International Journal of Mechanical Engineering and Robotics Research Vol. 9, No.

1, January 2020

Robust Adaptive Controller Design for Excavator

Arm
Nga Thi-Thuy Vu
School of Electrical Engineering, Hanoi University of Science and Technology
Email: nga.vuthithuy@hust.edu.vn

Abstract— Working in the dynamic environment makes the About the control design, during the earlier stage of
operation of excavator arm be affected by many factors. study on excavator, impedance control is considered as a
This leads to the unexpected operation of the automated popular control. In [11], a position-based impedance
arm. In order to overcome this problem, in this paper a controller is presented on mini-excavator. In [12]-[13],
robust adaptive controller that does not depend on the
authors present detail of robust impedance control for
system parameters is presented for an excavator arm.
Firstly, the model of the excavator arm is demonstrated in hydraulic excavator. The impedance control suits to apply
the Euler-Lagrange form considering with overall excavator for excavator because it can deal with both free and
system. Next, a robust adaptive controller has been constrain motion [13]. However, the algorithm is quite
constructed from information of state error to calculate the complicated. Recently, many modern control techniques
unknown factors. In this paper, the stability of overall are used in trajectory control of excavator arm. In [14], an
system as well as the convergence of the adaptive adaptive controller is presented in controlling excavator
components is mathematically proven by using Lyapunov arm. The stability of system is ensured through
stability theory. Also, the proposed controller is model free mathematical proof and verified by simulation results.
then the closed loop system is not affected by disturbances
However, the simulation as well as the explanation of
and uncertainties. Finally, the simulation is executed in
Matlab/Simulink to compare the performances of the simulation results is quite poor. In [15]-[16], the fuzzy
presented scheme and the PD controller under some controllers are employed to solve the tracking control
conditions (i. e. no load, full load, rated parameters, varying problem of the excavator. In these type of controllers, the
parameters). The simulation results show that the proposed information about the system does not require. However,
algorithm given the better performances (fast response, the stability of the overall system is not shown in the
small steady state error) for all cases. mathematic. Nowadays, the PID controller still being
used widely in the practice because of its simplify.
Index Terms— adaptive, excavator, Euler-Lagrange system, However, tuning of the PID gains to adapt with the
robust control
change of working conditions is difficult and depending
on the personal experiences. In recent times, in order to
deal with this problem, some optimization techniques
I. INTRODUCTION
such as artificial neural network (ANN), ant colony
Nowadays, robots are used commently in the industries optimization (ACO), etc., have been applied to optimize
because of their versatility and efficiency [1]-[3]. The the PID parameters. In [17], an genetic algorithm (GA) is
automatic remote control of the robot plays a significant used to determine the PID gains for trajectory tracking
role in real-life application, such as nuclear field, control of robotic excavator. The presented scheme gave
construction, rescue missions. For the excavator robot, in the good performances in comparison with some given
order to perform a specific duty, it needs to complete at methods. However, the gotten results still have problems
least two tasks: determining a feasible path from its initial needed to be discussed.
location to the destination and then executing the task In this paper, a robust adaptive controller is proposed
through control algorithm that has to be designed. for excavator arm system. The structure of controller
According to these requirements, the tracking control consists of two parts: the first part is responsible for
problem for the excavator robot system is constantly keeping the stability of the system and the second part is
receiving the interesting of scientists. used for adapting with the unknown parameters.
The earlier research work mainly focused on modelling Therefore, the proposed controller has ability to cancel
work including kinematic and dynamic model, modelling the effect of the uncertainties as well as to keep the
of interaction between the machine and the environment, tracking error go to zero. Also, the presented controller is
and parameter identification [4]-[10]. Modelling and simple so it is easy to implement. The feasible of the
parameter identification during the operation of machine algorithms is demonstrated by Lyapunov stability theory
is very helpful for the real-time monitoring and remote and verified through simulation models. The simulation is
control. executed in Matlab/Simulink for both presented scheme
and the PD controller under some conditions to ensure
that the proposed algorithm given the good performances
Manuscript received June 4, 2018; revised September 19, 2019. for all cases.

© 2020 Int. J. Mech. Eng. Rob. Res 42

doi: 10.18178/ijmerr.9.1.42-47
 International Journal of Mechanical Engineering and Robotics Research Vol. 9, No. 1, January 2020

II. CONTROLLER DESIGN FOR EXCAVATOR ARM to move the entire system on the x0O0y0, plane and the
arm is used for movement in the z0O0y0 and z0O0x0 planes.
A. Modelling of Excavator Arm This paper concern mainly on the motion control of
Consider an excavator system with structure as Fig. 1. excavator arm, so the base part and the rotation around
The system consists of two subsystems: the base is used O0z0 axis are consider unchanged.

Figure 1. Schematic diagram of an excavator.

The Euler-Lagrange model of excavator arm during the D23  D32  D33  M bu a3r4 cos  4   4 
digging process corresponding to the coordinates of each
D13  D31  D23  M bu a2 r4 cos 34   4 
join as shown in Fig. 1 is as follow [18]:
D12  D21  D13  I st  M st  r32  a2 r3 cos 3   3  
D( )  C ( , )  G( )  B( )    FL (1)
 M bu  a32  a2 a3c3  a3r4 cos  4   4  
where   2 3 4  is the position of each joint in
the joint space, D() represents inertial part, C ( ,  ) is C11   M st a2 r3 23 sin 3   3   M bu a2 a3 23 s3

the Coriolis and centripetal effects, G() is the gravity  M bu a2 r4 234 sin 34   4 
C12   M st a2 r3 23 sin 3   3   M bu a2 a3 23 s3
part, B( ) represents frictions;  is the corresponding
 M bu a2 r4 234 sin 34   4 
input matrix,   1  2  3  specifies the torques
C13   M bu a2 r4 234 sin 34   4 
acting on the shaft of 3 joints, FL represents the
interactive torques. The formulas of above parts are given C21  a2 2  M bu a3 s3  M st r3 sin 3   3  
 M bu a3r4 234 sin  4   4 
by the following expressions:

 D 11 D 12 D 13   C 11 C 12 C 13  C22   M bu a3r4 234 sin  4   4 

D( )   D 21 D 22 D 23  , C ( , )  C 21 C 22 C 23  ,
  C23   M bu a3r4 234 sin  4   4 
 D 31 D 32 D 33  C 31 C 32 C 33 
1 1 0 
C31  M bu r4 2  a2 sin 34   4   a3 sin  4   4  
  0 1 1 ,
0 0 1   M bu a3r43 sin  4   4 
C32  M bu a3r4 sin  4   4 
G( )  G2 G3 G4  , B( )   B bo2 B st3 B bu4  , C33  0
G2   M bu  M st  ga2c2  M bo gr2 cos  2   2 
where:
G3  M bu ga3c23  M st gr3 cos  23   3 
D33  I bu  M bu r42
G4  M bu gr4 cos  234   4 
D22  D33  I st  M st r32  M bu  a32  2a3r4 cos  4   4  
D11  D22  I bo  M bo r22  M st  a22  2a2 r3 cos 3   3   B. Robust Adaptive Controller Design
Define the error signal:
 M bu  a22  2a2 a3c3  2 a 2 r4 cos 34   4  
e = d -  (2)

© 2020 Int. J. Mech. Eng. Rob. Res 43

 International Journal of Mechanical Engineering and Robotics Research Vol. 9, No. 1, January 2020

where d is desired value of . The filtered error surface  

D   x  C  , x  M   y  C  , y  
is chossen as
 G    B   FL     (4)
x    d  e
(3)  
 C  , x  Tmm  
x    d  e where
y  d  e ,
where   diag (1 , 2 , 3 )  0 .
 m   D   C  ,     G   FL  ,
T
B 
Based on (3), (1) can be written as
m   y y 1 1 1
T

Asuming that

   
T
m   D   G   FL    1
T
C  , B  2  3  ? 4  5 6 
 
And i (i = 1,2,3,4,6) are unkown positive constans. Theorem: If there exist the scalars i (i = 1,2,3,4,6) so
Consider the following theorem. that

Then the following controller and adaptation law can 6

x ˆ k k2
make the dynamic error go to zero. V   xT Kx  xT Tm m  xT 
k 1 x k   k
x  k
2

6 2 6
6
x ˆ kk 2    k  k ˆ k
k

   Kx   k 1 x  k   k k 1  k
k 1 x k   k
x  k k2
2
6 6

˙ (5)   xT Kx  x Tm m     k  k ˆ k
x  k   k k 1  k
 x k 2
2
k 1

ˆ k  k  k' ˆ k , k  1,, 6 x  k k2
2
 (8)
x k   k
6 6
  xT Kx  x T      k  k ˆ k
k 1 x  k   k k 1  k
6 6
 k
where K = diag(k1, k2, k3) with ki > 0 (i = 1, 2, 3),   xT Kx   k  k    k ˆ k
  0,  k  0 are constant,  = diag (1, 2, 3) > 0, k 1  k
'
i 1
k
6
k 2 6   2
  xT Kx    k  (  k  k  k k )
T
        1 1
k 1 2 k i 1 2 k
Proof: Choose the Lyapunove function: 6
 k 2
 min  K  x   k  
2

6 k 1 2 k
1 T 1 2
V x D   x   k (6)   V  
2 k 1 2k
where
  min[min  K  ,k' / 2k ]/ max[ D    ,1/ k ]  0
where k  k  ˆ k , ˆ k are estimated values of k,
k is positive constant.    1 2 3 4 5 6  ,?
T

The time derivative of Lyapunov function using (4) is ´

6
 2
as follow:   ? k  k  k k )
k 1 2k
 
 
6 ˙
1 1
V1   xT C  ,  D    x    k  k  Multiplying (8) by e t gives:
 2  k 1  k

 xT [ D   y  C  , y   (7) d
Vet    et (9)
dt
 
 G    B   FL   ]
Integrating (9) leads to the following inequality:
       
6 ˙
1
x T T
 k ˆ k    
k
m m
k 1 0  V (t )   V (0)   e t   V (0) 
    (10)

Substituting (5) into (7), it is obtained: Based on above results and Barbalat’s lemma, all
error signals will go to zero when time goes to infinite-
time.

© 2020 Int. J. Mech. Eng. Rob. Res 44

 International Journal of Mechanical Engineering and Robotics Research Vol. 9, No. 1, January 2020

III. SIMULATION RESULTS With the purpose of comparison, the simulation is

exercuted for both proposed controller and PD controller
Based on above results and Barbalat’s lemma, all
[15].The simulation has been conducted for three cases:
error signals will go to zero when time goes to infinite-
- Case 1: The parameters of system are rated, the
time.
To evaluate the correctness and suitability of the machine works without payload.
proposed robust adaptive controller, the algorithm was set - Case 2: The parameters of system are rated, the
and simulated in Simulink software with an excavator machine works with full payload.
system given parameters as the following [18]: - Case 3: The parameters of model change, the
Mbo=1566; Mst=735; Mbu=432; Mload=500; machine works will full payload.
Ibo=14250.6; Ist=727.7; Ibu=224.6; In each case, the results were compared with the
response of PD controller which was mentioned in [18].
a2=5.16; a3=2.59; r2=2.71; r3=0.64; r4=0.65; Simulation results for Case 1, Case 2, and Case 3 are
Bbo=0.02; Bst=0.02; Bbu=0.02; shown in Figs. 2, 3 and 4, respectively. In these figures,
the solid line (Thetad) represents the reference value of
The parameters for the control law is chosen by and dashed line (Theta) is for real value of .
practical method (trial – and – error):

 = diag(1), K = diag(2x106, 1.5x106, 105), k = 1,  k' =

0.01, k = 0.1

Figure 2. System responses under condition of without payload and rated system parameters

Figure 3. System responses under condition of full payload and rated system parameters.

© 2020 Int. J. Mech. Eng. Rob. Res 45

 International Journal of Mechanical Engineering and Robotics Research Vol. 9, No. 1, January 2020

Figure 4. System responses under condition of full payload and system parameters variation.

In Fig. 2, when system works without load and the the PD controller was setup to evaluate the effectiveness
parameters of model is rated, the position response of of the two control schemes. The simulation results show
joints by using proposed controller and PD controller is that in comparison with PD controller, the proposed
absolutely tracked to desired value, and the tracking error algorithm guarantees the better characteristics under all
is trivial. In the second case (Fig. 3), the system conditions. i.e. all joints of excavator arm absolutely track
parameters remain unchanged but the excavator work to desired value even if there is the effect of the load
with full payload. The adaptive controller gives the disturbance and the impact of the uncertainties of system
position response of joints almost no deviation from model.
desired trajectory, while the PD controller gives a
tracking error about 0.05 rad. In case system works with CONFLICT OF INTEREST
full payload and the parameter of system change (figure
The authors declare no conflict of interest.
4), the results obtained for the adaptive controller are still
relatively good. The PD controller give a maximal
tracking error about 0.15 rad. AUTHOR CONTRIBUTIONS
From the simulation results, it can be seen that the PD Nga Thi-Thuy Vu conducted the research, setup the
controller can make the system only work well when the simulation, wrote the paper, and approved the final
elements of system are determined. When the system has version.
load disturbance, the tracking error appears (0.05 rad) and
will increase to 0.15 rad if the system is affected by REFERENCES
addition disturbances. This prove that the PD controller
[1] S. Kumar, K. Rani, and V. K. Banga, “Robotic arm movement
will increase tracking error when system has disturbance. optimization using soft computing,” IAES International Journal
Meanwhile, for adaptive controller, the responses of of Robotics and Automation, vol. 6, no. 1, 2017.
system under condition of system uncertainties as well as [2] A. A. Jaber and R. Bicker, “Wireless fault detection system for
payload noise are so good. an industrial robot based on statistical control
chart,” International Journal of Electrical and Computer
Engineering, vol. 7, no. 6, pp. 3421-3425, 2017.
IV. CONCLUSIONS [3] H. J. Azhar, F. Hadary, and S. Syaifurrahman, “Designing of
robot gamelan music using ATmega 16 microcontroller,” IAES
This paper has presented the robust adaptive controller International Journal of Robotics and Automation, vol. 6, no. 2,
that does not depend on the system model for excavator 2017.
arm. The controller includes two parts, the state feedback [4] H. Yu, Y. Liu, and M. S. Hasan, “Review of modelling and
remote control for excavators,” International Journal of
part for system stability and the adaptive part for Advanced Mechatronic Systems, vol. 2, no. 1–2, pp. 68–80, 2009.
compensation. The controller has simple structure, easy [5] A. J. Koivo, “Kinematics of excavators (backhoes) for
to implement but still guarantees good performances to transferring surface material,” Journal of Aerospace Engineering,
the uncertainty of the system. The stability and suitability vol. 7, no. 1, pp. 17–32, 1994.
[6] P. K. Vaha and M. J. Skibniewski, “Dynamic model of
of the overall system as well as the convergence of the excavator,” Journal of Aerospace Engineering, vol. 6, no. 2, pp.
adaptive component were demonstrated by Lyapunov 148–158, 1993.
stability theory and examined through simulation. Also,

© 2020 Int. J. Mech. Eng. Rob. Res 46

 International Journal of Mechanical Engineering and Robotics Research Vol. 9, No. 1, January 2020

[7] A. J. Koivo, M. Thoma, E. Kocaoglan, and J. Andrade-Cetto, [17] H. Feng, C. B. Yin, W. W. Weng, W. Ma, J. J. Zhou, W. H. Jia,
“Modelling and control of excavator dynamics during digging and Z. L. Zhang, “Robotic excavator trajectory control using an
operation,” Journal of Aerospace Engineering, vol. 9, no. 1, pp. improved GA based PID controller,” Mechanical Systems and
10–18, 1996. Signal Processing, vol. 105, no. 5, pp. 153-168, 2018.
[8] S. Tafazoli, P. D. Lawrence, and S. E. Salcudean, “Identification [18] Y. Liu and M. S. Hasan, “Modelling and remote control of an
of inertial and friction parameters for excavator arms,” IEEE excavator,” International journal of automation and computing,
Transactions on Robotics and Automation, vol. 15, no. 5, pp. vol. 7, no. 3, pp. 349-358, 2010.
966–971, 1999.
[9] Y. H. Zweiri, “Identification schemes for unmanned excavator
arm parameters,” International Journal of Automation and Copyright © 2020 by the authors. This is an open access article
Computing, vol. 5, no. 2, pp. 185–192, 2008. distributed under the Creative Commons Attribution License (CC BY-
[10] C. P. Tan, Y. H. Zweiri, K. Althoefer, and L. D. Seneviratne, NC-ND 4.0), which permits use, distribution and reproduction in any
“Online soil parameter estimation scheme based on Newton- medium, provided that the article is properly cited, the use is non-
Raphson method for autonomous excavation,” IEEE/ASME commercial and no modifications or adaptations are made.
Transactions on Mechatronics, vol. 10, no. 2, pp. 221–229, 2005.
[11] S. Tafazoli, S. E. Salcudean, K. Hashtrudi-Zaad, and P. D.
Lawrence, “Impedance control of a teleoperated excavator,” Nga Thi -Thuy Vu received the B.S. and M.S.
IEEE Transactions on Control Systems Technology, vol. 10, no. 3, degrees in electrical engineering from Hanoi
pp. 355–367, 2002. University of Science and Technology, Hanoi,
[12] Z. Lu and A. A. Goldenberg, “Robust impedance control and Vietnam, in 2005 and 2008, respectively, and
force regulation: Theory and experiment,” The International the Ph.D. degree in electronics and electrical
Journal of Robotics Research, vol. 14, no. 3, pp. 225–254, 1995. engineering from Dongguk University, Seoul,
[13] Q. P. Ha, Q. H. Nguyen, D. C. Rye, and H. F. Durrant-Whyte, Korea, in 2013. She is currently with the
“Impedance control of a hydraulically-actuated robotic Department of Automatic Control, Hanoi
excavator,” Automation in Construction, vol. 9, no. 5–6, pp. 421– University of Science and Technology, Hanoi,
435, 2000. Vietnam, as a Full Lecturer. Her research
[14] P. H. Yang, S. P. Wang, J. S. Chiang, and C. H. Wang, “The interests include control theory, robot control, motion control, and
adaptive control applied to the analysis of the excavator renewable energy system.
dynamics,” IEEE Conference on Innovative Computing
Information, 2008.
[15] H. M. Yousif and K. GaneshInterval, “Type-2 fuzzy position
control of electro-hydraulic actuated robotic excavator,” Int. J.
Min. Sci. Technol., vol. 22, pp. 437-445, 2012.
[16] D. L. Hanh, K. K. Ahn, N. B. Kha, and W. K. Jo, “Trajectory
control of electro-hydraulic excavator using fuzzy self tuning
algorithm with neural network,” J. Mech. Sci. Technol., vol. 23,
pp. 149-160, 2009.

© 2020 Int. J. Mech. Eng. Rob. Res 47