See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332586279

PID-SMC controller for a 2-DOF planar robot

Conference Paper · February 2019

DOI: 10.1109/ECACE.2019.8679436

CITATIONS READS

10 2,332

3 authors:

Umme Zakia Mehrdad Moallem

Simon Fraser University Simon Fraser University
20 PUBLICATIONS   79 CITATIONS    130 PUBLICATIONS   2,339 CITATIONS   

SEE PROFILE SEE PROFILE

Carlo Menon
Simon Fraser University
301 PUBLICATIONS   4,305 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

D2D communications View project

Hand gesture classification using force myography View project

All content following this page was uploaded by Umme Zakia on 26 April 2019.

The user has requested enhancement of the downloaded file.

 2019 International Conference on Electrical, Computer and Communication Engineering (ECCE), 7-9 February, 2019

PID-SMC controller for a 2-DOF planar robot

Umme Zakia1, Mehrdad Moallem2 and Carlo Menon1, Member, IEEE

Abstract— Sliding mode controller (SMC) is considered as a controlling tracking of 3- DOF end-effector for rehabilitation
robust control method. In this paper, a combination of SMC and [8]. A sliding mode controller based on sliding surface with
traditional proportional-integral-differential (PID) control PID for a 3- DOF spatial manipulator for tracking
scheme for a 2-degree of freedom (2-DOF) planar manipulator performance obtains higher accuracy and improved
is considered. Since the PID control does not require precise robustness [9]. A combination of adaptive PD and SMC
system modelling, a sliding mode surface for the SMC control is controller shows stable system design with appropriate control
designed using the PID component parameters of the law [10]. A non-linear sliding control algorithm for 6-DOF
manipulator. The PID control adjusts the errors of the closed arm-like manipulator shows efficient trajectory tracking over
loop system and the SMC control ensures fast convergence
the conventional PID control mechanism [11]. An SMC
towards the sliding mode surface. Hence, this hybrid controller
control for controlling major axes joint angles of the 3-DOF
possesses features of both control schemes and obtains better
system stability. Puma manipulator obtains good tracking characteristics [12].
A non-linear PD with SMC control for a 3-DOF planar robot
I. INTRODUCTION is designed for better tracking performances [13]. For pick-
and-place tasks with accurate position control, a fast and
Controlling systems in industrial environment is a robust sliding model controller is observed [14].
continuous process and requires monitoring for quality of
production management [1]. Closed loop systems are widely In this paper, we design a hybrid PID-SMC controller for
used in the industries where feedback mechanisms are a 2-DOF planar robot which has two revolute joints (RR).
implemented to adjust errors produced in the systems [1], [2]. First, the PID controller is developed, then it is used as the
PID controller is well known as a closed loop mechanism sliding surface of the SMC controller with a sign function. The
where proportional, integral and differential of the errors make combined scheme provides fast adaptation of the uncertainty
the system adjustable and delivers the desired output [3]. related to robot dynamics and better system stability. Our
perspective is to study the effectiveness of the proposed
Sliding mode controller is used in nonlinear systems that scheme and to compare the performance of the PID controller
modifies dynamics of the system and forces the system to slide with the proposed PID-SMC control scheme for the robot
along the system's standard behavior. It is robust and efficient system. Simulations are performed and results show
in maintaining stability for nonlinear dynamic systems. The improvements of the PID-SMC controller over the
characteristics and parameters of the SMC control design is conventional PID controller.
governed by the sliding mode surface. To obtain the sliding
mode, the system state has to satisfy the dynamic equation, II. BACKGROUND STUDY
and the sliding motion is ensured by infinite switching [4].
A. 2-DOF Planar Manipulator
Control mechanism in industrial environments are largely The dynamic equations of motion of a robot is described
designed with the popular PID controllers. Recent researches as,
focus on sliding mode control designs which are compared
with the traditional PID control. In recent years, researches are ( ) ″+ ( ′, ) + ( )= (1)
being carried out to study the performance improvements of
where B(q) is the mass inertia, C(q) is the Coriolis force on the
trajectory control of robotic systems by combining these two
manipulator, g(q) is the gravitational force acting on the
controllers. In 2005, a chattering-free controller eliminates
torque rippling problem for a 2- DOF direct-drive system and manipulator and q″, q′, q are acceleration, velocity and
improves performance during position and trajectory tracking displacement of the joints of the manipulator respectively, and
[5]. A sliding mode control scheme for a deep-sea manipulator F is the joint driving torque [1], [2].
is designed with self-adaptive velocity control and trajectory Our focus is to observe the performance of a hybrid PID-
tracking and obtains good performance [6]. For a 2-DOF SMC controller and we have chosen a simple 2 -DOF planar
robot, a fuzzy tuning process is developed to adjust gain of robot, a classic multi-input multi-output (MIMO) system and
PID and SMC controller where the dynamic equation is a benchmark for assessment performance of nonlinear
defined by a sliding surface [7]. An adaptive neuro-fuzzy dynamic system. For simplicity, we assume that there are no
inference system is proposed to solve inverse kinematics for external disturbances present in the system. The Lagrange
equation to obtain the dynamics of a planar robot with two
revolute joints is as follows
ℒ= − E (2)
Research supported by the Natural Sciences and Engineering Research
Council of Canada (NSERC), the Canadian Institutes of Health Research where KE is the kinetic energy and PE is the potential energy
(CIHR) and the Canada Research Chair (CRC) program. of the robot. Fig.1 shows a 2-DOF planar manipulator that has
1
Authors are with Menrva Research Group, Schools of Mechatronic two revolute joints with, 1 = 2 = masses of link 1 and link
systems Engineering and Engineering Science, Simon Fraser University,
2, and 1 = 2 = link length 1 and link length 2, respectively.
Metro Vancouver, BC, Canada (email: {uzakia, cmenon}@sfu.ca).
2
Author is in Schools of Mechatronic systems Engineering, Simon Thus, we can write
Fraser University, Surrey, BC, Canada (email: [email protected]).
1= 1sin 1 (3),
*Corresponding author: C. Menon (email: [email protected], phone: +1-
778-782-9338; fax: +1-778-782-7514).
1= 1cos 1 (4),

978-1-5386-9111-3/19/$31.00 ©2019 IEEE

 ( ) = [− ( 1+ 2) 1sin 1− 2 2sin( 1+ 2)
− 2 2sin ( 1+ 2)] (17),
2= 1sin 1+ 2sin( 1+ 2) (5),
and
2= 1cos 1+ 2cos( 1+ 2) (6).
T
=[ 1 2] (18).

The dynamics of the system can be expressed with the state-

space equations. To this end, let

1=  1 ,
which results in
1′ = 1 – 1 (19),
And, let

2=  2

=> 2′ = 2 − 2 (20).
Fig. 1. 2-DOF (RR) Planar robot with two revolute joints. Thus, the complete system equations are given by

B. Dynamics of the 2-DOF Planar Manipulator 1′= 1 − 1

The kinetic energy (KE) of the 2-DOF planar manipulator 2′= 2 − 2

could be formed as follows
[ ″1 ″2]T = ( )−1[ − ( ′, ) − ( )] +
1′ + ½ 1′ + ½ 2′ + ½ 2′
2 2 2 2
=½ 1 1 2 2 (7)
[ 1( 1 − 1) − 1 1′+ 1 1
Further simplification of (7) results in
2( 2 − 2) − 2 2′+ 2 2 ] (21)
= ½ ( 1+ 2) 12 1′2 + ½ 2 22 1′2
+ 2 22 1′ 2′+ 2 22 2′2 III. PROPOSED PID-SMC CONTROLLER
+ 2 1 2cos 2( 1′ ′2+ 1′2) (8) The system control scheme is based on the dynamics of
the 2-DOF planar robot and the motion of the robot is based
The Potential Energy is given by
on the system equation (1). We can derive the accelerations of
= 1 1cos 1+ 2 ( 1cos 1+ 2cos ( 1+ 2)) (9) the joints as
Next, we can obtain the Lagrange Dynamics based on (8) and ( ) ″ + ( ′, ) + ( ) =
(9) as follows
=> q″ = ( ) −1[− ( ′, ) − ( ) + ] (22)
ℒ = ½ ( 1+ 2) 1 1′ +½ 2 2 1′ + 2 2 2 2
2 2 1′ 2′
2
Let
+ ½ 2 22 2′2+ 2 1 2cos 2( 1′ 2 ′+ 1′ )
2

− 1 1cos 1− 2 ( 1cos 1+ 2cos( 1+ 2)) (10) ~= ( ) −1

Now, let us form the dynamics using the following => = ( ) ~ (23)
1,2 = / [ ℒ/ ′1,2]− ℒ / 1,2 (11) Now, we redefine the system to the new input
T
The dynamic equations can be simplified as follows ~=[ 1 2] (24)
(( 1+ 2) 12+ 2 22+2 2 1 2cos 2) 1″ Nevertheless, the torque inputs to the system are given by
+ ( 2 22+ 2 1 2cos 2) 2″ [ 1 2]
T
= ( )[ 1 2]
T
(25)
− 2 1 2sin 2(2 1′ 2′+ 2′2)
– ( 1+ 2) 1sin 1− 2 2sin( 1+ 2) = 1 (12) The error signals related to joint angles are given by

( 2 2
2
+ 2 1 2cos 2) 1″+
2
2 2 2″
( 1) = 1 − 1 (26),
− 2 1 2sin 2 1′ ′2− 2 2 sin ( 1+ 2) = 2 (13) ( 2) = 2 − 2 (27).
Finally, we define the motion of the robot by A. Developing the PID Controller
( ) ″ + ( ′, ) + ( ) = ; A proportional-Integral-differential (PID) controller for
any input would have the general structure as below [1],[2]
where
= P + ′+  (28)
T
= [ 1 2] (14),
where P, I, and D are the proportional, integral and
( ) = [( 1+ 2) 12+ 2
2 2 +2 2 1 2cos 2) ( 2
2 2 + 2 1 2cos 2) derivative control gains of the PID controller.
2
2 2 + 2 1 2cos 2 2 2
2
] (15), For the 2-DOF RR planar robot, we have
( ′, ) = [− 2 1 2sin 2 (2 ′1 ′2+ ′22 ) 1= P1 ( 1 − 1) − 1 ′1+ 1  1 (29)
− 2 1 2sin 2 ′1 ′2 ] (16), 2= P2 ( 2 − 2) − 2 ′2+ 2  2 (30)
Hence, the complete system equations with the PID control 1) States results of Parameter set 1 with PID control
would be given by The error forms of 1 & 2 for the PID controller is as
″ = ( ) −1[− ( ′, ) − ( ) ]+ ~ (31) follows
where,
Theta-1 error Theta-2 error
3.5 1

3 0.5

T
~=[ 2]
2.5
0
1 2
-0.5
1.5

=[ P 1( 1 − 1) − 1 1+ 1
-1

1 1
-1.5
0.5

− 2) − 2 2+  ] (32)
-2
0
P2( 2 2 2 -0.5
-2.5

-3

B. Developing the SMC controller

-1

-1.5 -3.5
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

The SMC controller is designed by establishing the

time (sec) time (sec)

(a) (b)
sliding mode surface and a switching function responsible for 100
Torque of theta 1
20
Torque of theta 2

reducing chattering effect. The sliding mode surface is built 80 15

using the PID controller parameters such as 60 10

40 5

s= P + ′+  (33) 20 0

0 -5

Here, s is the sliding mode surface and P, D and I are the -20 -10

control gains of the PID control as defined earlier. -40 -15

-60 -20

The control law of the hybrid PID-SMC is selected as follows

0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
time (sec) time (sec)

(c) (d)
u = s + Ksal Sw (s, σ) (34) Fig. 2. Error forms of 1 (a) & 2 (b), Torque of 1 (c) & 2 (d) for the PID
where Ksal is the control gain and Sw is the switching controller (for parameter set 1).
function. This function can be defined as follows
2) States results of Parameter set 1 with PID-SMC control
Sw (e, σ) = |e|σ sign (e) (35) The error forms of 1 & 2 for the PID-SMC controller is
where σ is chosen between 0 to 1. The switching function as follows
combines features of the sign and the exponential function. 3.5
Theta-1 error
0.5
Theta-2 error

When the system state is far from the sliding surface, it forces 3 0

the system to reach the sliding surface quickly. In contrary, 2.5

2
-0.5

when the system state is near the sliding surface, it restricts 1.5
-1

the approach speed towards the surface. Thus, it can

-1.5
1
-2

minimize the chattering effect of the system and achieves fast 0.5

0
-2.5

convergence towards stability. -0.5 -3

-1 -3.5

IV. SIMULATIONS & RESULTS

0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
time (sec) time (sec)

For the PID control gain scheme, the state variables x1′, (a) (b)
x2′, 1″ and 2″ are used to represent the dynamics of the 2- 250
Torque of theta 1
100
Torque of theta 2

DOF robot with MATLAB ode45 command. The simulations 200 80

are carried out for the PID and the PID-SMC controllers with 150 60

two sets of parameter values as stated in ‘Parameter set 1’ and 100 40

‘Parameter set 2’. For ‘Parameter set 1’, the final positions 50 20

T T
are [ 1 2 ] = [ /2 − /2] while the system has initial 0 0

positions of 0 = [− /2 /2 ]T. For ‘Parameter set 2’, the final -50 -20

positions are [ 1 2 ]T = [ /2 ]T while the system has initial -100

0 2 4 6 8 10 12 14 16 18 20
-40
0 2 4 6 8 10 12 14 16 18 20

positions of 0= [ /2 ]T.
time (sec) time (sec)

(c) (d)
A. Simulating with Parameter set 1
The first set of values for the parameters for the PID Fig. 3. Error forms of 1 (a) & 2 (b), Torque of 1 (c) & 2 (d) for the PID-
SMC controller (for parameter set 1).
control gains are selected as follows
P1=15, 1=7, 1=10, B. Simulating with Parameter set 2
P2=15, 2=10, 2=10.
The second set of values for the parameters are for the PID
controller after the trial-and-error are set to as follows
In the 2-DOF robot design, units for mass and link length
P1=30, 1=12, 1=20,
are kg and meter, respectively. For simplification, we have
considered mass and link lengths as follows P2=32, 2=22, 2=30.

1= 2= 1= 2=1. Now the mass and link lengths of the 2-DOF robot are
For the SMC control design, Ksal is set to 5 for the PID- considered as
SMC controller. The value of σ is selected by trial-and-error. 1=5, 2=5 & 1= 2=0.34.
It is observed that σ = 0.3 provides the best results for both
PID and PID-SMC controllers. where units are in kg and meter.
 Furthermore, we use the same values of Ksal and σ as the robot. Simulation results in Fig. 2 and Fig. 4 show the
mentioned in Parameter set 1 for the PID-SMC control design tracking performance and sensitiveness of the PID controller
such as Ksal = 5 and σ = 0.3. in favor of the discussion. Hence, designing PID control
parameters needs careful adjustments for system operations
3) States results of Parameter set 2 with PID control
such as undefined final positions, trajectory tracking, etc.
The error forms of 1 & 2 for the PID controller is as
follows TABLE I. SIMULATION PARAMETERS
Theta-2 error
1.6

1.4
Simulation Trials
1.2
Param. Parameter Set 1 Parameter Set 2
1
Mass &
0.8

Length 1= 2= 1= 2=1. 1=5, 2=5 & 1= 2=0.34.

0.6

P1=30, 1=12, 1=20,

0.4

0.2
PD P1=15, 1=7, 1=10,
0 component P2=15, 2=10, 2=10. P2=32, 2=22, 2=30 .
-0.2
SMC
-0.4
0 2 4 6 8 10
time (sec)
12 14 16 18 20
component Ksal = 5 , σ = 0.3 Ksal = 5 , σ = 0.3
(a) (b)
Initial = [ /2 - /2 ]T Initial = [ /2 ]T
8
Torque of theta 2
Position
7
Final = [- /2 /2]T Final = [ /2 ]T
6

4
The PID-SMC controller, on the other hand, shows
3
improvements by quickly converging the tracking errors to
2 zero. Fig. 3 and Fig. 5 show that the manipulator is capable to
1
rapidly trail the preferred path with the PID-SMC controller
0
0 2 4 6 8 10
time (sec)
12 14 16 18 20 in comparison to the PID controller, as shown in Fig. 2 and
(c) (d) Fig. 4. Chattering occurs when the 1 moves from /2 to
− /2 and 2 moves from − /2 to /2 in ‘parameter set 1’, as
Fig. 4. Error forms of 1 (a) & 2 (b), Torque of 1 (c) & 2 (d) for the PID
controller (for parameter set 2). shown in Fig. 3; while this effect is not present when 1 moves
from to /2 and 2 moves from /2 to in ‘parameter set
4) States results of Parameter set 2 with PID-SMC control 2’, as shown in Fig. 5. This is because of the smooth trajectory
The error forms of 1 & 2 for the PID-SMC controller is between the joints while we simulate the 2-DOF robot with
as follows ‘parameter set 2’. The PID-SMC controller allows faster
settling time than the PID controller and shows more stability.
Theta-1 error Theta-2 error
0.2 2.5

0
2 V. CONCLUSION
-0.2

-0.4 1.5
In this paper, a PID-SMC controller is implemented for a
-0.6

-0.8
1 2-DOF planar robot system with two revolute joints. The
-1 0.5 system dynamics is obtained with Lagrange method. Initially
-1.2
0
a PID controller is designed which becomes the sliding
-1.4

-1.6 -0.5
surface for the SMC controller. Simulation results illustrate
0 2 4 6 8 10
time (sec)
12 14 16 18 20 0 2 4 6 8 10
time (sec)
12 14 16 18 20
that PID-SMC controller is well in tracking performance of
(a) (b) the robotic arm compared with the traditional PID methods.
0
Torque of theta 1 The proposed PID-SMC control scheme will be applied on
-10 other spatial manipulator platforms in our future works.
-20

-30
REFERENCES
-40 [1] Stanislaw H. ˙Zak, Systems and Control, Oxford University Press,
-50
Newyork, 2003.
-60 [2] John J. Craig, Introduction to Robotics: Mechanics and Control, 3rd Ed.,
-70
Pearson Education International, 2005.
0 2 4 6 8 10 12 14 16 18 20
time (sec)
[3] P. Tomei, “Adaptive PD Controller for Robot Manipulators,” IEEE
(c) (d) Trans. Robotics and automation, vol. 7, pp. 565–570, Aug. 1991.
[4] A. A. Pervozvanski , L. B. Freidovich, “Robust stabilization of robotic
Fig. 5. Error forms of 1 (a) & 2(b), Torque of 1 (c) & 2(d) for the PID-
manipulators by PID controllers,” Dynamics and Control, vol. 9, pp.
SMC controller (for parameter set 2).
203–222, 1999.
[5] S. Bogosyan, P.R. Suravaram and M. Gokasan, “Sliding mode based
C. Discussion rejection of load and torque ripple in a direct-drive 2 DOF robot arm ,”
PID control gains such as P, and are related to the 31st Annual Conference of IEEE Industrial Electronics Society,
(IECON), Raleigh, NC, USA, Nov. 2005.
direct error & speed development, interaction speed during
[6] G. H. Xu, Z. H. Xiao, Y. Guo, and X. B. Xiang, “Trajectory tracking
state change and overall error cancelation, respectively. for underwater manipulator using sliding mode control,” IEEE
Though the PID controller produces stability through error International Conference on Robotics and Biomimetics (ROBIO), pp.
cancellation in the closed loop control mechanism, slight 2127–2132, Sanya, China, Dec. 2007.
deviations in the controller parameters may result in instability [7] D.l Vyas, J. Ohri and A. Patel, “Comparison of conventional & fuzzy
such as overshoot, oscillations etc. Similarly, the controller based sliding mode PID controller for robot manipulator,” International
Conference on Individual and Collective Behaviors in Robotics
performance is dependent on the initial and final positions of (ICBR), Sousse, Tunisia, Dec. 2013.
[8] F. Sado, S. N. Sidek, and H. M. Yusuf, “Intelligent trajectory [12] S. Kasera, A. Kumar, and L. B. Prasad, “Trajectory Tracking of 3-DOF
conversion and inverse dynamic control of a 3-DOF neuro- Industrial Robot Manipulator by Sliding Mode Control,” 4th IEEE
rehabilitation platform,” 10th Asian Control Conference (ASCC), Uttar Pradesh Section International Conference on Electrical,
Malaysia, Jun. 2015. Computer and Electronics (UPCON) GLA University, Mathura, India,
[9] Kaige Wan, Aiguo Wu, Haiting Liu, and Na Dong, “Trajectory 2017.
Tracking of 3-DOF Spatial Robot Manipulator,” 12th World Congress [13] J. Tang, P.R. Ouyang, W.H. Yue, and H.M. Kang, “Nonlinear PD
on Intelligent Control and Automation (WCICA), Guilin, China, 2016. Sliding Mode Control for Robotic Manipulator,” IEEE International
[10] P. R. Ouyang, J. Tang, W.H. Yue, and S. Jayasinghe, “Adaptive PD Conference on Advanced Intelligent Mechatronics (AIM), Munich,
plus Sliding Mode Control for Robotic Manipulator,” Proc. of IEEE Germany, July, 2017.
International Conference on Advanced Intelligent Mechatronics [14] A. A. Mohammed and A. Eltayeb, “Dynamics and Control of a Two-
(AIM), Banff, AB, Canada, July 2016. link Manipulator using PID and Sliding Mode Control,” International
[11] S. A. Ajwad, R. U. Islam, M. R. Azam, M. I. Ullah and J. Iqbal, Conference on Computer, Control, Electrical, and Electronics
“Sliding mode control of rigid-link anthropomorphic robotic arm,” 2nd Engineering (ICCCEEE), Khartoum, Sudan, Aug. 2018.
International Conference on Robotics and Artificial Intelligence
(ICRAI), Rawalpindi, Pakistan, Nov. 2016.

View publication stats