PID and Fuzzy Logic Controller for ROV depth

control
Adrian Wai Chen Leong 1, Quek Shiung Liew 2, Wei Yin Yong 3, Gabriel John Ting 4
Department of Mechatronics
Engineering Universiti Teknikal
Malaysia Melaka (UTeM) Durian
Tunggal, Melaka.
Email: (B011510100/B011510072/B011510098/B011510099)@student.utem.edu.my

Abstract – Remotely operated vehicle (AUV) is an unoccupied Two controllers, PID controller and Fuzzy
underwater robot that connected to a ship or terminal via cables. In
Logic Controller (FLC) are been compared to get the
order to obtain a better performance on its depth control,
proportional-integral derivative (PID) controller and fuzzy logic best result. PID controller become popular and gained
controller (FLC) are designed to analyze their performance. The widespread in industrial acceptance due to the benefits
objectives of this paper is to design a PID and FLC controller on the of this controller such as simplicity, robustness and
ROV depth control, and to simulate and analyze the performance of
wide applicability. P controller is mainly used to
the PID and FLC controller on the ROV depth control model.
Firstly, block diagrams based on PID and FLC controller on the decrease the steady-state error of the system but unable
model are designed, followed by tuning to obtain the results. PID to eliminate the steady-state error of the system. P-I
controller with initial and after tuning is done, goes by the Mandani controller manage to eliminate the steady-state error
FLC controller by tuning it until a satisfactory response is obtained.
and tracking small change error but lead to high
The performance of the ROV without controller, PID controller
after tuning and Mandani FLC controller are compared based on overshoot of the system. P-D controller able to
the parameters such as settling time, slew rate, rise time and increase the stability of the system as it handles large
overshoot value. The results show that FLC controller has better changes well with minimal overshoot but have poor
performance than PID controller due to its lower rise time, higher
performance on tracking small changes or errors. For
slew rate, and lower percentage of overshoot.
PID controller, it has the optimum control dynamics
Keywords – ROV, Depth control, tuning, PID controller, FLC since it has 3 parameter that including zero steady-state
controller error, fast response, no oscillations and higher stability.

I. INTRODUCTION FLC is widely used in machine control. The

term "fuzzy" refers to the fact that the logic involved
can deal with concepts that cannot be expressed as the
Underwater is still not completely explored fully and "true" or "false" but rather as "partially true". The
not easily accessible by humans. Underwater vehicle blocks of a FLC is shown in Figure 1.1. Each of the
has been developed and improved for decades in block has its own characteristic. The first block inside
various designs and new innovations. There two major the controller is fuzzification which used to converts
type of underwater vehicle such as remotely operated every input data to degrees of membership by a lookup
vehicle (ROV) and autonomous underwater vehicle in one or more membership functions. The
(AUV). ROV is an unoccupied underwater robot that fuzzification block will then compare the input data
connected to a ship or terminal via cables. The cables with the conditions of the rules to determine the fitness
transmit signals and commands for controlling and of each rule into the particular input instance [1].
feedbacks of the ROV allowing remote navigation.
AUV is also known as unmanned underwater vehicle
which is very suitable for survey missions. AUV is
operated with little to no operator intervention. ROV
are controlled by onboard sensors such as sonar, depth
sensor and inertia measurement unit (IMU). ROV has
thrusters, propellers or wings to control its motions. In
term of motion control of the ROV, depth control is
very important. To control the ROV’s depth, a
suitable controller is needed.
Fig. 1. Block Diagram of Fuzzy Logic Controller (FLC) [1]
 II METHODOLOGY

This section discusses about the procedure or

method to tune the PID and FLC controller for the ROV
depth control model.[2] In this paper, the transfer function
state space technique yields for ROV is given in equation
(1) below:

0.4871𝑠 2 + 1.37𝑠 + 67.31

𝑇𝐹 = (1)
𝑠3 + 4.911𝑠 2 + 8.309𝑠 + 76.09

First, the PID controller is designed based on the

transfer function that stated in equation (1). Then, Fuzzy
Logic Controller (FLC) is designed with following criteria
which are MISO system which is 2 inputs and 1 output,
Mamdani inference engine and design 5x5 membership
functions. The construction of block simulation will be
done by using Simulink. All the result will be compared
and analysed.
Fig. 3. PID tuning tool
Figure 2 shows the design of PID controller and
Fuzzy Logic controller for the model of ROV in the
Next, fuzzy command is executed in command
Simulink. Initially, the control system is designed with a
window of MATLAB to launch the Fuzzy Logic Designer
step input, PID controller block, transfer function as a plant
as shown in Figure 4.
of ROV and a scope which act as output. The step input is
set with 1s step time, start from 0 and 1 final value.

Fig. 2. The design of PID controller and Fuzzy Logic

controller in the Simulink
Fig. 4. Fuzzy Logic Designer

The tuning of PID controller is done by using the From the Figure 4, there are 2 input and 1 output
auto-tuning function of Simulink which is shown in for the Fuzzy Logic. For membership functions, 2 inputs
Figure 3 After the toolbox is linearize the plant for auto- which are error and change-in-error are represented in the
tuning process, then, the response time and transient form of linguistic variables. Each input has five linguistic
behaviour can be edited to tune the desired and best variables. Those five linguistic variables are negative big
output for the plant. The result of tuning parameter is (BN), negative small (SN), zero (ZO), positive small (SP)
generated and the simulation is run for 10 seconds. and positive big (BP). The terms such as BN, SN, SP, BP
Finally, the output result is recorded and analysed. and ZO are characterized via triangular-shaped membership
functions which is shown in Figure 5. Figure 5 also shows
that there are 10 parameters will be estimated by try and
error method for each input and output. Input 1 is inserted
with a range of [-10, 10] while input 2 is inserted with a
range of [-3 3]. Output is inserted with a range of [-2 2]. name is then inserted into the block of FLC in the
Simulink. The gains that are inserted in the control system
is act as the scaling gain for the output. Hence, the scaling
gains is used to tune the FLC. The simulation is run for 10
seconds. Finally, the result is recorded and analysed.

Fig. 6. The design of Fuzzy Logic controller in the Simulink

III RESULT AND DISCUSSION

A transfer function without controller is shown in

Figure 7. The performance of the ROV implemented
using different controllers is as shown in Figure 8.

Fig. 5. Membership function of error, change of error and

output

For rule-base, it is act as a correspondence

between the particular membership functions of inputs
and output. The rule-base is presented in a table form
which is shown in Table 1. However, it is same as the
rule-base that presented in If-Then form. There are 25
control rules in total.

Table 1: Fuzzy logic controller rule-base for the transfer function

BN SN Z SP BP Fig. 7. Transfer Function without Controller

BN BN BN BN SN Z
SN BN BN SN Z SP Fig. 8. The Performance of The Transfer Function using Different
Z BN SN Z SP BP Controllers
SP SN Z SP BP BP
BP Z SP BP BP BP Figure 8 shows the performance of the transfer
function using different controllers. The added controllers
After that, the file is saved in a file and export into are compared with the desired step function. From Figure 7,
the workspace. The control system with Fuzzy Logic it is found that the performance of the original system is far
controller is designed as shown in Figure 6. The file away from the desired steady-state value. Thus, a few
different types of controllers are added to the system to system reaches the steady-state value faster than the other
determine the suitable controller to be used in the system. controllers.
From Figure 2, it is found that the result showed for PID
controller is overshoot while Mamdani FLC is undershoot. IV CONCLUSION
Both of the controller will be compared in terms of
overshoot and settling time. In conclusion, PID and FLC controllers can be
designed to tune the ROV-Depth control model, for
Table 2: Comparison of Transition Parameters of the Transfer Function obtaining a better performance in terms of their
Parameters Without PID Mandani FLC parameters, which are rise time, slew rate, percentage of
Controller Controller overshoot and settling time. The smaller the parameters
High (mV) -162.70 1000.0 999.80 value, the better the performance of the controller. In this
Low (mV) -3798.0 5.988 5.024 paper, FLC controller shows better performance than PID
Amplitude (mV) 3635.0 994.0 994.80 with tuning. However, FLC controller has a longer
Rise Time (ms) 93.499 13.717 1.142 settling time compared to PID controller which means the
Slew rate (V/s) 37.258 57.974 696.787 system reaches the steady-state value slower than PID
controller. Overall, FLC controller shows better
performances than the other controllers due to its lower
Table 2 shows the comparison among the rise time, higher slew rate, and lower percentage of
controllers in terms of the transition parameter. Rise time overshoot.
refers to the time taken for the response to rise from 10% to
REFERENCE
90% of the steady-state response. Hence, the shorter rise
time indicates faster response to reach the step response. [1] Jantzen, Jan. "Design of fuzzy controllers." Technical University of
The Mamdani FLC has the lowest rise time among the Denmark, Department of Automation, Bldg 326 (1998): 362-367.
three. Therefore, Mamdani FLC is the most suitable
controller to be used in terms of rise time. Slew rate [2] Aras, M.S.M, S.S. Abdullah, Development and Modelling of underwater
Remotely Operated Vehicle using System identification for depth
indicates the change of voltage per unit time. Higher value
control, 2013
of slew rate indicates better rise time of the response. Thus,
Mamdani FLC with the highest slew rate has the best
performance.

Table 3: Comparison of Overshoot Parameters of the Transfer Function

Parameters Without PID Mandani FLC
Controller Controller
Preshoot (%) 68.107 0.602 0.505
Overshoot (%) 132.552 19.880 0.311
Undershoot -57.081 1.125 2.637
(%)
Settling Time - 4.0 5.5
(s)

Table 3 refers to the shoot parameters of the

response with the settling time. Overshoot is an
important parameter in evaluating the performance of the
controller. Percent overshoot is the amount of waveform
deviates from the desired output which is calculated in
percentage. A lower percentage of overshoot means the
curve of the response is deviating lesser to the step
response. Thus, Mandani FLC with a lower percentage of
overshoot has the least fluctuation compared to the other
controllers. Settling time refers to the time required for
the system to reach and stay within ±2% of the steady-
state value. The shorter settling time indicates the faster
the system to reach the steady-state value. Thus, PID
controller has the shortest settling time which means the