respect to the control param-
Furthermore, the handle involves an XY two-way eters, P, I, and D are adjusted to 1, 0, and 0, respect-
rocker with proportional output. When the emer- ively. After several iterations, the actual PID control
gency stop button is pressed, the entire control parameters for the excavator’s swing, boom, arm, and
system is shut down. The manual, automatic, and bucket are determined. The parameters are listed in
remote control functions are switched via a three- Table 1.
position switch. When the power switch is pressed, a
green light turns on. Furthermore, a blue light turns System control error. To understand the accuracy of the
on when the remote control sends a command. The automatic running program to control the movement
communication frequency of the remote controller is of the excavator in the correct position, the system
433 MHz, and the communication distance is more control error should be measured. In the entire

(a) (b)

0 45
1 2 3 4 5 6 7 43
41
Angle(e)

39
Angle(e)

-5 37
35
33
-10 31
29
27
25
-15
1 2 3 4 5 6 7
Experiment No. Experiment No.
Swing(TV) Swing(AV) Boom(TV) Boom(AV)
(c) (d)

-50 40
-60 1 2 3 4 5 6 7
-70
20
-80
Angle(e)

Angle(e)

-90
-100 0
-110 1 2 3 4 5 6 7
-120
-130 -20
-140
-150 -40
Experiment No. Experiment No.
Arm(TV) Arm(AV) Bucket(TV) Bucket(AV)

Figure 17. Comparison of theoretical and actual values of destination coordinates (TV: theoretical value; AV: actual value).
(a) Swing. (b) Boom. (c) Arm. (d) Bucket.
Liu et al. 13

automatic operation, the source coordinates (starting Positioning accuracy of the system. By using the direct
position) and destination coordinates (ending pos- kinematics correlation equation of the excavator,
ition) are the two most important positioning three-dimensional coordinates of a bucket centre cor-
points. Hence, in the measurement experiment, the responding to the actual angle coordinates can be
points are selected as subjects. Furthermore, seven obtained. A comparison of the theoretical values in
groups of tasks are set, where each group of tasks the task setting with the comparative data of the
runs five times, obtains the value of the actual angle three-dimensional coordinates (X-Y-Z) of source
of the excavator moving to the source and destination and destination points is shown in Figure 19.
points during each operation. The average of the five Furthermore, a comparison of the distance L between
actual angles is considered as the final actual angle. bucket centre and swing centre is shown in Figure 20.
A comparison of the angles is shown in Figures 16 The error analysis is shown in Figure 21. The data
and 17. show that within the operating range of the excavator,
The error diagram is shown in Figure 18. the error between the actual positioning coordinates
Specifically, Figure 18 shows that during the auto- (X-Y-Z) of the bucket centre and theoretical value in
matic operation of the excavator, there is a certain the task setting is within 7 cm. Simultaneously, the
error between the values of the automatic running distance L between the bucket centre and origin of
position angle and theoretical angle of the excavator. swing centre is solved, and the error between theoret-
The swing error is lower than those of other actuators ical value and actual value is less than 5 cm. Based on
due to the longer acceleration/deceleration time. the theoretical basis and further analysis of the test
However, the results show that the error between data, it is determined that the error along the Z-axis
the values of the actual angle and theoretical angle for the bucket centre is higher than those along the
is within 1 . other two-axis. This is because the angle of the

(a) SWING ANGLE ERROR (b) BOOM ANGLE ERROR

Source coordinates Destination coordinates Source coordinates Destination coordinates

ERROR (ABSOLUTE VALUE/°)

ERROR (ABSOLUTE VALUE/°)

0.65

0.54

0.52
0.48

0.47
0.46

0.40
0.37
0.31

0.32
0.28
0.21

0.20

0.21
0.19

0.18
0.17
0.15

0.13

0.13

0.13
0.11

0.11

0.10
0.08

0.05
0.03

0.01

1 2 3 4 5 6 7 1 2 3 4 5 6 7

ARM ANGLE ERROR BUCKET ANGLE ERROR

(c) (d)

Source coordinates Destination coordinates Source coordinates Destination coordinates

ERROR (ABSOLUTE VALUE/°)

ERROR (ABSOLUTE VALUE/°)

0.64

0.63
0.58

0.60

0.51
0.48

0.43
0.42

0.41
0.39
0.39

0.38

0.37
0.35
0.34
0.34

0.32
0.31
0.26

0.24

0.22
0.19
0.17

0.17

0.15

0.15
0.14
0.03

1 2 3 4 5 6 7 1 2 3 4 5 6 7
NO. NO.

Figure 18. Error in the angles. (a) Swing. (b) Boom. (c) Arm. (d) Bucket.
14 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

(a) (b)

6500 1100

Coordinate(mm)

Coordinate(mm)
5500 1050

4500 1000

3500 950

2500 900
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.
X (TV) X (AV) Y (TV) Y (AV)
(c) (d)

2100 6500
Coordinate(mm)

6000

Distance(mm)
2050
5500
2000 5000
4500
1950
4000
1900 3500
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.
Z (TV) Z (AV) L (TV) L (AV)

Figure 19. Comparison of theoretical and actual values of test source point (TV: theoretical value; AV: actual value). (a) X. (b) Y.
(c) Z. (d) L.

(a) (b)
-970
Coordinate(mm)

7000 1 2 3 4 5 6 7
Coordinate(mm)

-980

6000 -990
-1000
5000
-1010

4000 -1020
1 2 3 4 5 6 7 -1030
Experiment No. Experiment No.
X (TV) X (AV) Y (TV) Y (AV)
(c) (d)
2100 8000
Coordinate(mm)

7500
Distance(mm)

2050 7000
6500
2000
6000
1950 5500
5000
1900 4500
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.
Z (TV) Z (AV) L (TV) L (AV)

Figure 20. Comparison of theoretical and actual values of test destination points (TV: theoretical value; AV: actual value). (a) X.
(b) Y. (c) Z. (d) L.
Liu et al. 15

(a) (b)
X-COORDINATE ERROR Y-COORDINATE ERROR
OF BUCKET CENTRE OF BUCKET CENTRE
ERROR (ABSOLUTE VALUE/mm)

ERROR (ABSOLUTE VALUE/mm)

Source coordinates Destination coordinates Source coordinates Destination coordinates

62.44
40.88

39.72
27.30

25.42

23.49

31.54
22.03

24.45
17.70
15.52

20.10
19.98
11.67
11.30

10.59

14.82
13.25

13.20
12.48
6.32

7.88

7.63
6.39
3.06

1.37

1.97

1.02
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.

(c) (d)
Z-COORDINATE ERROR ERROR BETWEEN BUCKET
OF BUCKET CENTRE CENTRE AND SWING CENTRE
ERROR (ABSOLUTE VALUE/mm)
ERROR (ABSOLUTE VALUE/mm)

Source coordinates Destination coordinates Source coordinates Destination coordinates

47.72

42.71
59.12

65.72

39.09

36.78
36.10
35.11
45.60
43.87

43.07
32.48

31.73

15.24
16.76

10.48
12.60

9.23

8.82
5.02
8.76

2.43

1.58
4.22
4.17

4.13

0.08
0.46

1 2 3 4 5 6 7 1 2 3 4 5 6 7
Experiment No. Experiment No.

Figure 21. Error in the distances. (a) X. (b) Y. (c) Z. (d) L.

excavator boom, arm, and bucket mechanism changes Additionally, ideal PID control parameters are
rapidly relative to the horizontal plane when they also determined by conducting experimental
swing in the working plane. tests on the excavator automatic control system.
ii. During the measurement tests of the control
system, the errors in the values between the
Conclusions actual and theoretical angles of swing, boom,
In this study, the hardware for an automatic control arm, and bucket mechanism are within only 1
system is reformed and retrofitted to an existing exca- and the positioning errors are within 7 cm.
vator. Furthermore, kinematics modelling and control Hence, this verifies the accuracy of the automatic
system theory analysis and simulation are conducted control system and satisfies the requirements of
for the controller. A precise system control program the task design.
and software are designed to realise automatic moni-
toring and control of the excavator system. Thus, this In the study, the test prototype utilises typical
system satisfies the need of an unmanned automatic working actuators. Hence, this study has universal
driving system. The following functions are specific- significance to smooth and high-precision automation
ally implemented: operation of various types of hydraulic excavators.
However, improvements are required with respect
i. The speed of the step response of each mechan- to some shortcomings due to the limitation of condi-
ism is improved after adding PID control. tions. For example, the hydraulic system retrofit is not
Furthermore, the overshoot is low, and the thorough enough. This leads to a certain delay in the
system can reach a stable state quickly. control process. This will be further examined in a
16 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

future study. Furthermore, future studies will also 7. Back SJ, Son YK, Kim JH, et al. Reliability demonstra-
focus on the hardware retrofit and optimisation of tion of a robot system for an unmanned controlling
the control program to realise better working tasks. excavator. Appl Mech Mater 2014; 563: 215–218.
8. Mohammed YH and Athraa FS. Nonlinear modeling
and IT2 fuzzy control design of 4DOF robotic backhoe
Declaration of Conflicting Interests
excavator. In: 1st international conference in engineering
The author(s) declared no potential conflicts of interest with researches (ICER), Baghdad, Iraq, March 2017, pp.23–
respect to the research, authorship, and/or publication of 28. New York: IEEE.
this article. 9. Immonen M, Heikkilä R and Makkonen T. Suitability
of a three-axis inclinometer to the automated blade con-
Funding trol system of excavator. Int Symp Autom Robot
The author(s) disclosed receipt of the following financial Construct 2015; 28: 58–64.
support for the research, authorship, and/or publication 10. Yang L, Hasan MS and Yu HN. Modelling and remote
of this article: The authors acknowledge the support of control of an excavator. Int J Automat Comput 2010; 7:
National Key R&D Program of China (Grant No. 349–358.
2016YFC0802900), and the Fundamental Research Funds 11. Zhang J, Jiao S, Liao X, et al. Design of intelligent
for the Central Universities. hydraulic excavator control system based on PID
method. Comput Comput Technol Agricult 2010; 1:
ORCID iDs 222–230.
12. Glynne-Jones P, Tudor MJ, Beeby SP, et al. An elec-
Lei Yang https://orcid.org/0000-0002-6260-5801 tromagnetic vibration-powered generator for intelligent
Gaoyang Wu https://orcid.org/0000-0002-0581-5132 sensor systems. Sensors Actuat 2004; 110: 344–349.
13. Jun Z, Sheng-jie J, Gui-mao S, et al. Design of elec-
References tronic control system of hydraulic excavator with
1. Choi J. Tracking control of hydraulic excavator using CAN bus and PID method. In: 2010 International
time varying sliding mode controller with fuzzy system. Conference on Intelligent System Design and
Adv Sci Lett 2012; 15: 78–82. Engineering Application, Changsha, China, 13–14
2. Gu J, Taylor J and Seward D. Proportional-integral-plus October 2010, pp.570-573. New York: IEEE.
control strategy of an intelligent excavator. Comput Aid 14. Tomatsu T, Nonaka K, Sekiguchi K, et al. Model pre-
Civil Infrastruct Eng 2010; 19: 16–27. dictive trajectory tracking control for hydraulic excava-
3. Danko GL. Loading excavator analysis for trajectory tor on digging operation. In: 2015 IEEE conference on
control improvement. Proc Vol 2013; 16: 134–141. control applications (CCA), Sydney, Australia, 21–23
4. Tanimoto T, Shinohara K and Yoshinada H. Research September 2015. New York: IEEE, , pp.1136–1141.
on effective teleoperation of construction machinery 15. Zhang B, Wang S, Liu Y, et al. Research on trajectory
fusing manual and automatic operation. Robomech J planning and autodig of hydraulic excavator. Math
2017; 1: 14–20. Prob Eng 2017; 2017: 1–10.
5. Odolinski R, Teunissen PJG, et al. Combined BDS, 16. Dwyer AO. Handbook of PI and PID controller tuning
Galileo, QZSS and GPS single-frequency RTK. GPS rules. Automatica 2006; 41: 355–356.
Solut 2015; 19: 151–163.
6. Paziewski J and Sieradzki R. Multi-GNSS high-rate
RTK, PPP and novel direct phase observation processing
method: application to precise dynamic displacements
detection. Adv Sp Res 2017; 29: 1–16.