Kinematic Evaluation of a Series of Soft Actuators in Designing an

Eel-inspired Robot
Dinh Quang Nguyen and Van Anh Ho

Abstract— In this paper, we present the preliminary design

of a soft robot inspired by the eel which swims by undulating
its body in wavelike motion to generate thrust force. The
robot body comprises of a series of soft pneumatic actuators
controlled by Pulse-Width Modulation. A soft actuator was
simulated and designed to be of suitable shape to mimic an
eel. The robot eel body, was then, fabricated, incorporating
pairs of soft actuators. In fish locomotion analysis, kinematics
of the midline, which describes the path of the spine, is
frequently studied to represent the whole body for the purpose
of simplifying analysis. We investigated midline kinematics in
a model angulliform swimmer. Typical kinematic parameters
of the robot body during swimming were estimated to examine
their change with variation of frequency. A kinematic model of
the robot body was built to estimate the thrust force component.
The results revealed good potential to develop our proposed
design with midline robot body kinematic properties similar to
those in natural long body fish, with movement in the robot
body able to create thrust force.
Fig. 1. (a) Elongated body fish ambient fluid model when swimming [9],
[10]. The movement of body cause to create forward force. (b) Preliminary
I. INTRODUCTION design of elongated body fish soft robot with three main part: head contain
Soft robotics is an emerging research field, in which control parts; body is constructed from soft actuators which is controlled
to create flapping movement duplicating swimming eel/lamprey movement;
soft materials such as polymers, elastomers, hydrogels or spine is made of non-stretch flexible polymer
granules can be utilized for realization of tasks that cannot
be accomplished with conventional rigid robots. Recently,
novel designs of soft robots have been applied in man-
ufacturing, medical devices, search and rescue, and more information about the movement strategies of underwater
[1]. Among them, bioinspired aquatic robots are of interest animals.
in development of underwater mobile devices for search
activities, environment surveillance, and studying the physics B. The classification of fish swimming modes
of underwater movement by natural species.
Among behaviors of aquatic animals, locomotion of fish
A. Potential application of underwater soft robot has been well studied. Their swimming may be classified into
Most aquatic animals move by swimming. In swimming, two locomotion groups BCF (body and/or caudal fin) and
animals move the whole or part of their body, such as MPF (median and/or paired fin). Fish in BCF group employ
fins, or a tail, in interaction with the ambient water, to swimming strategies that include propagating a wave along
generate force for balance and/or locomotion. By mimicking their body to generate thrust force. Generally, BCF fish swim
such actions, soft robots of suitable structure are able to at high speed and with less complex body movement than
swim [2], [3]. Underwater soft robots are typically made MPF fish [5], which may explain why much research has
of soft flexible materials enabling interaction with fragile focused on BCF-inspired soft robots [6], [7], [8]. There are
objects without damage. Such characteristics facilitate their two main swimming strategies in the BCF group, namely
use in narrow and complex structures, such as coral reefs, undulation and oscillation. Undulatory movement is typified
and are useful in applications to monitor the development by entire body movement generating a series of sinusoidal
of ocean ecosystems or take sample at site of pollution waves from head to tail in long body fish, and just posterior
[4]. The development of aquatic robotics has been mainly part of the body in short body fish. Short body fish, especially
based on previous research on the movement mechanics of thunniform, swimmer where the undulation is generally but
aquatic animals. This research field has provided abundant not always confined to the caudal fin, show efficient swim-
ming at high speed for long periods. Angulliform swimmer,
Nguyen and Ho are with the School of Materials Science, Japan Advanced elongated body fish such as eels, are able to equally well
Institute of Science and Technology (JAIST), 1-1 Asahidai, Nomi, Ishikawa,
923-1292 Japan. Email: [email protected], forward and back ward by changing the direction of the wave
[email protected] that travels along their body [4].
C. Fish-inspired soft robots
In soft robotics, research inspired by fish that swim by
oscillation of their posterior body has been conducted. Wang
et al. [11] used a shape memory alloy (SMA) to create
the undulatory movement of a fish. The research showed
good results for robot swimming speed but limitations of
the material and swimming mechanics resulted in low per-
formance of biomimetic fin. Meanwhile, Katzschmann et al.
[12] reported an autonomous soft robotic fish that employed a
hydraulic soft actuator to oscillate its posterior body. The fish
robot demonstrated efficient swimming in three dimension
albeit at low speed (300 m/40 min). An undulatory swimming
soft robot eel was proposed by Christianson et al. [13] that
used three dielectric elastomer actuators. By applying a high
voltage, an actuator made a bending movement in the robot
body. Three actuators in sequence allowed the movement
to cascade from anterior to posterior sections of the robot.
Nonetheless, limitations of the robot eel fabrication material
adversely effected the robot’s swimming performance.

D. Contributions
In this work, inspired by previous work in the field and
the locomotion of elongated body fish, we investigated the
concept of a soft robot that used a series of pneumatic soft
actuators as shown in Fig. 1(b). Contributions of this research
are as follows:
1 Proposed the preliminary design of an elongated body
fish soft robot that included a number of new design
pneumatic soft actuators which provided undulatory Fig. 2. Design and simulation of soft actuators’ pressurized situation. (a, b)
movement via Pulse - Width Modulation. are the design of two different kinds of soft actuator with wide and narrow
2 Proposed a kinematic model to estimate the thrust force connecting channel with side view, isometric view of whole soft actuator
and isometric view of designed actuator without base plate, and (c, d) are
of the robot body using undulation frequency. the simulation results, respectively.

II. DESIGN AND FABRICATION OF THE SOFT

ACTUATOR AND ROBOTIC BODY
channel 112×39×3 mm long×wide×deep, while that of the
A. Design of a soft actuator other design (see Fig. 2(b)) is 112×5×3 mm. Abaqus CAE
In soft robotics, researchers have focused on two main 2018 was used to model how they react when pressurized
areas. One is the design, fabrication and control of robots at 0.03 MPa. Dragon skin 20 was chosen to fabricate the
able to perform tasks, the other is the design and fabrication soft actuators. This silicone may show nonlinear and hyper-
of soft materials suitable for soft robots. Research on de- elastic deformation, therefore, Yeoh model should be applied
signing soft actuators is one of the most interesting topics in to obtain appropriate simulation results [17]. Table I shows
the former [14]. Soft robots have been designed for different the material parameters and coefficients for simulating the
purposes and tasks that require various body shapes and sizes pressurized stage of the newly designed soft actuator. At the
and have either incorporated previously design actuators or meshing step, hybrid formulation was chosen for simulation
newly designed actuators to enable performance of novel of an object made of hyper-elastic material.
tasks. As mentioned above, elongated body fish swim by suc- TABLE I
cessively undulating their body in the shape of a sinusoidal M ATERIAL PROPERTIES OF D RAGON SKIN 20 [17]
wave from head to tail. To mimic this swimming style, a
robot would require a combination of actuators to bend its
body when pressurized. Interesting ideas can be gleaned from Parameters/ Mass Constant Constant Constant
Coefficients density C10 (MPa) C20 (MPa) C30 (MPa)
previous research with segmented, chambered soft actuators (tonne.mm−3 )
[15], [16]. Fig. 2 shows the proposed design of pneumatic
soft actuators, and simulation results when pressurized. Two Value 1028−12 0.44 0.005 -0.00002
actuators share the same segmented, chambered structure
apart from air channel which feeds the chambers. Fig. 2(a) Results are shown in Fig. 2(c) and (d) for the wide and
shows the soft actuator design with a wide connecting narrow connecting channels, respectively. In the former, the
soft actuator did not bend in the same curvature radius at
any actuator’s section. The wide connecting channel created
a large deformation area at the base plate which concentrated
stress at this side. This inhibited deformation at segmented
sections which play the main role in bending. However,
change in the narrowing connecting channel achieved good
results as shown in Fig. 2(d). In this design, pressurized,
stress accumulated at segmented, chambered section result-
ing in the actuator bending in a perfect curve shape. The
latter design was chosen to mimic muscle in the eel robot.
Fig. 3 shows a simulation of two actuators in two contrasting
situations - pressurized and unpressurized. The simulation
result indicated that the combination of the two soft actuators
with a suitable sequence of pressure alternation would induce
undulatory movement. This is the groundwork to design the
soft robotic eel’s body.

Fig. 3. Simulation of a pair of actuators (downer actuator in pressurized

stage of 0.06 MPa). The inset legend shows results in stress components
Fig. 4. Soft actuator fabrication process.

B. Development of a physical model of robotic body

Unit: mm
The elongated body fish robot comprises three main com-
ponents including a head, a spine, and three pairs of soft

45
actuators. The head would contain the controls including
122
control circuit, pump, and connecting pipes. The spine would 255
be made of non-stretch flexible polymer. And the three pairs
of actuators would be attached in symmetrical pairs at each Fig. 5. Fabricated actuator (a); A pair of soft actuators (b) and the robot
side of the spine. The preliminary design of the soft robot body with two pairs of actuators (c).
is shown in Fig. 1. During steady swimming, the unlulation
amplitude of an eel’s body increases from head to tail, this
strategy enables elongated-body fish to swim forward. To 0.35 mm thick which simulates an eel spine. Strong adhesive
mimic such swimming characteristics, each pair of actuators glue PPX was used to affix the actuators to the spine. Fig.
in the robot plays a different role. The first pair, which 5 shows a fabricated actuator, a pair of actuators and the
are near the head, actuate at the lowest amplitude during eel body robot with two pairs of actuators. As mentioned
stable swimming. In addition, this pair can change the robot earlier, in our design the two posterior actuator pairs play
swimming direction by bending the head to the left or right. the main role in creating thrust force, therefore, we intended
The other pairs of actuators play no role in swim direction. to test their operation first to make theoretical conclusion on
Increase in their amplitude and frequency leads to increase midline kinematics and thrust force.
in swimming velocity, with the tail actuators providing the
D. Design and assembly testing control circuit
greatest influence.
The motion of the actuators generated using Pulse-Width
C. Fabrication Modulation control. A performance program used to control
Semi-cylindrical soft actuators, of 122 mm length and the open/shut status of solenoid valves (VQD 1121-5L, SMC,
22 mm radius, were made from silicone rubber Dragon skin Japan) with a power circuit (MOSFET 4, UO 4) was written
20 (Smooth-On, Inc., USA). Liquid silicon was poured into a on Arduino (Arduino, Uno R3, Italy). Pressurized air was
3D printed mold to form the segmented, chambered structure generated by a JUN – AIR generator (JUN-AIR, USA),
after mixing type A and B in the ratio of 1:1 in weight and and fed through an adjustable flow control valve (AS2052F,
re-gassing in a vacuum chamber. The liquid silicon in situ on SMC, USA) to control the air flow rate to the solenoid
the mold was cured at room temperature of 22 Celcius degree valve which directed compressed air to the soft actuators
over 4 hours. All parts of the mold were designed using in sequence. The test system diagram is shown in Fig. 6.
SolidWorks then printed by 3D printer (M200, Zortrax). For the operation of the robot body (see Fig. 5(c)), actuator
After curing, the actuator parts were removed from the mold at the right side of the first pair and that at the left side of
and glued together using a silicone epoxy (Sil-poxy, Smooth- the second pair were connected. The similar connecting way
On, Inc., USA). The fabrication process is shown in Fig. 4. was performed with the two others. Therefore, solenoid valve
After the fabrication stage, pairs of actuators were attached will contribute pressure air to two soft actuators in each open
to either side of a thin plastic plate made of flexible acrylic cycle. This makes two pairs of soft actuator bend in similar
 Point P moves following the rule of Eq. (1) and creates force
F, therefore, F follows Newton’s Second Law:
F = mÿ. (3)
The weight m of each point is counted as virtual mass (per
unit length) [17] and is approximated to:
1
m = πρs2 , (4)
4
where: ρ is water density and s is the cross section area of
the robot body
From Eq. (1) to Eq. (4), action force F and thrust force
may be rewritten as Eq. (5) and Eq. (6), respectively:
 
3 2 2 2π
Fig. 6. The diagram of control system. F = Aπ ρs f sin x − 2π f t , (5)
λ
 
2π
FT h = Aπ 3 ρs2 f 2 sin x − 2π f t cos(θ ). (6)
frequencies with inverse phases, create sinusoidal wave body λ
during pressurization.

III. ROBOT BODY KINEMATICS

The elongated body fish robot achieves propulsion by

undulating its body to create thrust force for forward move-
ment. The body shape in this situation can be considered
as a sine wave. Using the Cartesian coordinate system
Fig. 7, each point P on the body may be determined by
the parameters (x, y,t) with the relationship between these
parameters expressed as:
Fig. 8. The variation of angle θ depending on x.
 
2π
y = A sin x − 2π f t , (1) The angle between action force F and thrust force FT h
λ
decreases with increase of x. When x increases from 0 to L/2
where: A is tail beat amplitude, λ is robot body wave length, (L - body length), θ decrease from π to 0. In this case, we
and f is tail beat frequency assume that the relation between x and θ can be considered
nearly linear. This means that:
y θ = ax + b. (7)
P’ FL
P’(x’, y’, t’) F Based on variation of the values of x and θ as above, an
FTh q equation describing the relationship between x and θ can be
P(x, y, t) P expressed as below:
O x 2π
θ =− x + π. (8)
L
Replacing θ in Eq. (6) by Eq. (8), we obtain a new
Fig. 7. The diagram of interaction force between swimming robot body
equation to calculate thrust force at a point along the midline
midline and ambient fuild [5]. of the robot body:
   
2π 2π
In building the kinematics model, the robot body midline FT h = Aπ 3 ρs2 f 2 sin x − 2π f t cos π − x . (9)
λ L
(spine) was taken to be a set of points with the movement
Total thrust force created by lateral movement of midline
of each point P(x, y,t) expressed as Eq. (1). In practical
may be determined as the sum of the forces created at all
application, these points impact on ambient water with force
points and can be calculated:
(F) which can be divided into two component forces (FL
Z L    
– Lateral and FT h – Thrust) as shown in Fig. 7. The 3 2 2 2π 2π
FT h = Aπ ρs f sin x − 2π f t cos π − x dx.
surrounding water provides a force that interacts with the 0 λ L
thrust force to propel the robot body forward. Thus: (10)
In Eq. (10), parameters A and λ are from the midline
FT h = F cos(θ ). (2) kinematic data, while L and s are from the robot body design.
The undulation frequencies were set at 0.8, 1.0, 1.1, 1.7 more compressed air yielding greater tail beat amplitude. The
and 3.3 Hz based on the variation of tail beat frequencies midline kinematics of the robot body at 1.7 and 1.1 Hz were
of natural elongated body fish up to 3.1 Hz [10] when roughly similar to that of an anguilliform animal as shown
swimming. Eq. (10) can be used to estimate whether the in Fig. 10. Nonetheless, the midline kinematics in the robot
robot body can create thrust force while undulating. revealed a difference in movement of the tail end, where
a natural animal showed more flexible movement than the
IV. T ESTING METHOD , EQUIPMENT AND RESULTS
fabricated robot body.
A. Experimental setup In comparison with a natural lamprey, which is categorized
as an anguilliform swimmer, as shown in Table II [18],
Camera swimming parameters of the robot body at testing frequency
y
1.7 Hz were approximately similar to an increase in ampli-
Unit: mm
Fixing rod
tude from the anterior to posterior body just over 2-fold.
The body wave speed and tail beat amplitude of the robot
800
body were also similar to a natural lamprey. However, the
z value of the body wave length was quite different. This may
60 Water level have been due to the size of the actuators and the stiffness
O of the robot body. At other frequencies, these parameters
x
Robot body
all showed differences with those of a lamprey. The thrust
(a) Diagram of experimental setup forces, created at time T (a completed flapping cycle) of
each undulation frequency were estimated using Eq. (10)
and parameters from Table II. The robot body generated
its highest thrust force of 0.297 mN when operating at a
(b) Max bending at 3.3 Hz (c) Max bending at 1.7Hz (d) Max bending at 1.1 Hz frequency of 3.3 Hz and its lowest of 0.029 mN at 0.8 Hz.
Thrust forces that robot body create in five different values
of tail beat frequency have medium value compared to these
of natural elongated body fishes which were estimated in the
(e) Max bending at 1 Hz (f) Max bending at 0.8 Hz
researches of Tytell [10] with eels of 208 mm average length
Fig. 9. Experimental setup and some related results (a) Diagram of experi-
(creating thrust force 0.79 mN) and Muller [21] with eels
mental setup; (b) Fabricated body of soft robotic eel fixed on supporting rod; of 80-100 mm length (creating thrust force 0.1-0.3 mN). The
and maximum bending situation of the body at the operating frequencies thrust force values indicated that undulatory movement of
of 3.3 (c), 1.7 (d), 1.1 (e), 1 (f) and 0.8 (g) Hz in the testing swimming
situation in water environment.
the robot body can create forwarding swimming. However,
the variation in thrust force was not dependent on body
The body model with two pairs of actuators was attached undulation frequency. Increase in frequency did not lead to
to a fixing rod from the interior of the spine and performed increase in thrust force. In further research, we will measure
in stable water conditions in a tank 460×360×260 mm long, thrust using a force sensor and construct a model to explain
wide and deep, respectively. All testing data were recorded this phenomenon.
using a Sony DSC-RX10M4 camera (Sony, Japan) at 960
frames per second. The tail beat frequencies were set at V. D ISCUSSION
3.3, 1.7, 1.1, 1.0 and 0.8 Hz. All frame data at each tail
beat frequency regime was extracted using Matlab R2018b The test results showed some differences in midline kine-
(MathWorks). Based on these data, the values of tail beat matics when swimming. The differences, mentioned above,
amplitude, amplitude growth rate and robot body wave length may have been due to the dimensions of the soft actuators
were estimated. Fig. 9(a) shows the experimental set up. and/or the stiffness of the robot spine. In this work we did not
The robot body fixed to a rod was submerged in the water investigate in detail the effects of these factors on swimming
to avoid surface water ripples, which would complicate the ability, these will be investigated in our future work. To esti-
task of extracting the midline kinematics. Fig. 9(b) samples mate these effects, the hydrodynamics of a swimming robot
of bending at five different frequencies with the body robot should be mentioned. The undulatory motion of elongated
fixed are shown in Fig. 9(c)-(g). fishes reacts with surrounding fluid to result in swimming.
As a result, ambient fluid will express different modes or
B. Results wake structures corresponding to bending/swimming char-
The midline kinematics of the robot body for five testing acteristics. A comparison of wake structures created by the
frequencies are shown in Fig. 11. The results indicated swimming of a robot, with those of an eel or lamprey, may
that the tail beat amplitude increased as tail beat frequency elucidate factors to estimate robot swimming efficiency. The
decreased. This may have been due to the port status of tail segment is the longest section in natural elongated body
the solenoid valves which directed compressed air to the fish and plays the most important role in creating thrust force.
actuators. A lower tail beat frequency corresponded to a Therefore, the length of each robot body segment will be
longer valve open time. As a result, the actuators received optimized by simulation.
Fig. 10. Kinematics of robot body midline in five different frequencies of the testing operation (dorsal view). XYZ coordinate system can be referred to
Fig. 9(a).

TABLE II
E LONGATED BODY FISH ROBOT BODY AND NATURAL LAMPREY SWIMMING PARAMETERS (L = BODY LENGTH )

Tail beat frequen- 0.8 1.0 1.1 1.7 3.3 Lamprey

cy, f (Hz) [18]

Tail beat ampli- 0.15 0.14 0.12 0.11 0.07 0.09

tude, A (L)

Amplitude 2.83 2.46 2.30 2.17 1.67 2.18

growth rate, α

Body wave 0.48 0.49 0.5 0.52 0.53 0.64

length, λ (L)

Thrust force 0.029 0.045 0.05 0.119 0.297

(mN)

pneumatic soft actuators which were designed and simulated

to determine a shape suitable for swimming. We successfully
created undulatory motion to generate thrust forces induced
by the pairs of actuators by Pulse-Width Modulation. Midline
kinematics revealed good potential for developing this model
from data at testing frequencies of 1.7 and 1.1 Hz, which
Fig. 11. Midline kinematics of aquatic animal swimming in anguilliform.
Modified from [20] is close to those in swimming natural elongated body fish.
Additionally, parameters of tail motion amplitude, amplitude
growth rate, and body wave speed at 1.7 Hz of tail motion
The ability of creating propulsion of robot body is the frequency were roughly similar to those of natural lamprey.
foundation for development of eel soft robot. In this research, The thrust force estimation results indicated that the move-
however, the effect of both lateral force component and ment of two pairs of actuators in sinusoidal waveform at
robot body morphology on propulsion are not mentioned. different frequencies can create propulsion.
It required to do research on the effect of lateral force
In future work, we intend to optimize the robot size
component, flapping frequency and robot body structure on
and swim tail frequency by simulation method. Further
thrust force to find the good parameters for completing robot
experiments will be conducted on robot swimming thrust
design and operation modes.
force to compare with natural animals and hydrodynamic
aspects in swimming situations to estimate robot swimming
VI. C ONCLUSION
efficiency. A kinematic model of the effect of lateral force FL
In this paper, we introduced the design of a soft elongated will also be developed to investigate further the relationship
body fish robot with a body constructed with three pairs of between operation regimes and thrust forces.
 R EFERENCES biomimetic underwater propulsion”, Smart Materials and Structures ,
Vol.17, 2008.
[1] R. J. Webster and B. A. Jones, “Design and kinematic modeling of [12] Robert K. Katzschmann, Andrew D. Marchese, and Daniela Rus, “Hy-
constant curvature continuum robots: a review”, The International draulic autonomous soft robotic fish for 3D swimming”, Conference on
Journal of Robotics Research, Vol.29, pp. 1661-1683, Nov. 2010. International Symposium on Experimental Robotics , Vol.14, pp.405-
[2] Andrew D. Marchese, Cagdas D. Onal, Daniela Rus, “Autonomous 420, 2014.
soft robotic fish capable of escape maneuvers using fluidic elastomer [13] Caleb Christianson, Nathaniel N. Goldberg, Dimitri D. Deheyn,
actuators”, Soft Robotics, Volume 1, Number 1, 2014. Shengqiang Cai, Michael T. Tolley, “Translucent soft robots driven
[3] Jennifer Frame, Nick Lopez, Oscar Curet and Erik D Engeberg, by frameless fluid electrode dielectric elastomer actuators”, Science
“Thrust force characterization of free-swimming soft robotic jellyfish”, Robotics, Vol. 3, Issue 17, Apr 2018.
Bioinspiration and Biomimetics, September 2018. [14] Pinar Boyraz, Gundula Runge, Annika Raatz, “An overview of novel
[4] Robert K. Katzschmann, Joseph DelPreto, Robert MacCurdy, Daniela actuators for soft robotics”, Actuators, MDPI, 16 August 2018.
Rus, “Exploration of underwater life with an acoustically controlled [15] B. Mosadegh, P. Polygerinos, C. Keplinger, S. Wennstedt, R. Shepherd,
soft robotic fish”, Biomimetic, March 2018. U. Gupta, J. Shim, K. Bertoldi, C. Walsh, G. Whitesides, “Pneumatic
[5] Michael Sfakiotakis, David M. Lane, J. Bruce C. Davies, “Review networks for soft robotics that actuate rapidly”, Advanced Functional
of fish swimming modes for aquatic locomotion”, IEEE Journal of Materials, 2014, 24, 2163–2170.
Oceanic Engineering, Vol.24, No.2, pp. 237–252, 1999. [16] Andrew D. Marchese, Robert K. Katzschmann, Daniel Rus, “A recipe
for soft fluidic elastomer robots”, Soft Robotics, 2015, Vol.2, No.1,
[6] Yong Zhong, Jialei Song, Haoyong Yu, Ruxu Du,“Towards a transform
7–25.
method from Lighthill fish swimming model to biomimetic robot fish”,
[17] S. Azim, A. Srinivasan, M. Pandaram, J. Kow, G. Boer, H. Wang, A.
IEEE Robotics and Automation Letters, Vol.3, Issue 3, July 2018.
Alazmani, P. Culmer, “Soft tactile sensors with variable compliance”,
[7] Cassandra M. Donatelli, Sarah A. Bradner, Juanita Mathews, Erin
Conference on IEEE sensors, pp. 663-665, 2017.
Sanders, Casey Culligan, David Kaplan, Eric D. Tytell, “Prototype of
[18] M. J. Lighthill, “Large – Amplitude elongated – body theory of
a fish inspired swimming silk robot”, IEEE International Conference
fish locomotion”, Proceedings of the Royal Society of London B:
on Soft Robotics, 2018.
Biological sciences, vol. 179, pp.125-138, 1971.
[8] Ardian Jusufi, Daniel M. Vogt, Robert J. Wood, George V. Lauder, [19] Brad J. Gemmell, Stephanie M. Fogerson, John H. Costello, Jennifer
“Undulatory swimming performance and body stiffness modulation in R. Morgan, John O. Dabiri and Sean P. Colin, “How the bending
a soft robotic fish-inspired physical model”, Soft Robotics, Vol.4, No. kinematics of swimming lampreys build negative pressure fields for
3, 2017. suction thrust”, Journal of Experimental Biology 219, 3884-3895,
[9] J. Carling, Thelma L. William, G. Bowtell, “Self-propelled anguil- 2016.
liform swimming: Simultaneous solution of the two-dimensional [20] BJ.L. Lim and T.M. Winegard, “Diverse anguilliform swimming
Navier-Stokes equations and Newton’s laws of motion”, The Journal kinematics in Pacific hagfish (Eptatretus stoutii) and Atlantic hagfish
of Experimental Biology, 201, pp.3143-3166, 1998. (Myxine glutinosa)”, Can.J.Zool, pp. 213-223, 2015.
[10] Eric D. Tytell, Geogre V. Lauder, “The hydrodynamics of eel swim- [21] U. K. Muller, J. Smit, E. J. Stamhuls, J. J. Videler, “How the body
ming: I. Wake structure”, The Journal of Experimental Biology, 207, contributes to the wake in undulatory fish swimming: Flow fields of
pp.1825-1841, 2004. a swimming eel (Anguilla Anguilla)”, The Journal of Experimental
[11] Zhenlong Wang, Guanrong Hang, Yangwei Wang, Jian Li and Wei Biology, Vol. 204, pp. 2751-2762, May 2001.
Du, “Embedded SMA wire actuated biomimetic fin: a module for