Modeling, Simulation and Control of a Robotic Arm

Nafiseh Ebrahimi
Department of Mechanical Engineering; the University of Texas at San Antonio

Abstract— the precise control upon each degree of freedom

of a robotic arm is a great challenge in implementing
industrial work. To simplify mechatronics motion control
systems design, this paper proposes mathematical modeling,
simulation and control of a given electric motor in terms of
input volt, Vin and output motions. The proposed model can
be used to select, design, test and validate both plant's and
motion control design to meet desired output performance. It
presents a basic example of PID control applied to a robotic
manipulator arm.
Keywords: Robotic arm, Simulation, PID controller, Full Fig.1. Equivalent representation of PMDC motor electromechanical
state feedback controller component
I. INTRODUCTION
mobile robot, and single joint robot arm, shown in Fig.2.
The term robotics is practically defined as the study, design
and use of robot systems for manufacturing. Robots are
generally used to perform unsafe, hazardous, highly repetitive,
and unpleasant tasks [1]. Most used actuator in mechatronics
motion control applications is DC motors. Despite a lot of
resources that propose different selections and design of
control strategies to control motions in desired fashion most
control system used are based on convention PID controller
[2].
II. DESCRIPTION OF THE SYSTEM
A. System Dynamics
Fig.2 Single joint robot arm.
Most used actuator in mechatronics motion applications is
DC motor, therefore, motion control can be simplified to DC The motor torque is given by Eq (1). The generated EMF
motor motion control. PMDC motor is an example of voltage, ea, is given by Eq (2).
electromechanical system, having both mechanical of electric
components, a simplified equivalent representation the 𝑇𝑚 = 𝐾𝑡 𝑖𝑎 (1)
armature controlled PMDC motor's two components is shown 𝑑θ
𝑒𝑎 = 𝐾𝑏 𝜔𝑚 = 𝐾𝑏 𝑚 (2)
in Fig.1. 𝑑𝑡

Applying Kirchoff’s law around the electrical loop,

substituting, result in Eq(3) .Taking Laplace transform and
rearranging, result in Eq(4):

𝑑𝑖 𝑑θ𝑚
𝐿𝑎 + 𝑅𝑎 𝑖 = 𝑉 − 𝐾𝑒 (3)
𝑑𝑡 𝑑𝑡

𝑉(𝑆)−𝐾𝑒 𝑆θ𝑚 (𝑠)

𝐼𝑎 (𝑠) = (4)
𝑅𝑎 + 𝐿𝑎 𝑆
 Performing the energy balance; the sum of the torques must Root Locus diagram for mentioned open loop system is:
equal zero, yields Eq.(5), Taking Laplace transform and
rearranging, result in Eq(6):

𝑑2θ 𝑑θ
𝐾𝑡 ∗ 𝑖 − 𝑇𝐿𝑜𝑎𝑑 − 𝐽𝑚 ( 2 ) − 𝑏𝑚 ( ) = 0 (5)
𝑑𝑡 𝑑𝑡
𝐾𝑡 𝐼(𝑠) = (𝐽𝑚 𝑠 + 𝑏𝑚 )𝑠 θ(s) (6)

substituting Eq.(4) in Eq.( 6) and rearranging, to have the

transfer function given by Eq.(7)in terms of the input voltage,
V(s) and output motor shaft angle θ𝑚 .

Ө𝑚 (𝑠) 𝐾𝑡
𝐺𝑚 = = (7)
𝑉(𝑠) 𝑠[(𝑅𝑎 + 𝐿𝑚 𝑠)(𝐽𝑎 𝑠 + 𝑏𝑚 )𝐾𝑏 𝐾𝑡]

the total equivalent inertia, Jeq and total equivalent damping,

beq at the armature of the motor are given by: Fig.3. Root locus diagram

The loot locus is drawn for just open loop transfer function
𝑏𝑙𝑜𝑎𝑑 𝐽𝑙𝑜𝑎𝑑
𝑏𝑒𝑞 = 𝑏𝑚 + , 𝐽𝑒𝑞 = 𝐽𝑚 + with gain K=1. However, if there is a variable gain (K) in the
𝑛2 𝑛2
system for maintaining stability of the system K would have a
Substituting Jeq and beq in Eq. (7), the total equivalent range of K< 415.
transfer function, relating input voltage Vin and Arm-load On the other hand, stability can be considered investigating the
output angular position θLoad, is given by Eq. (8). In this eigenvalues of A (system state matrix). if the eigenvalues are
transfer function, gear ration (n) which is the transfer function positive, the system will not satisfy the condition of BIBO
of gear system is included. stability, and will therefore become unstable.
The eigenvalues of matrix A are: 0, -0.9521, -4.3449, therefore
θ 𝐿𝑜𝑎𝑑 (𝑠) the output signal magnitude would not exceed a finite amount
𝐺=
𝑉𝑖𝑛 (𝑠) and it is stable.
𝐾𝑡 ∗ 𝑛
𝐺= (8)
𝐿𝑎 𝐽𝑒𝑞 𝑠3 + (𝑅𝑎 𝐽𝑒𝑞 + 𝑏𝑒𝑞 𝐿𝑎 )𝑠 2 + (𝑅𝑎 𝑏𝑒𝑞 + 𝐾𝑡 𝐾𝑏 )𝑠
B. System Performance Characteristics
The robot arm system to be designed, has the following
nominal values; arm mass, M= 8 Kg, arm length, L=0.4 m, Step response of open loop system to Vin= Step= 1V is:
and viscous damping constant, b = 0.09 N.sec/m. The
following nominal values for the various parameters of
eclectic motor used: Vin=12 Volts; Jm=0.02 kg·m²; bm
=0.03; Kt =0.023 N-m/A; Kb =0.023 V-s/rad; Ra =1 Ohm;
La=0.23 Henry; TLoad, gear ratio, for simplicity, n=1.

Substituting values of parameters into transfer function

gives Eq. (9):
0.023
𝐺= (9)
0.02913 𝑆 3 + 0.1543 𝑆 2 + 0.1205 𝑆

Regarding state space specification A, B, C and D matrix

could be obtain as below Eq. (10):

−5.2969 −4.1366 0 1
𝑋 . = [ 1.0000 0 0] 𝑋 + [0] 𝑈 (10)
Fig.4. Open loop step respnse
0 1.0000 0 0

𝑌 = [0 0 0.7896] 𝑋 + [0]𝑈
It is obvious that this system with no feedback from output
will continue to infinity. Using MATLAB, below system
III. OPEN LOOP SIMULATION performance characteristics obtained (table 1).
A. Stability
 Table1. Open loop system characteristics
Rise Time: NaN
Settling Time NaN
Settling Min NaN
Settling Max NaN
Overshoot NaN
Undershoot NaN
Peak Inf
Peak Time Inf

The system needs to be controlled with at least a feedback from

output. Simple potentiometer could turn the feedback in the
system. Therefore, we need to close the system.
Fig.6. Negative feedback closed loop system step response
IV. CLOSED LOOP SYSTEM
Closing the system by adding arm angular position sensor
(potentiometer) gives closed loop transfer function Eq. (11): For the closed loop one, system performance characteristics
are shown in table 2.
0.023
𝑇= (11)
0.02913 𝑆 3 + 0.1543 𝑆 2 + 0.1205 𝑆 + 0.001534
Table 2. Closed loop system characteristics
Rise Time: 169.82
A negative closed loop feedback control system with forward Settling Time 303.68
controller is shown in Fig.5
Settling Min 13.54
Settling Max 14.99
Overshoot 0
Undershoot 0
Peak 14.99
Peak Time 565.92
A. Design Specifications
Our design goal is to design, model, simulate and analyze a
Fig.5. Negative feedback closed loop system block diagram control system so that a voltage input in the range of 0 to 12
volts corresponds linearly of a Robot arm output angle range
of 0 to 180, that is to move the robot arm to the desired output
For the closed loop system K=12/180= 0.0667 is considered angular position, θL, corresponding to the applied input
for feedback gain. Step response for closed loop system is voltage, Vin, with overshoot less than 5%, a settling time less
depicted in Fig.6. than 2 second and zero steady state error. The error signal, e
is the difference between the actual output robot arm position,
θL, and desired output robot arm position. For 5% of
overshoot we have:
%𝑂𝑆
−ln( )
ζ= 100 = 0.69 (12)
√π + 𝑙𝑛 (
2 2 %𝑂𝑆
)
100
and for settling time of 2 seconds we have:
4
𝑇𝑠 = =2 (13)
ζωn
Which yields
ω𝑛 = 2.899
For a second order system we have
𝑆2 + 2ζωns + ω𝑛 2 = 0 ⇒
s1,2 = −ζωn ± jωn√1 − ζ2
and the characteristics equation would be
𝑆2 + 4s + 8.4 = 0 (14)
but we do not want to use second order approximation for
designing controller. Instead we are going to use PID
command and PID Tuner of MATLAB which are far more
powerful and precise in comparison to second order
approximation.
For using PID command it is better to have an insight about
the effect of different coefficients Kp, Ki, Kd on the system.
The effects of each of controller parameters, Kp, Ki,
and Kd on a closed-loop system are summarized in table 3.

Table 3. Effects of PID controller parameters Fig.7. Closed loop system with P controller step response
CL Rise Overshoot Settling S-S Error
Response Time Time
System specifications for the controlled system is shown in
Kp Decrease Increase Small Decrease table 4.
Change
Table 4. P controlled system characteristics
Ki Decrease Increase Increase Eliminate Rise Time: 63.70
Kd Small Decrease Decrease No Settling Time 114.72
Change change Settling Min 0.91
Settling Max 0.9993
Overshoot 0
Therefore, for the mentioned system which has settling time
more than 300 seconds the most important coefficient would Undershoot 0
be Kd which decrease settling time. On the other hand, using Peak 0.9993
Ki would worsen the condition of settling time. In terms of Peak Time 212.20
steady-state error, it is already zero and does not need any
action. Similarly, overshoot is already zero and does not need Trying larger Kp s like 100 and 1000 still does not reach
to be considered. desired specification. Then, we need to take into the picture
We are going to investigate two types of controller on the other coefficients Ki and Kd. Ki is used for steady state error
closed loop system in the following. elimination while we do not have any. The main problem for
the system is high settling time so as mentioned before we
need to work with Kd to reduce this quantity. In next section
B. P Controller we focus on PID controller design.
Using PID tuner MATLAB proposes Kp, Ki and Kd
coefficients which are used in PID command to draw the step
response of the controlled system. C. PID Controller
We chose P type of controller in MATLAB, PID tuner and We chose PID type of controller in MATLAB, PID tuner and
obtained following coefficient as first estimation: obtained following coefficient as first estimation:

Kp= 2.5927 Kp= 11.67

Ki= 2.19
Which yeilds settling time: 114.7 seconds and overshoot: 0% Kd= 15.56
that setteling time is still too high and needs more
modification. Step response for closed loop system with P Which results Settling time: 64.58, overshoot: 39.56% that is
controller is depicted in Fig.7. not acceptable. Only by putting Ki=0 the results got better as
setteling time: 26.59 sec, overshoot: 0 and steady state error:0
like previous time. After some manipulation on Kp and Kd
coefficients Finally we chose the coeffients below as final
controller:

Kp= 101
Ki= 0
Kd= 94
Step response for closed loop system with PID controller is matrix is equal to the order of system (n). The scheme of full
depicted in Fig.8. state feedback system is depicted in Fig.9.

Fig.9. Closed-loop system with full state feedback controller

The control vector U is designed in the following state

Fig.8. Closed loop system with PID controller step response
feedback form

System specifications for the PID controlled system is listed 𝑈 = −𝐾𝑋 (15)
in table 5.
This leads to the following closed loop system
Table 5. PID controlled system characteristics
Rise Time: 1.29 𝑋 . = (𝐴 − 𝐵𝐾)𝑋 = 𝐴𝐶𝐿 𝑋 (16)
Settling Time 1.98
Settling Min 13.58 where
Settling Max 15.26
Overshoot 1.74 𝐴𝐶𝐿 ≡ (𝐴 − 𝐵𝐾) (17)
Undershoot 0
The gain matrix is designed in such a way that
Peak 15.26
Peak Time 2.92 |𝑆𝐼 − (𝐴 − 𝐵𝐾)| = (𝑠 − 𝑃1 )(𝑠 − 𝑃2 ) … (𝑠 − 𝑃𝑛 ) (18)

Steady state error is zero as well. Then all designing where 𝑃1 , … 𝑃𝑛 are the desired pole locations.
specifications are satisfied with final chosen Kp, Ki and Kd. Solving for K, the gain matrix K is obtained such that the state
feedback control places the closed-loop poles at the loations
of desired poles. That means the eigenvalues of A – BK are
V. FULL STATE FEEDBACK CONTROL AND LINEAR equal to the desired poles.
QUADRATIC REGULATOR CONTROL According to characteristics equation obtained from desired
specification design (Eq.14) we have the following dominant
A. Pole Placement poles:
pole placement, is a method employed in feedback control
system theory to place the closed-loop poles of a plant in pre- P1,2 = -2.0±2.0976i
determined locations in the s-plane.[1] Placing poles is
desirable because the location of the poles corresponds The third pole is chosen such that its magnitude be more than
directly to the eigenvalues of the system, which control the three times of the dominant poles. Fig.10 shows step response
characteristics of the response of the system. The system must of the system after applying pole placement method.
be considered controllable in order to implement this method.
This technique is widely used in systems with multiple inputs
and multiple outputs, as in active suspension systems.
In order to implement pole placement method all state
variable must be measurable. Moreover, system must be
controllable. Using pole placement technique, we could get
the closed loop poles to the desired location. Not only the
“dominant poles”, but “all poles” are forced to lie at specific
desired locations. Our system state space model expressed in
Eq. (10) is controllable because the rank of its controllability
 The state feedback control gain matrix K is obtained as

𝐾 = 𝑅 −1 𝐵𝑃 (20)

Where matrix P satisfies the following reduced matrix

Riccati:

𝐴𝑇 𝑃 + 𝑃𝐴 − 𝑃𝐵𝑅−1 𝐵𝑃 + 𝑄 = 0 (21)

The weighting matrices are specified such that the closed loop
system is able to track the reference signal with a control
signal that does not significant violates the saturated actuator
limits. The minimum value of J in Eq. (19) is obtained as

𝐽𝑚𝑖𝑛 = 𝑥 𝑇 (0)𝑃𝑥(0)
Fig.10. Full State Feedback Control step response For our simulation we assumed Q and R as follows to design
optimal control.
System specifications for the controlled system is listed in
𝑄 = 104 ∗ 𝐶 ∗ 𝐶 𝑇
table 6.
Table 6. Full state feedback controlled system characteristics 𝑅 = 10−4
Rise Time: 0.81
Settling Time 2.24 In Fig.11 the step response of the system after applying LQR
Settling Min 0.90 technique is displayed. We manipulated R to obtain
Settling Max 1.04 acceptable system characteristics from the optimization.
Overshoot 4.19
Undershoot 0
Peak 1.04
Peak Time 1.73

B. Linear quadratic regulator (LQR)

The theory of optimal control is concerned with operating

a dynamic system at minimum cost. The case where the
system dynamics are described by a set of linear differential
equations and the cost is described by a quadratic function is
called the LQ problem. One of the main results in the theory
is that the solution is provided by the linear–quadratic
regulator (LQR), a feedback controller whose equations are
given below. The LQR is an important part of the solution to
the LQG (linear–quadratic–Gaussian) problem
For a time-continuous system, the state-feedback law
u = –Kx minimizes the quadratic cost function J(u).
System characteristics after implementation of LQR is
∞ shown in table 7.
𝐽(𝑢) = ∫0 (𝑥 𝑇 𝑄𝑥
+ 𝑢 𝑅𝑢)𝑑𝑡 𝑇
(19)
.
Subject to 𝑋 = 𝐴𝑥 + 𝐵𝑢. Table 7. LQR controlled system characteristics
Rise Time: 0.29
The goal is to design an optimal feedback K using the Linear
Settling Time 2.85
Quadratic Regulator (LQR) problem, whose parameters are
the state matrix A, input matrix B, state weighting matrix Q, Settling Min 0.90
and input weighting matrix R. Q is a symmetric positive semi- Settling Max 1.07
definite matrix and R is a symmetric positive-definite matrix. Overshoot 7.29
Matrices Q and R imposes constraints on states and control Undershoot 0
input, respectively. Peak 1.07
Peak Time 0.63
 [3] R. Agrawal, “Modeling a Controller for an Articulated
Robotic Arm”, Intelligent Control and Automation,
VI. CONCLUSION 2012, 3, pp. 207-210
[4] F. A. Salem, “Modeling, Simulation and Control Issues
Our first approach to control the robotic manipulator arm is to
derive a mathematical model for the system using physical for a Robot ARM; Education and Research”, I.J.
governing equations on system. After obtaining the state Intelligent Systems and Applications, 2014, 04, pp.26-39
space model and block diagram for the system, it is illustrated [5] T. A. Salih, O. I. Yehea, “A Novel Control System for
that this system is stable. Robotics Devices,” Journal of Theoretical & Applied In-
Using feedback to make a closed loop system and getting a formation Technology, Vol. 28, No. 1, 2011, pp. 48-53.
[6] M. Ruderman, J. Krettek, F. Hoffmann, T. Bertram,
preliminary insight from this system step response we
designed several controllers to control the robotic arm system “Optimal State Space Control of DC Motor”,
to meet system requirements. In order to meet the design Proceedings of the 17th World Congress, The
specifications of the response which were 5% overshoot, International Federation of Automatic Control Seoul,
settling time of 2 seconds, and zero steady-state error to step Korea, July 2008.
input, PD controller, and PID controller were designed using
root locus and their performance was compared. Next, full
state feedback control was implemented by assigning desired
closed-loop poles. Finally, linear quadratic regulator was
implemented to minimize the performance index.
Table 8 illustrates the characteristic of desired, open loop and
all designed controller. System specifications for all type of
applied controllers are illustrated in table 8. It can be inferred
that best performance is for PID controller since it satisfies all
of the desired specification the best and decries settling time
significantly.
Moreover, PP (pole placement) controller satisfies the
condition though its overshoot is more than PID and introduce
a trivial amount of steady state error to the system as well.
Also, LQR controller’s overshoot exceeds the desired amount
and this is unacceptable. In addition, P controller due to its
unsatisfactory settling time is not acceptable.

Table 8. Comparison of system characteristics for different controllers

OS% 𝑇𝑠 𝑒𝑠𝑠
Desired system 5 2 0
Closed loop 0 303.68 0
P Controller 0 114.73 0
PID Controller 1.74 1.98 0
PP 4.19 2.34 1.3x10-15
LQR 7.3 0.85 1.11x10-16

VII. ACKNOWLEDGMENT
I would like to take this opportunity to express my profound
gratitude and deep regard to Dr. Bing Dong for his guidance,
valuable feedback and constant encouragement throughout
the duration of the project. His valuable suggestions were of
immense help throughout my project work.

VIII. REFERENCES
[1] G. O. Young, “Synthetic structure of industrial plastics
(Book style with paper title and editor),” in Plastics,
2nd ed. vol. 3, J. Peters, Ed. New York: McGraw-Hill,
1964, pp. 15–64.
[2] W.-K. Chen, “Linear Networks and Systems” (Book
style). Belmont, CA: Wadsworth, 1993, pp. 123–135.