Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering

12(4) (2019) 0053~0066

HTTP://JSME.IAUKHSH.AC.IR
ISSN: 2008-4927

PID Tuning Method on AGV (Automated Guided Vehicle)

Industrial Robot
Ata Jahangir Moshayedi1*, Atanu Shuvam Roy1, Liefa Liao1

1. School of Information Engineering, Jiangxi University of Science and Technology, No 86, Hongqi Ave, Ganzhou, Jiangxi,341000,
China. *
Manuscript Received --- 14 Mar. 2020; Revised --- 24 Apr. 2020; Accepted --- 13 May. 2020)

Abstract
Controlling a system can be done in various ways and methods. The classical method which even now a day as a
solution works is PID which in that with some method three-parameter of controller called P (Proportional), I
(Integral), D (Derivational) tuned to have the best controlling response from a system. The AGV robot as the
abbreviation of the Automated Guided Vehicle is as a famous robot platform which used in various industries
relies on PID controllers in various ways. Each AGV or Machine has its own set of function, hence, in order to
accomplish the exact set of workload more efficiently one need to actually tune the PID parameters accordingly so
that there cannot be an intolerable amount of energy loss, inefficiency rate, lag, lack of robustness etc. In this
paper over than introduction of PID controller and see the effect of each parameter on the real system the
compassion between hired methods on AGV robot are investigated. As this review indicates that various PID tune
method are used based on system requirements with the help of Lyapunov Direct Method, traditional Ziegler
Nichols, Fuzzy controller, human immune system called the humoral, neural network, etc to control the speed and
steering of an AGV systems.

Keywords: PID, AGV, PID Tune, Ziegler-Nichols, Automated Guided Vehicle

1. Introduction accurate and stable. It can be proved through
A Proportional-Integral-Derivative controller mathematical methods that a PID feedback loop
(PID controller) is a generic controller widely can keep the system stable, when other control
used in industrial control systems. A PID methods cause the system to have a stable error or
controller continuously calculates an error and a process is repeated that is put the system on
applies a corrective action to resolve the error [1]. loop. The main reason for having a PID controller
On the other way, the controller compares the is error correction and control management and it
collected data with a reference value, and then can be applied in various field like: motor speed
uses this difference to calculate a new input value. control, car cruise control, temperature [2]. Along
The purpose of this new input value is to allow the with mentioned application and due to vast
system data to reach or maintain the reference application of this controller on motor speed, the
value. Unlike other simple control operations, the automated guided vehicle (AGV) is also hired the
PID controller can adjust the input value based on PID routine to correct tracking, power, speed,
historical data and the occurrence rate of the robustness, steering and a lot of other parameters
difference, which can make the system more that have uncertainty. The PID control equation
54
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

involves three separate parameters; the predetermined value from the past average error
Proportional, the Integral and Derivative terms. value can be defined as this term. [3].
AS the PID name shows, to tune this controller，
°C
three parameters of Scale (term P), Integral (Term
14
I), Differential (term D) as shown in equation (1) 13
12
11
are involved. 10

(1)

Times
Equation 1 is the general equation for PID 1 2 3 4 5
(sec)
controller which Kp is proportional, gain, Kd is
Figure (2). The I controller
derivative gain and Ki is integrator gain. PID is
named after its three correction algorithms. All As brief, in Integral only mode, the controller
three algorithms use addition to adjust the value multiplies the integral of the error (accumulation
being controlled. In fact, most of these additional of error or part under the error curve) by the
operations become subtraction operations because Integral Gain (Ki) to get the controller output.
the added number is always negative. The three
is just a mathematical term
Differential or Derivative
algorithms are:
meaning rate-of-change. This term used to control
Proportional (Scale) the future, calculate the first derivative of the
error, multiply it by a negative constant D, and
To control the present, the error value is
finally add it to the predetermined value. The
multiplied by a negative constant P (for scale) and
control of this derivative will respond to changes
then, added to the predetermined value. P is only
in the system. The larger the result of the
true when the output of the controller is
derivative, the faster the control system responds
proportional to the system error. This controller
to the output. The D- parameters help reduce
output change is proportional to the deviation of
short-term changes in the controller. Some
the input controller [3].
practically slow systems may not require D
(Subtract PV from SP)
parameters. In more specialized terms, a PID
SP + Error
Controller controller can be called a filter in a frequency
(Multiply Error by Kp)
Output
domain system. This is useful in calculating
whether it will eventually reach a stable result. If
-
PV the value is not selected properly, the input value
of the control system will oscillate repeatedly,
which may cause the system to never reach the
pre-set value [3].
Figure (1). The P controller
Integral In Derivative Only mode, the controller simply
To control the past, the error value is the sum of multiplies the rate of change of the error at that
the errors over a period. Afterwards, multiplied by instance (slope of the error curve) by the
a negative constant I and afterwards added to the Derivative Gain (KD) to get the controller output.
predetermined value. In the other words, the In this paper first, the Different Controller Types
average error of the output of the system and the with the effect on step response are described in
Section II. Then, the PID Tuning method is
 55
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

reviewed in Section III. In section iv, the AGV proportionality should be determined according to
PID Tune methods are discussed and Finally, the the specific situation. The proportionality is too
paper has concluded in section V with the large, and the control effect is too weak, which is
controller performance and PID methods used not conducive to the system to overcome the
AGV systems. disturbance. The remaining margin is too large,
and the control quality is poor. If the control effect
°C Slope of the curve is too strong, it will easily lead to the deterioration
at time = 1 sec
14 of system stability and trigger oscillation.
13
12
11
10
12 Step response(Ref) P
Ref
10

Times 8
(sec) P controller
1 2 3 4 5
6
Figure (3). The D controller

1- Different Controller Types 4

Although different types of controllers have 2

different structures and principles, there are only

three basic control laws: proportional (P) control, 0 Time( Sec)

0 1 2 3 4 5 6 7 8 9 10
integral (I) control, and derivative (D) control
(Figure (4)). Figure (5). The P controller Vs Step response

P
PI
Error PD Actual For controlled objects with sensitive response and
Signal PID Outpu
Desired
Output t strong amplification ability, in order to improve
r
the stability of the system, the proportionality
+ e u y
Controller Plant should be slightly larger; for controlled objects
- with slow response and weak amplification
ability, the proportionality can be selected to be
Figure (4). The controller Types and structure smaller. Increasing the sensitivity of the entire
system can also reduce the residuals accordingly.
As shown in Figure (4), these kinds of control
In simpler words, the proportional mode duty is to
laws can be used individually, but more occasions
multiply the Error by the Proportional Gain (Kp)
are used in combination. Such as proportional (P)
to get the controller output and the Proportional
control, proportional-integral (PI) control,
Gain is the setting that can tune to get our desired
proportional-integral-derivative (PID) control, etc.
performance from just the P controller. Simple
which are described as following
proportional control is suitable for occasions with
 Proportional (P) control small disturbances, small lags, small load
Separate proportional control is also called changes, and low requirements, which allow a
"difference control". The change in output is certain margin to exist. The use of proportional
proportional to the deviation of the input control law in industrial production is more
controller. The larger the deviation, the larger the common [4].
output. In practical applications, the size of the
56
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

 Proportional integral (PI) control: cannot overcome the influence of interference in a

The proportional control law is the most basic and timely and effective manner. It is also difficult to
widely used one among the basic control laws. Its stabilize the control system. Therefore, in
biggest advantage is that the control is on time practice, integral control is generally not used
and rapidly. If a deviation occurs, the controller alone, but is combined with proportional control
immediately produces a control effect. However, to form proportional-integral control. In this way,
the disadvantage of not being able to finally the strengths of the two, are made up for each
eliminate the residual limits its use alone. The other, and both the rapid and timely proportional
way to overcome the residual is to add integral control function and the ability of the integral
control to the proportional control. The output of control function to eliminate residuals are
the integral controller is proportional to the achieved. Therefore, proportional integral control
integral of the input deviation over time. can achieve a more ideal process control [4]. A
"Integral" here means "accumulation". The output proportional-integral controller is the most widely
of the integral controller is not only related to the used controller, which is mostly used in liquid
magnitude of the input deviation, but also to the level, pressure and flows control systems in
time that the deviation exists. As long as the industrial production. The introduction of integral
deviation exists, the output will continue to action can eliminate the residual, make up for the
accumulate (the output value becomes larger or shortcomings of pure proportional control, and
smaller), and the accumulation will stop until the obtain better control quality. However, the
deviation is zero. Therefore, integral control can introduction of integral effects will make the
eliminate the residual. The law of integral control system stability worse. For control systems with
is also called the law of no difference control. large inertia lag, it should be avoided as much as
possible.
P
12 I
Step response(Ref) PI controller
Pi  Proportional Derivative (PD) Control
10 Ref
Proportional integral control is not ideal for
8 controlled objects with a time lag. The so-called
P conteroller
6
"time lag" refers to: when the controlled object is
I controller
disturbed, the controlled variable does not change
4
immediately, but there is a time delay, such as the
2 capacity lag. Currently, the proportional-integral
control appears sluggish and untimely. For this
0

0 1 2 3 4 5 6 7 8 9 10 reason, people assume that the corresponding

control action can be made according to the
Figure (6). The P, PI, I controller Vs Step response
changing trend of the deviation. by having the
The integration time characterizes the strength of experienced operators, change the valve opening
the integral control effect. The smaller the degree (proportional effect) according to the size
integration time, the stronger the control effect; of the deviation, and one can predict the situation
otherwise, the weaker the control effect. that will occur according to the speed of the
Although, integral control can eliminate the deviation change. This is the law of differential
residual, it has the disadvantage of untimely control with "advanced" control. The output of the
control. Because the accumulation of the integral differential controller depends on the speed at
output is gradual, the control effect produced by it which the input deviation changes [2].
always lags the change of the deviation, and it
 57
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

eliminate residual capacity, and the advanced

P
12
P conteroller
D
PD
control function of the differential action [5].
Step response(Ref) Ref
10
P
12 I
8 Step response(Ref) PI controller Pi
PD conteroller 10 Ref
6

8
4 P conteroller
D controller 6
I controller
2

4
0
2
0 1 2 3 4 5 6 7 8 9 10

Figure (7). The P, D, PD Vs Step response 0

0 1 2 3 4 5 6 7 8 9 10

The differential output is only related to the rate Figure (8). The P, PI, PID controller Vs. Step response

of the change of the deviation and has nothing to When the deviation step appears, the derivative
do with the magnitude of the deviation and the immediately and greatly moves to suppress this
existence of the deviation. If the deviation is a jump in deviation. At the same time, the
fixed value, no matter how big it is, if it does not proportion also plays a role in eliminating the
change, the output change must be zero, and the deviation and reducing the deviation range.
controller has no control effect. The greater the Because the proportion is a control law that lasts
differential time, the longer the differential output and plays a major role, therefore It can make the
is maintained, and therefore the stronger the system more stable; the integral effect slowly
differential effect; otherwise, the weaker it is. overcomes the residual. If the three control
When the derivative time is 0, there is no parameters are properly selected, the advantages
derivative control effect. Similarly, the selection of the three control laws can be fully utilized to
of the differential time needs to be determined obtain a better control effect.
according to the actual situation. The
characteristics of differential control are fast 2- PID Tuning
action, advanced adjustment function, which can Parameter tuning of the PID controller is the core
effectively improve the control quality of the content of control system design. It is to
controlled object with a large time lag; but it determine the size of the proportional coefficient,
cannot eliminate the residual, especially for integration time and derivative time of the PID
constant deviation input, there is no control at all. controller according to the characteristics of the
Therefore, the differential control law cannot be controlled process. There are many methods for
used alone. Especially, for objects with large tuning PID controller parameters, and they can be
capacity lag, the magnitude of dynamic deviation summarized into two categories [6]:
can be reduced, control time is saved, and control
quality is significantly improved. The first is the theoretical calculation

 PID control The first is the theoretical calculation method. It is

It combines the strengths of the three mainly based on the mathematical model of the
P(Proportional), I(Integral), D(Derivational). Then system to determine the controller parameters
in this controller not only the proportional action through theoretical calculations. The calculation
is timely and rapid, but also the integral action to data obtained by this method may not be directly
58
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

used but must be adjusted and modified through parameters can be bypassed using a few tuning
actual engineering. rules. By name, there are two most common
methods to determine and tune PID. One is the
The second is the engineering setting method
Ziegler-Nichols method and the other is Cohen
the second is the engineering setting method, Coon method [23]. The Ziegler-Nichols method is
which mainly depends on engineering experience the most widely used because it's quite simple to
and is directly performed in the test of the control calculate and the Cohen Coon method involves a
system. The method is simple and easy to master. bit more calculations, but it is still easy to use, and
The engineering setting methods of PID controller it is suited to a wide array of processes. Both
parameters mainly include critical ratio method, methods work by generating a response curve
response curve method and attenuation method. from a step input in user control output. Some
Each of these methods has its own characteristics. parameter read from this curve and then using
The common point is to pass the test and then set calculations to determine the P, I and D constants
the controller parameters according to the with the following adjustments.
engineering experience formula. However, no
 Adjustment of the proportionality factor [3]
matter which method is adopted, the controller
parameters need to be adjusted and improved in The adjustment range of the proportional
actual operation. The method for setting the PID coefficient P is generally: 0.1--100. If the gain
controller parameters using this method is as value is 0.1, the output of the PID regulator
follows: (1) first select a sufficiently short changes to a deviation value of one-tenth. If the
sampling period to allow the system to work; (2) gain value is taken as 100, the output of the PID
only add the proportional control link until the regulator changes by a hundred times the
system responds to the input step response with deviation value. The larger the value, the greater
critical oscillation Note down the proportional the gain effect produced by the ratio. During the
amplification factor and critical oscillation period initial adjustment, choose a smaller value, and
at this time; (3) Under certain control degree, then slowly increase it until the system fluctuation
calculate the parameters of the PID controller by is small enough, and afterwards adjust the integral
the formula [7].In order to tune the PID controller, or differential coefficient. Too large P-value will
some parameter is important should be based on cause the system to be unstable and continue to
Table (1)[9]. oscillate; too small P-value will make the system
unresponsive. The appropriate value should make
Table (1): PID Controller Parameters the system have enough sensitivity but not be too
NO Parameter Definition sensitive. The delay of a certain time depends on
1 Rise Time the amount of time the system takes to go
from 10% to 90% of the steady-state, or the integration time [8].
final value
2 Percent the amount that the process variable  Adjustment of integration coefficient
Overshoot overshoots the final value, expressed as a
percentage of the final value
The integration time constant is defined as the
3 Settling time the time required for the process variable to time that the deviation causes the output to
settle to within a certain percentage
increase. The integration time is set to 1 second,
(commonly 5%) of the final value.
4 Steady-State the final difference between the process and the time required for the output to change by
Error variable and set point 100% is 1 second. During the initial adjustment,
Most PID controllers are tuned on-site due to set the integration time longer, and then slowly
machine and process variations. The theoretical decrease until the system is stable [8].
calculations for an initial setting of PID
 59
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

 Adjustment of differential coefficient Table 2: The Ziegler method

The differential value is the rate of change of the NO Type of Controller Kp Ti Td

deviation value. For example, if the input 1 P T/L Infinity 0
deviation value changes linearly, a constant 2 PI 0.9T/L L/0.3 0
adjustment value is superimposed on the regulator 3 PID 1.2T/L 2L 0.5L
output side. Most control systems do not need to According to the algorithms used for the PID
adjust the derivative time. Because only the controllers, the most common way to tune the
system with time lag needs to add this parameter. variables is:
If one adds this parameter to the system, the
control of the system will be affected. If the ideal  Start with a low proportional and no integral
control requirements are still not received through or derivative.
the adjustment of the proportional and integral
 Double the proportional until it begins to
parameters, the differential time can be adjusted.
oscillate, then halve it.
Set this coefficient small during initial tuning and
then, increase it slowly until the system is stable  Start implementing Derivative until it
[8]. For the Ziegler-Nichols method as shown in becomes flat line with the steady step.
figure (9). The main tuning parameters of PID are
Rise time, Settling time, Steady-state and  Slowly implement the integrals until it
overshoot via the adjustment of proportionality becomes stable.
factor, integration coefficient and the differential
Assuming the steady step is like this, that the
coefficient.
rise time is from 0 to 1 and then goes flat. Then
accepted range of rising time can be from 0 to 1.2
c(t) Tangent Line at
inflection point considering there is overshoot. The settling time
k must be within 5% range of the final value for the
system to be optimum. And the aim has to be that
the steady-state error becomes zero or very close
to zero depending on the algorithm used.

L T
t
Very Fast response (good), but poor stability (bad)

Acceptable stability and Medium fastness

Figure 9: Reference Curve Setpoint (a step)

As it is shown in figure 9 reference curve, the S-

shaped reaction curve can be designated by two
constants, delay time L and time constant T,
which are determined by drawing a tangent line at Very Stable (good) but poor response (bad)

the inflection point of the curve and finding the

intersections of the tangent line with the time axis
and the steady-state level line. Then, the tuning
method for controlling should be as Table (2). 0 2 4 6 8 10 12 14 16 18 20

This is one of the methods to do the trick. [10] Figure 10: Sample PID Curve

As it shows in figure (10) the middle curve is the

most acceptable form of the curve as it has
 60
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

acceptable stability and medium fastness hence a the expected value of the internal ring hence, it
good compromise for the whole system. Based on establishes a relationship between the kinematics
the main aim of this paper on AGV robot and dynamics of the whole system directly.
platform, PID tuning is very important, such as,
Each AGV or Machine has its own set of function, In their research, they did some test tracking
hence, in order to accomplish the exact set of under MATLAB/SIMULINK environment. Their
workload more efficiently one needs to actually simulation result verifies that AGV can trace the
tune the PID parameters accordingly so that there given geometric path more rapidly and correctly
cannot be an intolerable amount of energy loss, compared to other systems out there. They used
inefficiency rate, lag, lack of robustness etc. The PID here to have different control modes for their
parameter before and after PID Tune are shown in AGV tuning, which are mentioned before as
Table (3). hierarchical control and Lyapunov direct method.
Having different modes on PID is the novel thing
Table 3: the PID Parameter and performance they have achieved. Jorge Villagra et al [14] in
Before After Tuning [Ziegler their research have shown a detailed comparison
No Parameters
Tuning Nielsen Method]
between different control methods for AGV
1 Rise Time Variable Decrease
2 Overshoot Exists Reduced robust path tracking where they ensured low jerk
Settling variations when docking AGVs in industrial
3 Slow Reduced
Time environments. Their proposed control system
Steady State Low focused on both robustness and performance at
4 Little to No error
Error Error
one side and easy configuration capabilities of the
Then, based on the PID parameter, the effects of controller on the other. Two control techniques
increasing each controller parameters Kp, Ki, and namely, fuzzy and vector pursuit and a data-
Kd can be summarized as shown in Table (4): [10] driven model for feedback control combined with
differential flatness control were compared
Table 4: the PID Parameter and performance
Steady
together using exhaustive Monte-Carlo
Rise Settling Simulation. Their research simulations concluded
No Response Overshoot State
Time Time
Error that a model-based controller has excellent
Minor
1 Kp Decrease Increase Decrease tracking quality under a high degree of
Change uncertainty while a non-model-based controller
2 Ki Decrease Increase Increase Eliminate
(fuzzy and vector pursuit) has an acceptable level
Minor Minor
3 Kd Decrease Decrease
Change Change of transportability regardless of high amplitude.
They added that having an Intelligent PID (I-PID)
3- AGV PID Tune method
controller compensates for skidding, slipping and
Hauling Zhang et al [13] in their paper have actuators dynamics without the need of physical
aimed at AGV motion model of two-wheeled parameters.
differential drive control modes. They have
divided this mode into two other modes. They are Young Jin-Lee et al [15] in their research have
defined as double closed-loop control based on designed an experimental AGV system for
Lyapunov direct method and sliding mode control automated container terminal with intelligent PID
technology and speed control based on the PID (I-PID) controllers. They aimed for a more
method through hierarchical control thought. The effective system using the idea of the biological
double closed-loop control structure has two rings immune system which is cell-mediated immunity.
or loops and they investigated that the virtual This system intelligently adjusts its parameters
speed vector output by the external controller is just like the self-regulating mechanism of the
 61
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

immune system. The researchers used such type reach for this field. W.S. Wijesoma et al [17] in
of PID to enhance control and let the AGV learn their research have portrayed stable fuzzy
as it goes. The parameters for the designed controllers' designs for an AGV. In their paper,
controller have been optimized using HIA or the they have stated 3 controllers namely, Sliding
humoral immune algorithm. The proposed system Mode Controllers (SMCs), Fuzzy SMCs, and
with this new PID controller was put on par with Fuzzy PD controllers. Experiments on all 3 were
existing classical PID and NNPID systems carried out by giving each controller a steering
respectively and was thoroughly compared with angle profile having a trapezoidal steer angle
each other using various constant loads having starting from zero. Results showed that all the
both maximum displacement errors and average three controllers' performance were relatively
displacement errors. The result showed a 17% and similar to each other with nothing much of a
16% and, a 27% and 20% improvement with the difference other than the fact that fuzzy controller
systems respectively. It was also concluded that had slightly fewer tracking errors meaning higher
the proposed controller reduces the energy tracking accuracy. The paper concludes the same
required for controlling the steering and speed and adds that it gives more degrees of freedom in
settling. The novel thing about this is the usage of performance tuning while using SMCs and
a self-regulating hence self-learning mechanism in inclusion of heuristic control knowledge in simple
PID. Jin-Woo Lee et al [16] experimented in their state variables, whereas using fuzzy control
paper for AGV Steering Control and Identification permits treatment of linear combinations of state
using Vision from a color CCD camera. In their variables. Laurens Jacobs et al [18] in their
designed AGV they set up the camera on the front research have proposed an LPV (linear parameter
and grabbed the 3D image of the designated path varying) control design strategy for trajectory
of the vehicle. It was then converted into a 2D tracking using a varying linear parameter control
image from which the AGV processes its driving i.e. a non-linear kinematic control. They
conditions. Using the filtering process, they drew approximated the tracking problem for the
out the ratio of the color of the path which is nonlinear model by a tracking problem for a linear
converted into a simple X-Y Co-ordinate system parameter-varying model. Their controller design
and through that the steering angle is calculated. method had a closed-loop characteristic. The
This whole process takes about 20 frames/sec and research shows the pristine usage of varying linear
thus the driving state is recognized and is sent to control over the closed-loop designed PID
the AGV's microprocessor. Neural network controller. The control software of the robot was
identifier has been used to eliminate the time programmed in an OROCOS environment which
delay issue from the vision system. Hence, the could communicate with the sensors and motors
modified PID controller has been designed using through Ether CAT protocol at a rate of 100HZ.
the neural network algorithm and applied to the They used LCToolbox from MATLAB to solve
steering control of the AGV. The conclusion was the control problem. They validated this controller
made that the result of this attempt did not require on a prototype of an autonomous mobile platform
any consideration of line detection, labelling in a greenhouse. The results show that the
algorithm etc rather only the vision sensor was tracking performance is centimeter precise thus
enough to ensure faster guideline detection using achieved a satisfying precision.
only image processing. For this particular
research, the researchers used PID but only to Yu Dianyong el al [19] in their paper has adopted
work with the help of visual identification. Having a self-adjusting fuzzy control method that can
a different sense to work with PID is a new idea to make AGV trace the reference path at a definite
62
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

speed but at a higher precision and the speed of Neuro-Fuzzy Inference System (the combination
the vehicle tend to the tangent direction of its of fuzzy logic controller and neural network
trajectory. Their designed guideway of the AGV system), the small oscillations in the system
used a photoelectric guideway and the usage of response can also be reduced.Hong-Jie tang et al
such a mechanism with PID controller is [21] stated in their paper that their proposed
unhackneyed. In their simulation, they used an system based on their designed PID controller has
AGV with the speed of 0.3m/s and, the results good tracking performance in a straight line as
showed higher precision and rapid response from well as it as strong adaptability and high
the system. The experimental results were much robustness while tracking circular path. In their
closed to the actual simulation also. Even when paper, they used two DC brushless motors and
having the bigger deviations of the parameters the hence their AGV goes with two differential
vehicle was able to return to its original position speeds. The vehicle has a magnetic navigation
in a shorter time. Lastly, the researchers have sensor which both detects declination or
concluded that their method has a higher control inclination of the relative tape as well as the tape
precision and better dynamic performance when it as its directed path. They used MATLAB
comes to controlling AGV. Because, they do this Simulink platform to simulate the mathematical
through a fuzzy controller. They further conclude model of the design. Their PID control algorithm
that using photoelectric guideway has better is an incremental PID control expression which
advantages of easy construction and easy does not require accumulation while calculation;
assigning of variables. the incrementation corresponds to the change of
the position of the actuator and this is the only
Priyam Parikh et al [20] have researched on their value outputted by the computer hence the
paper about implementing fuzzy logic controller influence of error when occurred is small. The
and PID controller to a DC Motor to achieve algorithm with collaboration with the sensor only
constant RPM. In their project, they have calculates the deviation from the path which is the
designed an AGV in which they implemented a tape. The results showed that by using magnetic
PID controller technique using the Ziegler Nichols navigation PID control overshoot is small and this
algorithm and Mamdani technique for the fuzzy control method can effectively correct AGV
logic controller. Then, they programmed and deviation and track the given path.K.R.S.
managed delays using a microcontroller and Kodagoda et al [22] in their paper have proposed
applied the AGV motor to it via Microcontroller. a control structure for uncoupled lateral and
Finally, they have interfaced a MATLAB longitudinal control for an AGV. They achieved
Simulink. Their proposed system after lateral control using fuzzy controllers and
experimentation concluded the result that the longitudinal control using fuzzy drive controller
error, overshot and rise time in PID controller is and fuzzy braking controller. A supervisory
much lesser for the fuzzy logic controller. The controller which implemented the heuristics that
whole system is more stable and more accurate the AGV will only break if and only if the actual
due to the fuzzy logic controller. Finally, the speed was greater the requested speed was used as
researchers conclude alongside their experiment's the switch between the different controller of the
result that using their method of implementing longitudinal control. Their experiment shows that
fuzzy logic controller and PID controller lets the the fuzzy logic controller schemes outperform the
AGV achieve constant speed in 0.8 seconds with traditional PID schemes when it comes to tracking
only 1% overshoot and they state that in the future accuracy and steady-state errors. Furthermore, the
using ANFIS controller which stands for Adaptive results showed that the fuzzy logic controllers are
 63
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

robust to load changes, coupling effects etc. When

coupled DC motor to the designed controller
showed smooth speed tracking performance with AGV robots as the controlling system need to
no jerk. have a controlling loop to follow its task and
mission in various goals. The comparison between
4- Conclusion PID and aim in different researchers are shown in
To control the systems, generally different Table (6).
combinations of three terms of P, I, and D are Table (6): Comparison Between Existing PID Controllers
used which intended to the controller like: P Other

Controller

Differential
Two Wheel
Controller

Controller
(Occasionally used), PI (Most frequently used) Controllers

Fuzzy
Specificatio

i-PID
No. based on
PD (rare but can be useful for controlling in some ns novel
algorithms
applications like servomotors etc) and PID. So,
Rapid
between them, the PID due to its better ability is 1
Path
Accurate
High
Tracking of
Satisfying
Tracking Accuracy precision
of more demand. Even some researcher believes geometry
Jerk
that using the PID under different control schemes 2 Managemen
Low to no
x x x
Jerk
to ease their work and get the desired result. The t
Depends on
mentioned controller has some merit and demerits 3
Uncertainty
High
Based on
Not Good the
Tolerance reference
and applications which are shown in Table (5) algorithm
Depends on
briefly. 4
Self-
Good Very Good
Can be
the
Adjustment better
algorithm
Table (5): The PI, PD and PID Controller Very Good
Better for
Controller with
NO Advantages Disadvantages Application 5 Robustness load Very good Not good
type incremental
changes
algorithm
moving Process variables

Speed control of motors,

control applications use
Stability and Maximum
peak overshoot is better

External
generators, turbines etc
Zero steady state error,

Majority of the speed

PI control action. Eg.

controller for slow
than Integral only

We cannot use PI

6 Data No No Yes Not always

Controller

required
1 PI
7 Overshoot Small x x Small
Outperform
Depends on
Steady-State Fewer s traditional
8 Parameter x
Error Errors PID
Changes
used for fast moving process variables like
flow, pressure, highly affected by external

Controllers
noise. A PD controller is analogous to a
Steady state error is not zero, cannot be

PD controllers are most widely used in

amplifies high freq noise if present.
Stability increased, Maximum peak
overshoot decreased, Settling time

Very
HPF along with amplifier. This it

Definite
Good with Smooth
temp control applications.

Speed Speed with

9 energy with Smooth
Tracking higher
saving differential
decreases

precision
2 PD speeds

As the Table (6) shows, the reviews on previous

papers show, most of the researchers are used some
form of PID with some form of tuning method
based on their system requirements. Huaqiang
cost is high, Complexity and tuning,

Zhang et al [13] have used PID to have differential

can be used for controlling both fast
peak overshoot, moderate stability,
steady state error is zero, moderate

for almost every application by

and slow process variables.

drive control and he used a tuning method based on

design complexity

Lyapunov Direct Method of control to achieve their

proper tuning

3 PID
goal. Jorge Villagra et al [14] have used PID for
their AGV's robust path tracking. They used an
intelligent PID controller alongside a fuzzy
controller which is a novel way of implementation
– the combination of two PID controllers. Young
 64
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

Jin-Lee et al [15] takes it to a step further because [5] http://m.eadianqi.com/view.php?aid=4090

they use intelligent PID controllers and have tuned [6] https://wenku.baidu.com/view/b5c17fb9d4d8d15abe23
a PID algorithm based on the human immune 4ef9.html?re=view

system called the humoral immune algorithm. [7] http://m.eadianqi.com/view.php?aid=4615

While till now all the researchers have used some [8] http://ask.zol.com.cn/x/4170860.html
[9] Devdas Shetty; Richard A. Kolk (17 June 2010).
form of PID tuning, the next group of researchers
Mechatronics System Design. Cengage Learning. pp.
namely, Jin-Woo Lee et al [16] have not used any 330–. ISBN 1-4390-6198-X
major form of PID tuning. Rather they used visual [10] http://faculty.mercer.edu/jenkins_he/documents/Tuning
images and a neural network does the trick, which forPIDControllers.pdf
is a very novel thing. Then, again Priyam Parikh et [11] Qian, Zhikang et al. “Design of a Low Power
al [20] have researched with fuzzy logic controller Consumption Control System of Permanent Magnet
and PID controller. But they have used the Synchronous Motor for Automated Guided Vehicle.”
2019 22nd International Conference on Electrical
traditional Ziegler Nichols algorithm to tune their
Machines and Systems (ICEMS) (2019): 1-5.
fuzzy PID, so nothing new is presented here in
[12] https://arduinoplusplus.wordpress.com/2017/06/21/pid-
terms of tuning other than implementation. Hong- control-experiment-tuning-the-controller/
Jie tang et al [21] have used a PID tuned with [13] Zhang, Huaqiang, Dandan Wu, and Tong Yao.
incremental PID control algorithm which is also "Research on AGV trajectory tracking control based on
relatively new. So, most of the researchers who double closed-loop and PID control." Journal of
have been mentioned here have done something Physics: Conference Series. Vol. 1074. No. 1. IOP
Publishing, 2018.
novel to tune their PID controller in their systems.
[14] Villagra, Jorge, and David Herrero-Pérez. "A
The most noteworthy novel things here are using a
comparison of control techniques for robust docking
combination of two or more PID controllers, using maneuvers of an AGV." IEEE Transactions on Control
different control techniques and tuning methods for Systems Technology 20.4 (2011): 1116-1123.
each of the controllers and then combining them, [15] Lee, Young Jin, et al. "Driving control of an AGV for
using visual aid and intelligent control network to an automated container terminal using an immunized
reduce controller tuning, using intelligent PID PID controller based on cell-mediated immunity."
Artificial Life and Robotics 9.2 (2005): 90-95.
controllers and finally even when using traditional
[16] Lee, Jin-Woo, et al. "A study for AGV steering control
tuning method, they have implemented them in a
and identification using vision system." ISIE 2001.
non-traditional way like in a fuzzy logic controller. 2001 IEEE International Symposium on Industrial
This work can be implemented on various robotic Electronics Proceedings (Cat. No. 01TH8570). Vol. 3.
applications [24-28] especially odor tracking [29- IEEE, 2001.
30] and pipe inspection robot [31-34]. The future [17] Wijesoma, Wijerupage Sardha, and K. R. S. Kodagoda.
work will be focus on effect of various PID tune on "Design of stable fuzzy controllers for an AGV."
Proceedings. 2000 IEEE/RSJ International Conference
AGV dynamic model.
on Intelligent Robots and Systems (IROS 2000)(Cat.
No. 00CH37113). Vol. 2. IEEE, 2000.
5- References
[18] Jacobs, Laurens, et al. "Trajectory tracking of AGVs by
[1] https://www.controleng.com/articles/fixing-pid/ [visted linear parameter-varying control: a case study." IFAC-
on 2/2/20] PapersOnLine 51.26 (2018): 43-48.
[2] https://www.techbriefs.com/component/content/article/t [19] Dianyong, Yu, and Xue Hui. "Application of fuzzy
b/features/articles/20013 control method to AGV." IEEE International
[3] https://baike.baidu.com/item/PID%E6%8E%A7%E5% Conference on Robotics, Intelligent Systems and Signal
88%B6%E5%99%A8/1888589?fr=aladdin#ref_[1]_104 Processing, 2003. Proceedings. 2003. Vol. 2. IEEE,
644 [visited on 2/2/20] 2003.
[4] https://zhidao.baidu.com/question/18668630767727744 [20] Parikh, Priyam, et al. "Implementing Fuzzy Logic
67.html Controller and PID Controller to a DC Encoder Motor–
 65
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066

“A case of an Automated Guided Vehicle”." Procedia [32] Moshayedi, Ata Jahangir, Damayanti C. Gharpure, and
Manufacturing 20 (2018): 219-226. Arvind D. Shaligram. "Design and Development Of
[21] TANG, Hong-jie, et al. "PID Control of Magnetic MOKHTAR Wind Tracker." International Journal of
Navigation Differential AGV Trajectory." DEStech Soft Computing and Engineering (IJSCE) (2013): 167-
Transactions on Engineering and Technology Research 173.
apop (2017). [33] A. J. Moshayedi and D. C. Gharpure, “Development of
[22] Kodagoda, K. R. S., W. Sardha Wijesoma, and Eam Position Monitoring system for studying performance
Khwang Teoh. "Fuzzy speed and steering control of an of wind tracking algorithms,” in Robotics; Proceedings
AGV." IEEE Transactions on control systems of ROBOTIK 2012; 7th German Conference on, 2012,
technology 10.1 (2002): 112-120 pp. 1–4.
[23] https://www.youtube.com/watch?v=zkWUFOajrVk [34] A. J. Moshayedi, A. Abbasi, L. Liao and S. Li, "Path
[visited 4th april, 2020] planning and trajectroy tracking of a mobile robot using
[24] O. T. Esfahani and A. J. Moshayedi, “Accuracy of the bio-inspired optimization algorithms and PID control",
2019 IEEE International Conference on Computational
Positioning Systems for the Tracking of Alzheimer’s
Intelligence and Virtual Environments for Measurement
Patients-A Review,” Int. J. Appl. Electron. Phys.
Systems and Applications (CIVEMSA), Tianjin, China,
Robot., vol. 2, no. 2, pp. 10–16, 2014.
2019, pp. 1-6.
[25] M., Bozorg, H., Jalili, S.A. Eftekhari, “Robustness of
autonomous underwater vehicle control in variable
working conditions”, Journal of Marine Science and
Technology, Vol. 12, No. 4, pp 232-239,November,
2007.
[26] M. Bozorg, S.A. Eftekhari, H. Jalili, “Robust control of
autonomous underwater vehicles in uncertain operating
conditions”, 5th IFAC/EURON Symposium on
Intelligent Autonomous Vehicles, Instituto Superior
Técnico, Lisboa, Portugal, July 5-7, 2004.
[27] M. Bozorg, S.A. Eftekhari, H. Jalili, “Robustness of
Underwater Vehicle Control under Parameter
Perturbations”, 8th International and 12th Annual
Mechanical Engineering Conference, Tarbiat Modares
University, I.R. Iran, May 2004.
[28] Moahamad Khaje khabaz, S. Ali Eftekhari, Mohamad
Hashemian, Davood Toghraie, “Optimal vibration
control of multi-layer micro-beams actuated by
piezoelectric layer based on modified couple stress and
surface stress elasticity theories”, Physica A: Statistical
Mechanics and its Applications, Volume 546, 15 May
2020, 123998
[29] A J. Moshayedi and D. C. Gharpure, “Evaluation of bio
inspired Mokhtar: Odor localization system,” in 2017
18th International Carpathian Control Conference,
ICCC 2017, 2017.
[30] A.J. Moshayedi, D. C. Gharpure, and A. D. Shaligram,
“Design and Development Of MOKHTAR Wind
Tracker.”
[31] A.J. Moshayedi, S. S. Fard, L. Liao, and S. A.
Eftekhari, “Design and Development of Pipe Inspection
Robot Meant for Resizable Pipe Lines,” Int. J. Robot.
Control, vol. 2, no. 1, p. 25, 2019.
66
A.J. Moshayedi et al. /Journal of Simulation & Analysis of Novel Technologies in Mechanical Engineering 12 (2019) 0053~0066