2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

Design, Development and Fuzzy Logic Based Control of a Remotely

Operated Underwater Vehicle
Joe Johnson, G Madhumitha, Niyas Yousuf, and Shyam Harikrishnan
Department of Mechatronics Engineering, SRM University, India
[email protected], [email protected]

Abstract Underwater marine exploration still remains a A single or dual %.()!
(.()#
%$)'%"(
)
+"5(

mystery. The purpose of this paper is to address the issues in thrusters, enabling the pilot to alter the heading, depth or
design and development of underwater vehicles with obstacle position of the ROV. In this paper, three major aspects related
avoidance and maneuvering assistance for operator in marine to the design of ROV is discussed in detail and is represented
environment using fuzzy logic controller. A symmetrical, cost
as block diagram in the Fig.1.
effective and small sized underwater Remotely Operated Vehicle
(ROV) with three thrusters is designed for testing control The objective of this paper is to design and develop a low cost
algorithm and performance of system. Kinematics of the ROV is and upgradable ROV (for accomplishing complicated tasks in
developed relating speed of rotation of thrusters with  
future) which overcomes navigational complications.
linear and angular acceleration. A comparative study of the
   response for simple if-else logic and with fuzzy logic
Mechanical design is made using SolidWorks software. User
controller for navigation is made and evaluated. A Graphical interface is designed using Python and obstacle avoidance
User Interface (GUI) with live feed is provided to the operator feature is incorporated using Fuzzy logic. Flex sensors are
for wider range of visual observation of underwater used to sense obstacles $
5(
&'%-#).
and interfaced
environment and navigational aid. with microcontroller to avoid obstacles and to achieve better
maneuverability. An inbuilt power source is used to enable
Keywords
ROV; Fuzzy logic controller; Obstacle avoidance; the ROV to come up to the surface in case of tether cut or
computer simulation failure.

INTRODUCTION
II.
MECHANICAL DESIGN
An ROV is a tethered underwater robot which is Design parameters to maintain symmetry of the ROV
controlled remotely from a device usually placed in a boat or which in return affects the buoyancy, propellers to achieve the
ship. Tether is a group of cables that carry electrical power, required thrust and to have low drift coefficient are selected.
control signals, video and data signals back and forth between The hull is made out of acrylic tube to have an aerodynamic
the operator and the ROV. Additional equipment like flow that decreases drag resistance. Such a choice is made to
SONAR, magnetometer, a still camera, a manipulator or have less weight and the structural dimension of the ROV is
cutting arm, water samplers, instruments that measure water finalized to 40 cm length, 30 cm width and 10 cm height, with
clarity, light penetration and temperature are commonly an estimated weight of 3.2 kg with all electromechanical
added to exp$
)
+"5(
&")(
components and circuit accommodated within the ROV itself.
The weight was determined using SolidWorks Model during
the design phase of the ROV by using datasheet of all the
components and weighed after assembly.
A.
CAD Modeling
Mechanical structure is the backbone of the ROV [1] and
is built from components as simple as possible without
compromising any functionality in such a way that the
dynamics of the vehicle is not affected. Transparent acrylic
tube of 3 mm thickness, 300 mm length, and 100 mm diameter
is chosen for making hull because of its strength and ability to
repel water. The circular shape allows for smooth
aerodynamic flow of water. A camera housed inside the
transparent hull would give a clearer and smooth picture. Two
such tubes with different diameters, one for circuit housing
and other for battery housing are used. End Caps are machined
from a thick block of propylene to fit the acrylic tube tightly
and a groove is made to fix the O-rings to seal the tube
Fig. 1. Major Aspects in ROV Design
properly.

978-1-5090-5203-5/16/ $31.00 ©2016 IEEE

 2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

the distance between thruster 2 and 3 (0.15 m). For the thrust
Flex sensor 2 Flex sensor 1 coefficient Kt = 0.35 (from datasheet using P/D ratio 1.4) and
465, the viscous factor of water (1.002 * 10-3 Ns/m2), the thrust
45
$')
.
)
)*)%'
(
+$
.
Thruster 2
* I &A 70; , = (1)
Hence maximum thrust from a single thruster having four
Thruster 3
blades with high angles is,
Thruster 1
* I   ><
' (2)

Flex sensor 3
"')%$
45
))$
.
)

(
+$
.
Flex sensor 4
+ I */ (3)
Fig. 2. Top View of 5(

Model
Maximum linear acceleration of ROV = 2*a = .6313 mm/s2
Holes are drilled on it to accommodate for the screw threads
to pass through. Polypropylene sheets are used in construction Acceleration in all the three translational degrees of freedom
of the ROV5( outer structure in order to give it high strength, surge, sway and heave is as follows,
rigidity, durability and less weight. 1 I 4N I Q&A 7, = O0;; G 0<; PR/ (4)
2I (5)
3 I -N I Q&A 7, = 0:; R/ (6)
Acceleration in all the three rotational degrees of freedom roll,
pitch and yaw is as follows,
5I (7)
8I (8)
9 I Q&A 7, = O0;; H 0<; PR/. (9)
where n1, n2 and n3 are the speed of rotation of the thrusters
T1, T2 and T3 in RPM respectively. Fig.5 gives the Cartesian
parameter representation of the ROV in 3 axes [6] and thruster
Fig. 3. Isometric ,
%
5(

odel position in the physical system.
The sheet was cut out and holes were drilled as per design in
order to accommodate the hull, battery pod and the motors.
The top view and isometric view of CAD model of the ROV
are shown in Fig. 2 and Fig. 3 respectively. The completely
assembled ROV system with components mounted on it is
shown in the Fig.4.

Fig. 5. Cartesian Representation of the ROV.

Rotational matrix about Z axis is given in equation 10,
 9 H  9

Rotational matrix, R =K  9
 9


 M (10)













From equations 4, 5, 6, 7, 8, 9 and 10 generalized kinematic
model is derived as follows,
1 0 ; G 0< ;
2  ; 
  
3  ) <?< = P

   0 ;
Fig. 4. Complete ROV System  5  I
J<?< L  O  6   : (11)
   
8  
B.
System Modeling 9
 
; ; .
0; H 0<
)
4#5

)
#((
%
)


!
4$5

the speed


45
 the diameter (0.1 m) of the propeller, 4"5


 2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

To simulate the movement of the robot [4], [5], a GUI is shows the flowchart for ROV control in traverse mode.
created in MATLAB using the above kinematic equation and Control is accomplished in such a way that previous
path of the ROV can be calculated from the same by giving actuation is dismissed before any new actuation takes place.
thruster speed as input. The simulated path for sample thruster
inputs is shown in the Fig.6.

Fig. 6. GUI for 5(
#*")
ath.

II.
ELECTRICAL DESIGN Fig.8. Flowchart of ROV Control in Traverse Mode

A robust electrical design has been implemented to control B.
Tether Cut or Recovery Mode
the actuator direction by using both GUI and semi-automated In this mode, tether cut or failure would have occurred
control using flex sensors for obstacle avoidance. ATMEGA so that input command through serial connection will not be
328P microcontroller is chosen for interfacing sensors, available. So, microcontroller makes its own decision to
actuators and serial communication. Three H-Bridge motor control the motor by relay actuation using analog reading
driver with ULN2003A current driver IC is used for driving from each of the flex sensors. The level of analog voltage
the thruster motors. The electrical components integration of received from the flex sensors will be proportional to the
the ROV is shown in the Fig. 7. physical bend caused by an obstacle if present. So, the ROV
will try to avoid the obstacle by taking reverse path for a
preset time and continuously moves upward to reach the water
surface. Fig. 9. shows the flowchart for ROV control in
recovery mode.

Fig. 7. Block Diagram of Electrical Components Integration

III.
CONTROL DESIGN

A.
ROV Control Algorithm
A GUI for operator control and serial communication is Fig. 9. Flowchart of ROV Control in Recovery Mode
developed using tkinter and serial library in Python. Upon
the availability of serial communication, the ROV will run in IV.
FUZZY LOGIC CONTROLLER
traverse mode, where the microcontroller will give signal Apart from simple if-else algorithm discussed above, a
that makes the ROV to maneuver according to the input fuzzy logic aided navigation control of underwater ROV is
received. Otherwise, the ROV will run in recovery mode in implemented. Operator5( input to the ROV is based on the
which it behaves as a semi-autonomous robot [2]. Fig. 8. visual feedback given by the camera attached to the ROV. The
 2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

fuzzy logic controller acts as brain of the ROV in taking 60 150 160] and [140 150 240 255] for both thruster 2 and 3.
decisions [3]. Fuzzy logic controller reduces the operator Fig.12 shows output membership function in fuzzy toolbox.
fatigue and strain, as obstacle detection and avoidance is taken
care of. The Obstacle Avoidance Control (OAC) layer is
implemented using fuzzy logic. T1, T2 and T3 represent the
thrust of center, left and right thrusters respectively, which are
calculated from each layer. Of the two signals reaching the
ROV, one from the operator and the other from flex sensors
on detecting the obstacle, the human control input has the Fig. 11. Trapezoidal Input Membership Function for Flex Sensor
highest priority. Simulation of membership functions and and Operator Input
rules are done using fuzzy editor tool in MATLAB.
OAC layer consists of two modes. The first is 4dodge mode5
which manages the ROV speed and turning to achieve
collision-free navigation. The other is 4escape mode5 in which
ROV heads upwards, if the ROV finds no space in front or
rear directions. After completing escape mode, OAC layer
changes back to dodge mode. Each of the modes has the same
number of fuzzy behavior rules. The last fuzzy rule in dodge
mode represents the activation condition for escape mode [8].

Fig. 10. Block Diagram of Fuzzy Logic Control Implementation

Block diagram of fuzzy logic control implementation in ROV

Fig. 12. Trapezoidal Output Membership Function for Thruster1,
is shown in the Fig.10. Limitations on the implementation of Thruster2 and Thruster3
fuzzy logic control scheme for ROV are imposed on the
design by physical constraints in hardware. Usage of In dodge mode, the robot will perform rotation until it avoids
computationally expensive methods of fuzzification and obstacle which is detected in front. The direction of rotation
defuzzification needs to be simplified as large number of depends on the magnitude of sensed values by flex sensors
calculations make them impractical during real-time and sensor position. The anticipated Fuzzy Interface System
implementation. After numerous fuzzy rule simulations with (FIS) simulated using MATLAB fuzzy toolbox is shown in
real time scenario, the number of rules are reduced from more the Fig.13. Signal from four flex sensors are given as
than hundred to twenty-one which can easily be coded to directional input to FIS. Output membership functions give
controller and implemented for faster response and decision directional signal to thruster 1, 2 and 3.
making. ATMEGA 2560 microcontroller is used to
implement the fuzzy logic control scheme. Table I, gives the For the ROV, rules are created in mamdani fuzzy type and the
simplified fuzzy logic rule implemented in the ROV5( control defuzzification method used is centroid [9]. Input parameters
scheme which has resulted in substantial development of the and output variables are related with fuzzy logic rule as given
fuzzy logic behavior for the robot. in equation 12 and 13. The centroid defuzzification equation
is given by equation 14 [3].
A.
Obstacle Avoidance Control Layer Design
The primary goal of this layer is to obtain the obstacle !'$
6!  "O4P I 
Q6!O4P 6"O4PR (12)
information. The flex measured by the front-left, front-right, ()
6!  "O4P I  Q
6!O4P 6"O4PR (13)
back-left and back-right flex sensors are denoted as
flex_sensor1, flex_sensor3, flex_sensor2 and flex_sensor4 F
E CD
OBPB
@B
respectively [7]. Corresponding operator input is given #(% I
F (14)
E CD OBP
@B
through joystick. Input trapezoidal membership function for
flex sensors are given as array [0 0 140 150], [120 140 255
255] for near and far proximities (low and high) respectively.
Membership function for operator inputs are given as [0 30 70
80], [60 70 160 170], [140 160 255 255] for low_value,
med_value and high_value respectively for different
directions. The simulated input membership function is
shown in the Fig.11. Membership function for output i.e. for
thruster are given by trapezoidal membership function as [0
10 40 50], [20 40 200 255] for thruster 1 and [0 10 60 70], [50 Fig. 13. Anticipated FIS for Underwater ROV
 2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

TABLE I. Fuzzy Rule Implemented in Underwater ROV

GUI GUI GUI GUI Sensor 1 Sensor 2 Sensor 3 Sensor 4 Motor M2 M2 M3 M3
Front Back Left Right 1 Front Back Front back
1 High Slow Fast Fast
2 Med Slow Med Med
3 Low Low Low Low Low Low Low Low Slow Slow Slow Slow Slow
4 High Slow Fast Fast
5 Med Slow Med Med
6 High Slow Fast Fast
7 Med Slow Med Med
8 High Slow Fast Fast
9 Med Slow Med Med
10 High High Fast Med Fast
11 High High Fast Fast Med
12 Med High Fast Med Med
13 Med High Fast Med Med
14 High High Fast Fast Med
15 High High Fast Med Fast
16 Med High Fast Med Med
17 Med High Fast Med Med
18 High High High Fast Fast Fast
19 Med High High Fast Med Med
20 High High High Fast Fast Fast
21 Med High High Fast Med Med

TABLE II. Comparison of Control Signal Results from Simple If-Else and Fuzzy Controller
Ex. GUI INPUT (Max value 255) FLEX SENSOR (Max 255) M1 (Max 255) M2 (Max 255) M3 (Max 255)
No Front Back Left Right S1 S2 S3 S4 IF-Else Fuzzy IF-Else Fuzzy IF-Else Fuzzy
1 200 5 5 5 15 15 15 15 0 25.4 255 197 255 197
2 5 5 200 5 15 15 15 15 0 25.4 -255 -197 255 199
3 5 5 5 200 15 15 15 15 0 25.4 255 197 -255 -199
4 5 200 5 5 15 15 15 15 0 25.4 -255 -197 -255 -197
5 200 5 5 5 190 15 15 15 255 136 255 105 -255 -199
6 200 5 5 5 15 190 15 15 255 136 -255 -199 255 105
7 5 5 200 5 190 15 15 190 255 136 255 197 -255 -199
8 5 5 5 200 15 190 190 15 255 136 -255 -199 255 197
9 100 5 5 5 150 15 15 15 255 136 255 105 -255 -105
10 150 5 5 5 190 15 15 15 255 136 255 105 -255 -154

TABLE III. Comparison of Linear and Angular Displacement Results for Simple If-Else and Fuzzy Controller
IF-ELSE CONTROL FUZZY LOGIC BASED CONTROL
Direction, Linear Angular X-axis Y-axis Linear Angular X-axis Y-axis
Obstacle displacement displacement Motion Motion displacement displacement Motion Motion
Start 0 0 0 0 0 0 0 0
Front, No 5 0 5 0 5 0 0 5
Front, s3(255) 0 3.038 0 0 3.25 .021 -1.63 4.37
Front, s3(150) 0 3.038 0 0 1.85 0.009 -2.32 3.15
Front, s3(190) 0 3.038 0 0 1.05 0.014 -2.89 2.59
Front, s3 0 3.038 0 0 -.7 -0.021 -1.26 1.97
Left, No 0 -3.038 0 0 -.65 -.032 -1.31 1.95
Left, s1,s4 0 -3.038 0 0 -.7 0.032 -1.36 1.97
Right, No 0 3.038 0 0 -.75 0.032 -1.41 1.98
Right,s2,s3 0 -3.038 0 0 -1.25 -0.032 0.48 0.13

V.
RESULT AND DISCUSSION operator had more ease of operating and navigating the ROV
Multiple input 0 multiple output fuzzy logic controller with the aid of fuzzy logic controller. The results of the test
is implemented for semi-autonomous control of ROV. run through test environment is shown in Table II. The
Comparative study on performance of simple If-Else loop and position results in X and Y plane obtained in test run for both
fuzzy logic controller that provides control signal to thrusters the control logics are shown in Table III. Using the positional
with same sensor input conditions is done. During test trial, information, the path followed by the ROV with the aid of
simple If-Else logic is shown in blue line and with the aid of
 2016 International Conference on Robotics and Automation for Humanitarian Applications (RAHA)

fuzzy logic controller is shown in dashed red line. It can be conventional skill and experience is not required for the
observed from Fig. 14 that test run with fuzzy logic control is operator. The wireless camera provides better view of the
more convenient for operator to navigate through underwater environment. The failsafe mode is made so as to
unstructured environment than simple If-Else logic, due to the give the ROV a backup power when there is loss of tether
reason that If-Else logic uses binary values as sensor inputs which makes it reach the surface by obstacle avoidance using
whereas fuzzy logic uses range of sensor input values. flex sensors. Further, better semi-autonomous control for
underwater vehicle is proposed. Fig.16 shows the proposed
fuzzy-PID controller for ROV with heading direction
correction for semi-autonomous capability[7].
Implementation of proposed system can increase the ease of
operator control further more.

Fig. 14. Path Traveled by ROV during Test Run.

An actual image of test run through known obstacle course is

shown in Fig. 15. The test run was successful for both the Fig. 16. Proposed Block Diagram of Fuzzy-PID Controller
algorithms implemented separately in the ROV5(
%$)'%"

loop. During test run using If-Else and fuzzy logic control REFERENCES
algorithms, both traverse and recovery mode, dodge and
escape mode respectively were tested. [1]
Nima Harsamizadeh Tehrani, Mahdi Heidari, Yadollah Zakeri, Jafar
Ghaisari
1Development, Depth Control and Stability Analysis of an
Underwater Remotely Operated Vehicle (ROV)3
2010 8th IEEE
International Conference on Control and Automation Xiamen, China,
June 9-11, 2010.
[2]
A. R. Marzbanrad, M. Eghtesad, R. Kamali
1A Robust Adaptive Fuzzy
Sliding mode Controller for Trajectory Tracking of ROVs3
2011 50th
IEEE Conference on Decision and Control and European Control
Conference (CDC-ECC) Orlando, FL, USA, December 12-15, 2011.
[3]
%
%$(%$
$
(*
%,$

1 $door Navigation of Mobile Robot
Using F*//.
%
%$)'%""'3
3rd International Conference on
SignalProcessing, Communication and Networking (ICSCN),2015.
[4]
W. P. Lin, C.S. Chin, E. Mesbahi
1Remote Robust Control and
Simulation of Robot for Search and Rescue Mission in Water3
School
of Marine Science and Technology, 2014.
[5]
D. Gracanin, K.P. Valavanis, N.C. Tsourveloudis, And M. Matijasevic,
1Design And Implementation Of A Virtual-Reajity-Based Testbed For
Navigation And Control Of Underwater Vehicles, With The Ultimate
Goal Of Improved Coastal Ecosystem Management3
IEEE Robotics
And Automation Magazine, 1999.
[6]
A. J. Healey, S. M. Rock, S. Cody, D.Miles, and J. P. Brown
1Toward
an Improved Understanding of Thruster Dynamics for Underwater
Fig. 15. Actual Images of Underwater ROV in Test Run Vehicles3
IEEE Journal Of Oceanic Engineering, VOL. 20, NO. 4,
October 1995.
VI.
CONCLUSION AND FUTURE SCOPE
[7]
A W Mussa, W ASandham and T S Durrani
1Underwater Robotics
This paper is an attempt to design and develop cost Navigation Using Multi-Sensor Fusion3
The Institutionof Electrical
effective underwater robotic system for implementing and Engineers, 1998.
understanding the working of different control algorithms. [8]
Jee-Hwan Ryu, Dong-Soo Kwon, Pan-Mook Lee
1Control of
Underwater Manipulators Mounted on nn ROV Using Base Force
This underwater test platform is excellent for demonstration Information3
Proceedings of the 2001 IEEE International Conference
of concepts and academic research. Both navigational control on Robotics & Automation Seoul, Korea .May 21-26, 2001.
logic is implemented and from the results it is clear that fuzzy [9]
Mohd Shahrieel Mohd Aras, Anuar Mohamed Kassim
1Tuning Factor
logic controller can increase the ease of navigation and the Single Input Fuzzy Logic Controller to Improve the Performances
of Depth Control for Underwater Remotely Operated Vehicle3

operation of ROV through complex environment and hence European Modelling Symposium, 2013.