Modelling and control of bioinspired robotic fish underwater vehicle and its propulsion mechanism

Modeling and Control of a Bioinspired Robotic
Fish Underwater Vehicle and its Propulsion
Mechanism
ABHRA ROY CHOWDHURY
(M.Tech., Indian Institute of Technology B.H.U., Varanasi India)
A THESIS SUBMITTED
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF ELECTRICAL AND COMPUTER
ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2014
I)eclaration
I
hereby declare
that
this
thesis
is
my original
work
and
it has
been written
by
me
in its entirety.
I have duly
acknowledged
all
the sources
of

information
which
have
been
used
in the
thesis.
This thesis
has
also
not been
submitted
for any
degree
in
any
university
previously.
NAMtr:
Abhra
Roy Chowdhury
DATtr:
16
lL2
12074
Acknowledgements
I take this opportunity to infinitely thank my supervisor, Assoc. Prof. Dr.
Sanjib Kumar Panda, by including me as a PhD student in his esteemed Elec-
tric Machines and Drives (EMDL) research group. He has reposed tremen-
dous faith in me and always encouraged by letting me explore my own ideas.

He has been extremely kind, considerate and supportive for the entirety of my
thesis work. Moreover, he gave me the opportunity to go to many conferences
and workshops as well as take an active part in the STARFISH2 project. He
always took the time to promote my work during his presentations to the in-
dustry and academia. Furthermore, he is a wonderful human being to know
and to work with, being friendly and supportive even in his busy schedule. I
have learnt from him to be independent, inquisitive, open-minded and most
importantly patient in research. I admire him for many reasons but the most
important habit I have tried to inculcate from him is the positive attitude
and work-life balance.
I would also like to immensely thank Dr. Sangit Sasidhar for the great
collaboration I had found with him for a main part of my thesis. I wish to
express my sincere gratitude to Mr. Y. C. Woo, and Mr. M. Chandra of
Electrical Machines and Drives lab, NUS, for their constant and selfless sup-
port. Their continuous support have made a noticeable contribution towards
my research progress.
I would like to thank Mr. Alok Agrawal, Mr. Vinoth Kumar, Mr.
ACKNOWLEDGEMENTS
Bhuneshwar Prasad for their key contributions for the hardware develop-
ment of the project. I would like to acknowledge useful suggestions and
feedback given by Dr. Rajesh Kumar of MNIT Jaipur, Dr. Wang Xue, Dr.
Manasa Behera of Tropical Marine Science Institute (TMSI) NUS, Mr. Shail-
abh Suman of Acoustic Research Lab NUS, Assoc. Prof. Mandar Chitre of
Acoustic Research Lab NUS, Dr. Pablo Alvaro Valdivia of Singapore-MIT
Alliance for Research and Technology (SMART), Assoc. Professor Marcelo
H. Ang Jr., Professor Xu Jianxin and Assoc. Professor Abdullah Al Mamun
of Department of Electrical and Computer Engineering. I sincerely thank the
office of Defense Science and Technology Agency (DSTA), under the Ministry
of Defense (Singapore) for their support of the present research.
I wholeheartedly thank Dr. Parikshit, Subhash, Saurabh, Dr. Zhaoqin,

Dr. Chinh, Tran, Jeevan, Jayantika, Kalpani, Amit, Sicong and all my col-
leagues in Electrical Machine and Drives Laboratory for useful discussions
and assistances.
Pursuing my research would not have been possible without a good circle
of friends around me in Singapore. I wholeheartedly thank all of them for all
the great times during the period of my study. It certainly helped me a lot
in these last four years.
I would like to express all my gratitude to my parents Mrs. Sati Roy
Chowdhury and Mr. Smaran Roy Chowdhury for their unfailing support,
unconditional love and unbound patience. This thesis would not have been
possible without them. I would like to dedicate this thesis to my parents.
Last but certainly not least, I would like to thank the Almighty for helping
me to learn innumerable life lessons and showing me the direction during the
course of my research.
iii
Contents
Summary xi
List of Figures xiii
List of Tables xvii
List of Acronyms xix
List of Symbols xxi
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The STARFISH 2 Project . . . . . . . . . . . . . . . . . . . . 3
1.3 Fish Swimming Mode Classification . . . . . . . . . . . . . . 8
1.4 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Motivation and Problem Statement . . . . . . . . . . . . . . . 20
1.5.1 Challenge 1: Bio-inspiration from Fish Swimming Modes
for Underwater Vehicle Propulsion and Maneuvering . 22
1.5.2 Challenge 2: Improvement of Energy Efficiency vis-a-

vis the Capability of Propellers . . . . . . . . . . . . . 23
v
CONTENTS
1.5.3 Challenge 3: Stealth and Noise Signature left by AUVs 25
1.5.4 Challenge 4: Bio-inspired Control and Navigation Sys-
tem of Underwater Vehicles . . . . . . . . . . . . . . . 26
1.5.5 Challenge 5: Learning from Group Behaviour and Dis-
tributed Senses of Aquatic Animals . . . . . . . . . . . 28
1.6 Thesis Contributions . . . . . . . . . . . . . . . . . . . . . . . 30
1.6.1 Dynamics modeling of the robotic fish based on biology
inspired principle . . . . . . . . . . . . . . . . . . . . . 30
1.6.2 Kinematics modeling of the robotic fish and mathemat-
ical input waveform design under Lighthill framework . 30
1.6.3 Hydrodynamic Modeling matched with Kinematics of
actual fish . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.6.4 Control Design Methodologies and Comparison . . . . 32
1.6.5 Behaviour based control architecture . . . . . . . . . . 32
1.7 Overview of the Thesis . . . . . . . . . . . . . . . . . . . . . . 33
2 Dynamic Modeling of the Robotic Fish 39
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.1 Robotic Fish (Anterior) Head . . . . . . . . . . . . . . 48
2.2.2 Robotic Fish (Posterior) 2-Link caudal tail as Thruster 50
2.2.2.1 Velocity and Acceleration vectors . . . . . . . 51
2.2.2.2 Forces and Torques . . . . . . . . . . . . . . . 52
vi
CONTENTS
2.2.2.3 Hydrodynamic Forces: Added Mass (Lighthill’s
Reactive Force) . . . . . . . . . . . . . . . . 52
2.2.2.4 Pressure Drag (Resistive) Force . . . . . . . . 63

2.2.2.5 Buoyancy Force . . . . . . . . . . . . . . . . . 64
2.2.2.6 Control forces and Servo motor Dynamics . . 64
2.2.3 Equation of Motion in Earth-fixed Frame . . . . . . . . 65
2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3 Kinematic Modeling of the Robotic Fish based on Lighthill’s
Slender Body Theory 71
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2 Lighthill

s Slender Body Theory based Mathematical Framework 75
3.2.1 Oscillating Sine with Linear Amplitude Wave . . . . . 78
3.2.2 Undulatory Lighthill Quadratic Amplitude Body Wave 80
3.2.3 Undulatory Lighthill Cubic Amplitude Body Wave . . 81
3.2.4 Non-Uniform Rational B-spline (NURB) Quadratic and
Cubic Body Wave (Tadpole-like Motion) . . . . . . . . 82
3.2.5 Undulatory SINC and DIRIC Body Wave . . . . . . . 86
3.2.6 Undulatory Anguilliform Body Wave (EEL-like Ma-
neuvering Model) . . . . . . . . . . . . . . . . . . . . . 88
3.3 Lighthill Control Parameters . . . . . . . . . . . . . . . . . . . 92
3.3.1 Tail Beat frequency (TBF) Based Control . . . . . . . 93
3.3.2 Caudal Amplitude (CA) Based Control . . . . . . . . . 96
vii
CONTENTS
3.3.3 Propulsive Wavelength (PW) and Propulsive Body-wave
Speed Effects . . . . . . . . . . . . . . . . . . . . . . . 98
3.3.4 Yaw Angle effects . . . . . . . . . . . . . . . . . . . . . 101
3.3.5 Determination of Lighthill’s Coefficients . . . . . . . . 105
3.4 Experimental Results and Discussions . . . . . . . . . . . . . . 106
3.5 Operating Region (ORE) . . . . . . . . . . . . . . . . . . . . . 114
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

4 Hydrodynamics Modeling of the Robotic Fish 121
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.2 CFD Modeling of Lighthill

s theory based Undulatory Motion 125
4.3 Simulation Results and Discussion . . . . . . . . . . . . . . . . 133
4.3.1 Pressure and Velocity Field Distributions . . . . . . . . 133
4.3.2 Tail-beat frequency (TBF) Effects . . . . . . . . . . . . 142
4.3.3 Caudal Amplitude (CA) Effects . . . . . . . . . . . . . 145
4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
5 Control System Design of the Robotic Fish 151
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.2 Control Methodologies . . . . . . . . . . . . . . . . . . . . . . 155
5.2.1 Computed Torque Control Method with Dynamic PD
compensation . . . . . . . . . . . . . . . . . . . . . . . 157
5.2.2 Computed Feed-forward Control Method with Dynamic
PD compensation . . . . . . . . . . . . . . . . . . . . . 160
viii
CONTENTS
5.2.3 Computed Feed-forward plus Computed Torque Method162
5.3 Experimental Results and Discussion . . . . . . . . . . . . . . 165
5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6 Behaviour Based Control Design of the Robotic Fish 181
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
6.2 Kinematics based Brain Map and Control Architecture . . . . 190
6.2.1 Distance based Priority / Action Selection . . . . . . . 192
6.2.2 Error based Priority / Action Selection . . . . . . . . 194
6.3 Kinematics Behaviour based High level Control . . . . . . . . 196
6.3.1 Tail Beat Frequency (TBF ) and Phase Shift . . . . . . 197
6.3.2 Caudal Amplitude (CA) Shift . . . . . . . . . . . . . . 198

6.3.3 Mixed Parameters Shift . . . . . . . . . . . . . . . . . 199
6.4 Central Pattern Generator (CPG) Model . . . . . . . . . . . . 201
6.5 Inverse Dynamics Based Low Level Control . . . . . . . . . . . 204
6.6 Modulated Pattern Generators (MPG) Model . . . . . . . . . 212
6.6.0.1 Caudal Amplitude (CA) Parameter Modulation213
6.6.0.2 Tail-beat Frequency (TBF ) Parameter Mod-
ulation . . . . . . . . . . . . . . . . . . . . . . 216
6.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
7 Conclusions and Future Works 221
7.1 Final Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 221
7.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
ix
CONTENTS
7.2.1 Sensory-inspiration of aquatic animals . . . . . . . . . 227
7.2.2 Group behaviors of aquatic animals . . . . . . . . . . 228
Bibliography 231
Publications 251
x
Summary
The research objective is to understand how unmanned underwater vehi-
cles (UUVs) running on the laws of physics can mimic fish biology principles
and deliver better locomotion. Fishes have evolved for over 4.5 billion years
leading to more than 32,000 species, exhibiting greater species diversity than
any other group of vertebrates [1]. This makes them the front-running design
choice for this bio-inspired machine design. The main focus is the locomotion
pattern, energy efficiency and maneuverability of these species that can be
suitably modified and translated to the vehicles performance improvement.
The evolutionary process of fishes was studied [2], specifically their distinct
classifications and morphological traits. The relation between these factors
and the fish locomotion pattern was observed in different fluid environments.

Firstly, it was imminent to understand the internal and external body pa-
rameters that can be translated to a fish inspired vehicle design. The major
kinematic parameters of the robotic fish were chosen based on the kinemat-
ics and energetics study of yellow fin tuna by experimental biologists. These
parameters were found to produce the undulating patterns in the spinal col-
umn of the fish body with optimal thrust. The major parameters were studied
based on a yellow fin tuna comparable to the robotic fish scale. The next step
was to find a bio-fluid-dynamic equation that can accommodate these param-
eters as well as be used as a plug-in body wave generator. The answer was
found with the Lighthills slender body theory. This equation was integrated
to redefine the dynamics of the robotic fish to understand the kinematic re-
lationship with different locomotion patterns. The accuracies of hydrostatic
and hydrodynamic forces were tested to understand the efficacy of the new
SUMMARY
dynamic model in water. Each hydrodynamic coefficient was validated sepa-
rately against one of these kinematic parameters. A novel dynamic model of
a robotic fish underwater vehicle was proposed by unifying conventional rigid
body dynamics and bio-fluid-dynamics of a carangiform fish swimming given
by Lighthill’s (LH) slender body theory. An inverse dynamic control method
based on non-linear state function model including hydrodynamic parame-
ters is proposed to improve the tracking performance. Further, the dynamic
motion closed loop control strategies for the robotic fish were developed and
compared based on three different nonlinear control schemes. These are CTM
(Computed-Torque Method), FF (Computed Feed-Forward) controllers both
with dynamic PD compensation and finally a proposed combination of CTM
with FF. A Matsuoka based non-linear oscillator CPG structure is used to
generate the desired rhythmic pattern preserving control properties like sys-
tem stability and synchronization. A two level locomotion control architec-
ture based on vertebrate fish biology is proposed, where a high level controller
plans the desired trajectory and the synchronization of these trajectories at

each joint, generates the fish-like locomotion behavior. Subsequently, a lower
level control scheme uses an inverse dynamics model based policy for tracking
this locomotion pattern on the nodes along the spinal column of robotic fish.
xii
List of Figures
1.1 AUV and Bio-inspired AUV . . . . . . . . . . . . . . . . . . . 5
1.2 Fish Swimming Modes Classification . . . . . . . . . . . . . . 10
1.3 Robo-Tuna II built in MIT 1994 with lever, pulley and ball
bearing mechanisms . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 PPF-04 Uni-link robot fish design by NMRI, Japan . . . . . . 14
1.5 Robotic Fish Chinois SPC-03 developed in BUAA-CASIA China
for underwater exploration . . . . . . . . . . . . . . . . . . . 15
1.6 Robotic Eel Angulliform Fish (Robea Project) developed in
CNRS France . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7 Essex G8 Robotic Fish diving mode in water . . . . . . . . . 17
1.8 Jessiko V4 Robotic Fish developed in France by RobotSwim . 18
1.9 NTU Robotic Fish . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1 Solidworks Robotic Fish Model . . . . . . . . . . . . . . . . . 44
2.2 Relative orientations and locations of local coordinate frames
at the CM of the head and the Inertial reference frame . . . . 46
2.3 BCF mode Carangiform swimming and travelling wave gener-
ation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1 BCF mode carangiform . . . . . . . . . . . . . . . . . . . . . 75
3.2 Block diagram showing integration of LH Model in the robotic
fish kinematics and dynamics model . . . . . . . . . . . . . . 79
3.3 Undulatory Cardinal Sine travelling wave function . . . . . . 87
3.4 Power Spectral Density of Mathematical Waveforms . . . . . 91
3.5 Forward Velocity vs. Tail beat Frequency . . . . . . . . . . . . 95
3.6 Lighthill Amplitude Variation and Forward velocity (u)at dif-
ferent values of envelope coefficients c

1
and c
2
. . . . . . . . . . 96
xiii
LIST OF FIGURES
3.7 Forward Velocity vs. Amplitude at different wavelength values
(Operating Region) . . . . . . . . . . . . . . . . . . . . . . . 97
3.8 Forward Velocity vs. Wavelength (Operating Region) at a
given TBF = 0.3 . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.9 Propulsive Wavelength and Body Wave speed variations . . . 100
3.10 Yaw Angle vs. Tail beat Frequency at different wavelength
values (Operating Region) . . . . . . . . . . . . . . . . . . . . 101
3.11 Forward Velocity vs. Yaw Angle . . . . . . . . . . . . . . . . 102
3.12 Actual Robotic Fish Path under LH Cubic Undulatory Wave-
function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.13 Light-Hill Implementation Envelope (Undulation) Formation 106
3.14 Relative Path comparison relative to a Centre point (origin) . 107
3.15 Trajectory traversal between two fixed points using different
oscillatory/undulatory wave-functions . . . . . . . . . . . . . . 111
3.16 Time Trajectory for different oscillatory/undulatory wave-functions
112
3.17 Relative Path comparison between Tadpole-like, Carangiform
and Anguilliform undulatory body waveforms . . . . . . . . . 114
3.18 Kinematic Experimental Results . . . . . . . . . . . . . . . . 117
3.19 Closed-loop Hardware prototype motion in different frames . . 118
4.1 Mesh Structure and View of 2D Dynamic meshing for the three
segment robotic fish model . . . . . . . . . . . . . . . . . . . . 129
4.2 Schematic diagram for undulation of fish tail at different time
instants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

4.3 Pressure field and Velocity Field contour distribution around
the fish body . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.4 Hydrodynamic coefficients C
d
and C
l
at different kinematic
parameter values . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.5 Hydrodynamic coefficient C
d
variation with Tail-beat frequency
(TBF) parameter . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.6 Hydrodynamic coefficient C
d
variation with Amplitude Span
(AS) parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.7 Forward Velocity at different kinematic parameter values in
the operating region . . . . . . . . . . . . . . . . . . . . . . . 148
5.1 Computed-Torque Control (CTM) Model of Robotic fish . . . 158
xiv
LIST OF FIGURES
5.2 Feed-forward Control (FF) Model of Robotic fish . . . . . . . 161
5.3 Feed-forward plus Computed Torque Control (FF) Model of
Robotic fish . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.4 Trajectory Tracking using Inverse Dynamics based PID, CTM,
FF and FFCT schemes . . . . . . . . . . . . . . . . . . . . . 166
5.5 Actual Robotic Fish Path with LH cubic spline wave under
CTM, FF and FFCT schemes . . . . . . . . . . . . . . . . . . 174
5.6 Closed-loop Hardware prototype motion in different frames . . 177
6.1 Carangiform Swimming Model showing undulation motion pat-

tern in 1/3rd of posterior body generated by coupled motoneu-
ron in mid-line. . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.2 Behaviour architecture using DES model . . . . . . . . . . . . 191
6.3 Fish biology based Speed Profile of Robotic Fish . . . . . . . 193
6.4 Stimuli based Action Selection . . . . . . . . . . . . . . . . . . 195
6.5 Block diagram of Brain-map modulated Behaviour Feedback
based Control Scheme . . . . . . . . . . . . . . . . . . . . . . 196
6.6 Block diagram of Brain-map modulated Behaviour Feedback
based Control Scheme . . . . . . . . . . . . . . . . . . . . . . 197
6.7 Kinematic Parameter Adaptation . . . . . . . . . . . . . . . . 200
6.8 CPG Signals and generated Trajectory Tracking using Inverse
Dynamics based CTM scheme . . . . . . . . . . . . . . . . . . 208
6.9 Proposed Bio-inspired Distributed Control Dynamic Gains . . 211
6.10 Block diagram of Neurobiology inspired Distributed Control
Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
6.11 MPG Signals and generated Trajectory Tracking using pro-
posed Control structure . . . . . . . . . . . . . . . . . . . . . 214
xv
LIST OF FIGURES
xvi
List of Tables
3.1 Trajectory (Geometric) Points for mathematical oscillatory/undulatory
propulsive waveforms . . . . . . . . . . . . . . . . . . . . . . . 108
3.2 Trajectory (Geometric) Points for mathematical oscillatory/undulatory
propulsive waveforms . . . . . . . . . . . . . . . . . . . . . . . 108
xvii
LIST OF TABLES
xviii
List of Acronyms
AUV

BAUV
BCF
CA
CFD
COM
CPG
CTM
DES
DH
EAP
FF
FFCT
FSI
FSM
LH
MPF
MPG
NACA
NURB
ORE
PI
PPy
PSD
PW
SD
SMA
STARFISH
TBF
UDF
UUV

Autonomous Underwater Vehicle
Bioinspired Autonomous Underwater Vehicle
Body Caudal Fin
Caudal Amplitude
Computational Fluid Dynamics
Centre of Mass
Central Pattern Generator
Computed Torque Method
Discrete Event Systems
Denavit Hartenberg
Electroactive Polymers
Feedforward
Combination of Feedforward and Computed Torque
Fluid Structure Interaction
Finite State Machine
Lighthill
Median Pectoral Fin
Modulated Pattern Generator
National Advisory Committee for Aeronautics
Non Uniform Rational Bezier-spline
Operating Region
Performance Index
Polypyrrole Polymer
Power Spectral Density
Propulsive Wavelength
Standard Deviation
Shape Memory Alloy
Small Team of Autonomous Robotic FISH
Tail-beat Frequency
User Defined Function

Unmanned Underwater Vehicle
ACRONYMS
xx
List of Symbols
θ
i
X
i−1
Z
i−1
F
I
F
B
r
I
r
B
v
B
v
I
M
h
C
h
g
h
τ
h

η
ν
A
i
R
i
T
i
F
i
h(x,t)
c
i
ω
w(x,t)
Joint angle
Horizontal axis
Rotational axis
Inertia frame of manipulator-base system
Base Frame located at the centre of mass of the base
Position of inertia frame
Position of base frame
Velocity of point wrt inertia frame
Velocity of point wrt base frame
Inertia matrix (head) with added inertia
Coriolis-centripetal matrix
Gravity matrix
Propulsion forces vector
Position and orientation vector in earthfixed frame
Linear and angular velocity vector bodyfixed frame

Transformation matrix
Reaction Force
Torque
Generalized Force
Lateral push of body-wave
Caudal amplitude parameter
Tail-beat frequency
Lateral velocity
SYMBOLS
a
P
Q
J
x
t
V(x,t)
u(x,t)
η
f
P
i
C
d
C
l
K
p
K
v
w

y
i
φ
G(jω)
v
c
(t)
v
m
(t)
v
o
(t)
Lagrangian coordinate
Thrust (longitudinal)
Thrust (lateral)
Jacobian matrix
Position in cartesian space
Fluid flow velocity
Forward velocity of the body
Froude’s efficiency
Control points (vector)
Drag Coefficient
Lift Coefficient
Tuned position gain matrix
Tuned velocity gain matrix
External noise
System states
Phase angle
System frequency response

Carrier body-wave signal
Modulating body-wave signal
Modulated body-wave signal
xxii
1
Introduction
1.1 Background
There is currently an increased interest in the use of long range/long duration
Unmanned Underwater Vehicles (UUV’s) for oceanographic observation, mil-
itary surveillance and commercial search missions. Existing Autonomous Un-
derwater Vehicles (AUVs) [1,2] are relatively small vehicles for three reasons:
low cost (fully autonomous vehicles have a significant probability of being
lost), ease of deployment (to allow operations from conventional ships), and
safety (to minimize the danger to manned ships and installations). They are
powered by small rotary propellers driven by electric motors. The propellers
typically operate at fairly low efficiencies and suffer from serious lag times
in transient response. These problems lead to short mission times, restricted
payloads, and control problems. On the other hand, the fishes are found to
be highly maneuverable and effortless swimmers. It has taken more than
160 million years to superbly adapt themselves to the watery environment.
As society becomes ever more technologically advanced, electromechanical
1
1.1 Background
systems are playing an increasingly important role in performing hazardous
tasks; tasks that human otherwise could not perform, and even mundane
tasks such as household chores. Not only should robotic systems be able
to perform these tasks, but they should be able to do so as reliably and as
robustly as possible.
One of the primary areas of focus for any robotic system is its means
of locomotion. Recent advances in robotics have led to the development of

underwater robot systems. The state-of-the-art research work presents contri-
butions to developing an underwater propulsion system through a bioinspired
design process with a focus on a rather simple, yet efficient swimming mech-
anism. The problem of synthesizing the engineering approach for a highly
biologically inspired system for locomotion is addressed in liquid environ-
ment by combining millions of years or evolution with modern engineering
knowledge. The motivation and objective as well as the outline of the re-
search work are provided in this chapter. The field of robotics is undergo-
ing rapid changes as technology continues to allow the development of ever
more complex systems. One of the fastest growing areas of development is
’Biomimetics’[3,4,5], in which robots are designed with the intention of mim-
icking nature. However, mimicking natural systems presents an enormous
challenge to any designer as nature has had millions of years to develop the
complex mechanisms present in the bodies of animals. Take for example, the
human arm. From the shoulder to the wrist, it has 7 degrees of freedom.
To mimic such motion, traditional modern design would require a skeletal
system of complex linkages, joints, hinges cables, pulleys and up to 7 dif-
ferent actuators. Even slightly more advanced designs, which may utilize
improvements such as under-actuated system and the latest developments in
actuation, would still require multiple parts and would require a high degree
2
1.2 The STARFISH 2 Project
of mechanical complexity.
Almost 40 years ago, M.J. Lighthill [6] published a study of the superior
hydrodynamic efficiency of aquatic animal propulsion. Since that time, there
has been a growing effort to capture the benefits of fish-like modes of propul-
sion for use in man-made vehicles. It is generally undisputed that the fish-like
propulsive motion is possibly superior in efficiency relative to propellers, yet
it has not been attempted in the design of large scale submersible vehicles.
In fact, modern submarine design has focused on reducing the disturbance

of the flow around the hull, whereas fish create large disturbances in the wa-
ter. This incredible, nonintuitive design demands our attention and makes
the study of fish-like propulsion very interesting and worth studying. Water
occupies 70% of the planet, thus it is clear that in times of constantly acceler-
ating human activity and rapidly decreasing energy resources it is needful to
improve the current liquid environment propulsion mechanism, which have
remained fundamentally the same since Archimedes designed his screw in
the 3rd century BC. The screw propellers are still used on most of marine
and underwater applications such as ships, submarines, underwater robots
etc. Although using high amount of engineering skills and human knowl-
edge these devices will work sufficiently for most purposes, complex systems
designed by nature still outperform them in various aspects.
1.2 The STARFISH 2 Project
The STARFISH 1 (Small Team of Autonomous Robotic FISH) project [7]
was started in 2006 with an aim to develop a platform to test advanced
Autonomous Underwater Vehicle (AUV) technology including algorithms for
cooperative AUV missions. The first major milestone in this project was
3