You are on page 1of 6

Proceedings of the 2011 IEEE

International Conference on Robotics and Biomimetics


December 7-11, 2011, Phuket, Thailand

Improvement and Testing of a Robotic Manta Ray (RoMan-III)


K. H. Low, Chunlin Zhou, and Gerald Seet Shusheng Bi and Yueri Cai

Abstract—Manta Ray generates thrust force by flapping two Brower [3]. To mimic the locomotion of manta rays, a robot
pectoral fins, which inspires the parametric study of the fin that should be autonomous so that it can switch among different
affects the swimming performance. In this paper, the design of a locomotion patterns when activated by sensors.
robotic manta ray (RoMan-III) will be presented. A biomimetic Palmisano et al. [4] identified four major independent
flapping fin model will be discussed. The forward flapping parameters that can be varied to change the force generation
motion, turning and gliding motions are considered in the study time-history during a stroke cycle: fin surface curvature, root
of Manta Ray’s locomotion. Parameters related to thrust angle of attack, bulk rotation angle and frequency. Fin surface
generation include the fin flapping amplitude, frequency, fin curvature refers to the shape of the fin with respect to time.
area, and fin shape. The fin model is derived based on a The angle of attack is the angle of the fin root with respect to
simplified model by Bernoulli equation. A scheme of motion the horizontal axis. The CFD analysis showed no thrust
control is also suggested for the fish locomotion. Experimental benefit to varying angle of attack over time. The bulk rotation
results have been obtained to demonstrate the validity of the angle is the peak to peak amplitude of a stroke, and it was
proposed model. found that increase in rotation angle leads to increase in thrust
force.
Keywords- Biomimetics, robotic fish; manta ray; flapping fin. The advantages of ray-like swimming have encouraged
the development of fish robots. An excellent design of
ray-like fish robot was illustrated by the FestoTM Aqua Ray
I. INTRODUCTION [5]. Driven by fluidic muscle, Aqua Ray can perform elegant
Scientists and researchers try to analyze and mimic the swimming and gliding motions in water. In [6], the flapping
actuation of aquatic creatures in order to invent the most motion was driven by multiple parallel-arranged bending
energy-efficient propulsion. The area of biomimetic robots pneumatic rubber actuators. In both robot designs, the fin
has been greatly explored due to the increasing awareness to motions can exhibit excellent dexterity under the control of
preserve environment. In this paper, a biomimetic robot fluidic actuators. However, such actuators are inevitably
(RoMan-III) mimicking manta rays will be introduced. Manta bulky in size and slow in movement. Electric motors are
rays exhibit rajiform motion (thrust mainly comes from adopted in new designs to tackle the problem. Extensive
movements of two pectoral fins) during its swimming in the efforts have been made on the ray-like fish robots in Beihang
ocean [1]. There are several reasons motivating the study of a University, which results in four generations of bionic fish
manta ray robot. The flapping motion of manta ray is propelled by wide pectoral foils (Robo-ray I to IV). Both
considered as energy-efficient as energy is mainly used for pneumatic artificial muscle and electric servomotors were
propulsion with less water dissipation. Its gliding motion adopted as actuators to drive pectoral fins [7, 8]. Ray-like
would be extremely significant in energy saving since the flapping motions were successfully implemented on those
robot needs no more flapping. Instead of oscillating their robots. Besides the flapping motions, the basic gliding motion
bodies and caudal fins, manta ray has a big and flat body that has been achieved in the robotic manta ray developed at the
keeps stable in swimming. High maneuverability in water Nanyang Technological University (NTU). The first
exhibited by manta ray is another advantage [2]. prototype (RoMan-I) driven by servomotors has been
Manta ray could perform several locomotion patterns developed and tested [9].
including the forward/backward swimming, turning and
gliding, which will be implemented and elaborated by
flapping two pectoral fins, modulating propagation directions
of the fin waves and holding stable attack angle of fins,
respectively. During swimming and turning motions,
sinusoidal wave will be applied on the fin, as described by

Manuscript received on October 25, 2011. The financial supports by the


NTU-Mindef/06/01 and MoE AcRF RG23/06 Research Grants (Singapore)
are acknowledged.
K. H. Low , Chunlin Zhou and Gerald Seet are with the School of
Fig. 1. Scaled diagram of a manta ray. Dimensions of the two fins and body
Mechanical and Aerospace Engineering, Nanyang Technological University
are normalized to the wing span [9]
(NTU), Singapore (phone: +65-67905755, email: mkhlow@ntu.edu.sg)
Shusheng Bi and Yueri Cai are with Robotics Institute, Beihang University
(BUAA), China (email: biss_buaa@163.com)

978-1-4577-2138-0/11/$26.00 © 2011 IEEE 1730


Manta ray could be recognized from its diamond-shaped
flat body. The shape and size of its two wing-like fins and the TABLE I Comparisons between RoMan-II [10] and RoMan-III
streamlined body play important roles in the swimming. The Specifications RoMan-II RoMan-III
scaled diagram of a manta ray is shown in Fig. 1. The Mass(kg) 7.3 5
proportions of the body shape for the developed prototype Wingspan (cm) 100 88
approximately follow those of the real manta [9, 10]. Body Length (cm) 50 37
The paper is organized as follows. The mechanical design Velocity(m/s) 0.4 0.3
of RoMan-III will be reviewed in Section II. The fin model Velocity(BL/s) 0.8 0.81
will be elaborated in Section III. In Section IV, the control Coloring Nippon Paint Acrylic Glue
methods and sensors will be introduced. Experiments and
results will be presented in Section V. III. ANALYSIS OF FIN MODEL FOR FLAPPING MOTION
In analyzing the hydrodynamics of RoMan-III, the
II. ROBOTIC MANTA RAY (ROMAN-III) following assumptions are made to simplify the problems: (1)
Water turbulence could be ignored; (2) Water is still; (3) The
A. Design of RoMan-III angle of attack is zero;(4) The fins are not deformable under
It is necessary to review the mechanical design as pressure; (5) Water near the fin surface will has same speed as
dynamic analysis requires the dimensions and parameters of the fin.
the RoMan-III, especially the fin design. As shown in Fig. The fin design is simplified as shown in Fig. 3, where the
2(a), the robot has three parts mainly: body, water tank and curves at distal is assumed as straight, width of rays is
fin. assumed zero, thickness of silicon rubber is assumed zero.
First Middle Last

Fig. 2 Design of fish robot (a) Overview of robotic manta ray (RoMan-III) (b)
Fin design of RoMan-III
Fig. 3 Simplified fin model and dimensions
The fin design consists of three servo motors on each side
and with three ribs attached to the shaft. The three ribs are A. Thrust
named the first ray, middle ray, and third ray from the left in To generate propulsion, sinusoidal wave will be used. For
Fig. 2(b). The rays are parallel-connected and actuated each ray, its oscillation will be expressed as angle position
independently by the servo motor. All the six rays are made of respect to time:
Teflon (Polytetrafluoroethylene) to add compliance of θ = ϕ sin(2π ft − nα ) (1)
motion. The structure of manta ray’s pectoral fin was divided where θ is the angle with respect to the horizon; ij is the
into basal row, the first third of fin adjacent to the body, and amplitude; f is the frequency; n is the nth ray; Į is the phase
distal row, the rest of the fin up to the tip by Klausewitz [11]. lag and it is 48 degrees. The thrust force will be analyzed
The basal part was discovered to be much more rigid than based on this model. There are three steps: (1) find the
distal one. Therefore, the rays are designed to have a pressure distribution over the fin; (2) find the relationship
decreasing thickness from proximal to distal to have larger between the thrust force and total force generated; (3)
deflection at the distal. The fin membrane is made of silicon integrate the pressure over the fin area for thrust force
rubber and attached to the three fin rays. The control signals generated in the fin-water-interaction. The foundation to
for all six fin rays will be generated by an artificial CPG build up the hierarchy in the model is Bernoulli Equation
model. The buoyancy control is essential for gliding. A which is valid at any arbitrary point along a streamline. It
bi-directional pump is utilized to suck in or pump out water so relates dynamic pressure, static pressure and depth.
as to change the weight of the robot and a pressure sensor is 1 2
adopted to detect the depth. Closed-loop buoyancy control is ρ v + ρ gz + p = C (2)
2
therefore implemented for gliding motion control of where C is a constant; v is velocity of the fluid; ȡ is fluid
RoMan-III. density; z is the depth; p is the pressure. Bernoulli Equation is
B. Comparison between RoMan-II and RoMan-III applicable as the fluid is incompressible and friction by
RoMan-III was designed and improved based on viscous forces could be negligible. For our analysis, a point
RoMan-II [10]. Comparisons between these two designs are faraway will be selected, whose velocity is zero, depth is zero.
shown in Table I. It is obvious that the size of RoMan-III is Therefore, C = P0 , P0 is the ambient pressure, 1.01 ×105 Pa.
much more compact while maintaining the velocity.

1731
Based on the direction that two adjacent rays move in, two 1l +a

cases will be analyzed: two rays in the same direction; two Ft = ³ ρ (4ϕ fy ) 2 [ 2 yϕ sin(α 2) cos(2π ft − α 2) ] dy
l2
rays in opposite directions. (12)
1) Two rays in the same direction Simplify the equation above with the dimensions given
When the fin moves, water against its movement will and α = 48D
exert pressure along one side of the fin, l +a
1 Ft = ³ 6500 f 2ϕ 3 y 3 cos(2π ft − α / 2) dy (13)
p1 = C − ρ v 2 − ρ gz (3) l
2 b) Thrust generated by triangular area
Water following its movement will impose pressure along Here, the ratio will be required to obtain the length in the
the other side of the fin form of triangle:
p2 = C − ρ gz (4) l +b b + l − y 1
Since the directions of these two pressures are opposite, F=³ ( ρ v 2 ) h 2 + i 2 dy (14)
l +a b−a 2
the net pressure on the fin will be: l +b b+l − y
1 Ft = ³ 6500 f 2 y 3ϕ 3 cos(2π ft − α / 2 dy (15)
p = ρv2 (5) l +a b−a
2 2) Two rays in different directions
Since the angular velocity of the ray is versatile, average Similarly, the integration of pressure over different area
angular velocity will be used. As the robot will complete can be made. No matter in rectangular or triangular area,
several strokes in a second, average angular velocity can be however, the net thrust will be zero, since a pair of thrusts of
expressed as: the same magnitude and of the opposite directions is
w = 4ϕ f and v = 4ϕ fy (6) generated by the fin.
The thrust angle (ȕ) shown in Fig. 4 is defined as attack As a result, the total thrust force will be
angle of the fin at a certain cross section. Hence the thrust
force will be expressed in terms of force generated by water
Ft = 6500 f 2ϕ 3 cos(2π ft − α / 2) (³l
l+a
y 3 dy
on the fin as, b+l − y 3
l +b l +c l +b b + l − y ·
Ft = F × sin β (7) +³ y dy + ³ y 3 dy + ³ y 3 dy ¸
b−a
l+a l l +c b−c ¹
˄16˅
Based on the values in Fig. 3, Ft can be obtained:
Ft = 17.01 f 2ϕ 3 cos(2π ft − α / 2) − cos(α / 2) (17)

B. Drag
Fig. 4 Projection of fin for thrust angle Reynolds number is defined for a number of different
If the two rays are projected onto the front plane, they will situations where fluid is in relative motion to a surface. For an
form an angle dθ object in fluid, Reynolds number called the particle Reynolds
dθ = ϕ × sin(2π ft ) − ϕ × sin(2π ft − α ) number and denoted as Re p , is important when considering
dθ = 2ϕ × sin(α / 2) cos(2π ft − α / 2) (8) the nature of flow around that grain, whether or not vortex
shedding will occur, and its fall velocity.
Since dθ is small enough in our project, the approximation
uLρ
of dθ ≈ tan( dθ ) is applicable. Thus, h can be expressed as Re p = (18)
μ
h = ydθ (9)
where u is the velocity of approaching stream, and in our
Therefore from the triangle formed by h, i and β , we have
project, it’s 0.29 m/s. ȡ is the density of water which is 1000
h kg / m 3 . The viscosity of water, μ , is 0.001002 kg/(m·s).
sin β = (10)
i + h2 2 Since the plate length is 0.3 meter, therefore the particle
By integrating the pressure over the area, the expression Reynolds number will be 89820. Since the Reynolds number
of the force F generated by the undulating movement of the is large enough, the pressure drag dominates and it is given by
fin can be obtained. Since the area of the fin is irregular, the 1
Fd = − ρ ACd u 2 (19)
integration will be made separately. 2
a) Thrust generated by rectangular area where Cd is the drag coefficient, which is assumed to be
1 2 2 2
l +a 1.28 [12]; A is the reference area; ρ is the water density; u is
F=³ ρ v h + i dy (11)
2 l the relative speed of the object.
The substitution of Eqs. (6), (8), (9), (10) and (11) into (7) The reference area A is often defined as the area of the
yields orthographic projection of the object. For the final design of
our robot, the reference area will be 0.045 m2.

1732
Therefore the drag force expression could be simplified in Fig. 5.Flexible rays oscillate about x-axis, which expands
as: the fin membrane to achieve flapping motion. Based on the
Fd = −28.8u 2 (20) symbols in Fig. 5, the propagating sinusoidal wave generated
From Newton’s second law and integration, u can along the manta ray body can be defined by
therefore be expressed in terms of t and m as: 2π
θ i ( xi , t ) = Ai sin(ωt − Φ d ,i ) = Ai sin(2π ft − xi ) (25)
u (t ) =
1 (21) λ
28.8
t +C where t is the time, xi (i = 1, 2, 3) is the horizontal position of
m the motors, Ai is the amplitude of sinusoidal wave, Ȧ is the
The mass of RoMan-III is 5 kg, and the maximum angular frequency, ĭd,i is the phase difference, Ȝ is the
velocity is 0.29 m/s. Hence C = 1/0.29=3.45 s/m. For the wavelength generated by flapping fins, and f is the flapping
robot reduces speed to 0.01 m/s, the time T required will be T frequency.
= 16.76 s.By integrating velocity with respect to time, the
displacement can be expressed as:
ln(5.76t + 3.45)
D (t ) = + C1 (22)
5.76

When the robot starts to decelerate, the displacement and


time is assumed zero. Therefore,
C1 = − ln 3.45 / 5.76 = −0.215 . As t = 16.76, the displacement
is
ln(5.76 ×16.76 + 3.45) Fig. 5 Din ray arrangement on one side
D= − 0.215 = 0.58m
5.76
This result is consistent with our observations during the B. Buoyancy Control
test where the robot will continue in 0.5 m due to inertia after Water pump and bladder are utilized in buoyancy control
stopping flapping motion. of Roman-III in order to change the overall buoyancy of the
robot for ascending and descending motions. Besides, an
C. Dynamic Equation absolute pressure sensor with the sensing range of 15 kPa to
Based on Newton’s Second Law and relate the thrust and drag 115 kPa is used to provide water depth feedback to the
forces with acceleration, we have microcontroller. The flow rate of the water pump is not
constant as the pressure in water bladder is varying. This will
du cause the variation of ascending or descending speed and lead
m = 34.02 f 2ϕ 3 cos(2π ft − α 2) − cos(α 2)
dt (23) to difficulty of controlling the best time to change the polarity
2 of the water pump. A proportional-derivative (PD) controller
− 28.8u
is essentially a controller which responds to the rate of change
The velocity will be maximized at zero acceleration as of the process error. With the help of PD controller, process
error is taken into consideration and hence reducing the
umax = 1.087 f ϕ 1.5 cos(2π ft − α / 2) − cos(α / 2) (24) maximum overshoot for the desired water depth.
Defining u(t) as the controller output, the final form of the PD
IV. MOTION CONTROL algorithm is:
d
Swimming by flapping pectoral fins, turning and gliding u (t ) = K p e(t ) + K d e(t ) (26)
dt
are the three main locomotion patterns of the manta ray. In
this section, buoyancy control for ascending and descending where Kp is the proportional gain, Kd is the derivative gain, e
motion will also be discussed. is the error (error = desired depth – instantaneous depth), t is
the instantaneous time.
A. Flapping Control
C. Turning Control
For RoMan-III, each side fin is controlled by three
oscillators using three brushless servo motors. Therefore, Turning motion is useful in navigation and obstacle
totally six oscillators are needed to generate the swimming avoidance. The turning or steering locomotion of the robot
gaits for six fin rays. Sinusoidal wave function is commonly can be achieved through unbalanced thrust forces generated
used in oscillating control of fin motion [13, 14]. In order to by two fins. The thrust control can be achieved by applying
achieve a coordination of multiple fin ray motions, a different amplitude and/or flapping frequency to the two fins.
well-defined phase difference between adjacent motors is For example, if the amplitude at the left side is larger than that
controlled. Along the fin chord length, a sinusoidal wave is of the right side, the robot will turn right. The turning radius
generated by three flexible rays at one side of the manta ray depends on the thrust difference between two fins.
robot. The arrangement of a single sided fin rays is illustrated In Equation (27), the negative sign at the phase difference
term implies that oscillation of the former fin ray leads the

1733
back ray and hence the robot performs forward swimming. deduced in Equation (24), the speed against frequency should
Simply changing the sign of the phase difference term to be linear. However, this is not observed in Fig. 7. As the
positive, the robot can perform backward swimming without frequency increases, the servo motor and microprocessor are
turning its body. not able to process so quickly, and the motion of fin cannot
For obstacle avoidance purpose, a distance sensor with follow the frequency exactly due to resistance of water.
effective sensing range of 30 to 80 cm in water, which is Further, relation between velocity square and amplitude can
sufficient enough in our application, is installed at the front be obtained by squaring both sides of Equation (24) as
part of the robot to provide distance feedback to the u 2 = 1.18 f 2ϕ 3 cos(2π ft − α / 2) − cos(α / 2) (28)
microcontroller. This sensor is composed of an integrated
Therefore, polynomial fitting in order three was used. As
combination of position sensitive detector (PSD), infrared
expected, the curve fits most points nicely in Fig. 8, which
emitting diode (IRED) and signal processing circuit.
confirms the validity of the derived model.

43 cm 0.35
Linear fit of 10
Linear fit of 15
5cm 0.3 Linear fit of 20
Linear fit of 25
Linear fit of 30

Velocity (m/s)
0.25
Linear fit of 35
Linear fit of 40
0.2
10 degree
15 degree
Fig. 6 Position of distance sensor 0.15 20 degree
25 degree
Figure 6 shows the position of the distance sensor on the 0.1 30 degree
35 degree
robot. The length of half wingspan is 43 cm while the distance 40 degree
of the sensor placement is about 5 cm from the centre. If zero 0.05
0.5 1 1.5 2
radius turning is achieved, the robot has to perform turning Frequency (Hz)
motion, when it senses an obstacle at the distance of 38 cm or Fig. 7 Effect of frequency on the speed at different amplitudes
more, to avoid collision between the fish robot and the
2

obstacle.
Velocity Square (m/s)

0.08
poly. fit of 0.5
poly. fit of 1.0
D. Gliding Control 0.06
poly. fit of 1.5
poly. fit of 2.0
Underwater gliding motion might not be as fast as normal poly. fit of 2.5
0.5 Hz
swimming by flapping motion, but it consumes much less 0.04
1.0 Hz
energy, which makes the robot more endurable in underwater 1.5 Hz
2.0 Hz
cruising. Such a requirement is particularly crucial for long 0.02
2.5 Hz
distance travelling. In gliding motion, the longitudinal line
0
forms a non-zero attack angle, called gliding angle with the 10 15 20 25 30 35 40 45 50 55

horizontal line. In order to maintain specific gliding angle to Amplitude (degree)


Fig. 8 Effect of amplitude on the speed at different frequencies
support gliding motion, the following equations are used in
gliding motion control: In addition to the swimming speed test, turning radius of
­ bA i = 1, 4 four turning modes of the manta ray robot was also tested to
° determine the best turning method for obstacle avoidance.
și (t ) = ® 0 i = 2,5 (27)
° −bA i = 3, 6 The experiment was conducted by performing a half circle
¯ turning motion, turning radius being estimated as half of the
where Ai is the amplitude of fin ray, while b (= ±1) represents distance travelled by the robot. All parameters of different
the buoyancy status. modes and experimental results are listed in Table II (F
V. EXPERIMENTS AND RESULTS denotes forward and B backward).
As shown in Table II, the smallest turning radius was
Experiments have been conducted to obtain the achieved in Mode 4, while Mode 3 is with a relatively small
swimming speed and turning radius. Testing of swimming turning radius. Thus, turning Mode 4 was selected to be
speed was also conducted to determine the effects of implemented in the turning control for obstacle avoidance
amplitude and frequency. In the experiments, flapping due to its ability to perform sharp turning. On the other hand,
amplitude and flapping frequency were independent variables, single fin turning in Mode 3 could be utilized as energy
while the swimming speed was dependent variable. During saving turning mode for the case that turning space is not
the speed test, the robot will complete two meter swimming in critical.
water tank. The time to complete was recorded and converted Ideally, pivot turning (with zero turning radius) should be
to the speed. Figure 7 shows the experiment results at achieved in Mode 4. However, experimental error will occur
different frequencies and amplitudes, indicating that the as the method used to calculate turning radius was only
maximum speed the robot can achieve is 0.294 m/s. As

1734
estimation. Assumption was made that the manta ray robot Future works should include the design and modeling for
could perform a complete round turning circle but this might better performance on gliding and payload carrying.
not happen in the experiments due to the disturbance. Moreover, some key questions and suggestions raised by
Lauder [15], Flammang and Porter [16] are worth exploring.
TABLE II Four Modes for Turning Radius Test and Results
ACKNOWLEDGMENTS
Left Fin Right Fin Turning
Mode Radius
The authors would like to express their thanks to
Ampl. Freq. Ampl. Freq. Dir
(deg) (Hz)
Dir.
(deg) (Hz) . (m) Wanchao Chi for his help in the paper preparation. Thanks are
also due to Shuo Liu, Yik-Cai Yong, Lei Li, and Jianfang
Mode
20 1.0 F 10 1.0 F 0.84 Xiao for their help in the construction of prototype, model
1
Mode
20 1.0 F 20 0.5 F 0.71
derivation, testing, and data collection.
2
Mode [1] Jana M. Parson, Frank E. Fish, and Anthony J. Nicastro, "Turning
20 1.0 F 0 0.0 F 0.05
3 performance of batoids: Limitations of a rigid body," Journal of
Mode Experimental Marine Biology and Ecology, 402 (2011) 12–18.
20 1.0 F 20 1.0 B 0.01
4 [2] L. J. Rosenberger, "Pectoral fin locomotion in batoid fishes: Undulation
In the experiments, smooth swimming gaits have versus oscillation," Journal of Experimental Biology, vol. 204, pp.
successfully been implemented on the developed manta robot 379-394, 2001.
[3] T. P. L. Brower, "Design of a manta ray inspred underwater propulsive
(RoMan). The steady forward swimming speed reaches 0.8 mechanism for long range, low power operation," M.Sc, Tufts
BL/s. As the robot performs gliding motion, the buoyancy University, 2006.
module is able to control the fish robot slowly moves up and [4] J. G. J Palmisano, R Ramamurti, K J Liu, J Cohen, T Mengesha, J Naciri,
W Sandberg, and B Ratna, "Design, development, and testing of
down in water. Meanwhile, the fins can hold steady postures flapping fins with actively controlled curvature for an unmanned
without flapping. The gliding motion is achievable in the underwater vehicle," in Bio-Mechanisms of Swimming and Flying: Fluid
initial testing (see Fig. 9). The frequency f becomes zero in Dynamics, Biomimetic Robots, and Sports Science, S. K. N Kato, Ed., ed
Berlin: Springer, 2008, pp. 283-294.
this case, which means that oscillators in the CPG model do [5] W. Stoll. (2009). Aqua ray: water-hydraulic manta ray with
not evolve. Note that the amplitude A is set to ʌ/6. flapping-wing drive. Available:
http://www3.festo.com/__C1256D56002E7B89.nsf/html/Aqua_ray_en.
pdf/$FILE/Aqua_ray_en.pdf
[6] K. Suzumori, S. Endo, T. Kanda, N. Kato, and H. Suzuki, "A bending
pneumatic rubber actuator realizing soft-bodied manta swimming
robot," in 2007 IEEE International Conference on Robotics and
Automation (ICRA), Rome, Italy, 2007, pp. 4975-4980.
[7] Y. Cai, S. Bi, and L. Zheng, "Design and experiments of a robotic fish
Gliding down Gliding up imitating cow-nosed ray," Journal of Bionic Engineering, vol. 7, pp.
(a) (b) 120-126, 2010.
[8] L. Zheng, S. Bi, and Y. Cai, "Design and optimization of a robotic fish
Fig. 9. Snapshots of RoMan-III in gliding locomotion: (a) gliding up with
mimicking cow-nosed ray," presented at the IEEE International
positive buoyancy (b=1) and (b) gliding down with negative buoyancy (b=
Conference on Robotics and Biomimetics (ROBIO), Tianjin, China,
-1). Red lines indicate the inclination of fins. For the video of RoMan-III
2010.
swimming performance (visit http://www.youtube.com/watch?v=q8jJ84
[9] Y. Zhong, D. Zhang, C. Zhou, C. W. Chong, T. Hu, L. Shen, and K. H.
y_82k)
Low, "Better Endurance and Load Capacity: An underwater Vehicle
Inspired by Manta Ray," presented at the Fourth International
VI. CONCLUTIONS Symposium on Aero Aqua Bio-Mechanisms, ISABMEC, Shanghai,
China, 2009.
This paper has presented the development of a robot [10] C. Zhou and K. H. Low, "Better Endurance and Load Capacity: An
imitating the movement of manta rays. Based on several Improved Design of Manta Ray Robot (RoMan-II)," Journal of Bionic
Engineering, vol. 7, pp. S137-S144, 2010.
assumptions, the mathematic model of flapping fin for manta [11] W. Klausewitz, "Der lokomotionsmodus der flugelrochen
ray’s propulsion and turning was derived. Relations of (myliobatoidei)," in Natur-Museam und Forschungsinstitude
different parameters were also explored. Improvements on Senckenberg, ed, 1963, pp. 111-117.
[12] Drag coefficient. available: http://en.wikipedia.org/wiki/Drag_
hardware, fin and control were made so that a better prototype
coefficient (accessed: 20 April 2011)
with high reliability was attained, which helped us obtain [13] S. N. Toda Y, Sanada Y, and Danno M., "The motion of a fish-like
better experiment results. For example, the neutral positions under-water vehicle with two undulating side fins.," Proceedings of
of motors were adjusted to the same level so that the the 3rd International Symposium on Aero Aqua Bio-mechanisms,
2006.
synchronization of two fins was achieved and the imbalance [14] M. R. J. Kelly S D, Anhalt C T, Murray R M, and Burdick J W.,
problem was remedied. "Modeling and experimental investigation f carangiform locomotion
It was found from the experiments that the maximum for control," Proceedings of American Control Conference, pp.
1271-1276, 2004.
speed will be achieved, as the frequency and amplitude of [15] George V. Lauder, "Swimming hydrodynamics: ten questions and the
rays are maximized. However, the imperfection of hardware, technical approaches needed to resolve them," Exp Fluids, (2011),
which refers to the servo position and response to the 51:23–35.
[16] Brooke E. Flammang and Marianne E. Porter, “Bioinspiration:
program, might have affected the performance the robotic Applying mechanical design to experimental biology,” Integr. Comp.
manta ray. Pivot turning with small turning radius was Biol., (2011), 51 (1): 128-132.
observed during turning experiments.

1735

You might also like