2019 International Conference on Robotics and Automation (ICRA)

Palais des congres de Montreal, Montreal, Canada, May 20-24, 2019

Compliant four degree-of-freedom manipulator

with locally deformable elastic elements for minimally invasive surgery
Jumpei Arata13 , Yosuke Fujisawa1 , Ryu Nakadate2 , Kazuo Kiguchi1 ,
Kanako Harada3 , Mamoru Mitsuishi3 and Makoto Hashizume4

Abstract— Minimally Invasive Surgery (MIS) is one of the

most successful applications of surgical robots. Although the
introduction of robotic technology has brought a number of
benefits, further advancements in MIS are limited by the size
and bending radius of instruments. In this paper, we present a
compliant four degree-of-freedom manipulator that consists of
elastic elements with partly thinner structures. The proposed
mechanism allows the elastic element to deform locally, thus
minimizing its bending radius while the low number of me-
chanical parts greatly contributes to its compactness. This paper
describes the design strategy, optimization method using FEA,
prototype implementation, and evaluations. The evaluations re-
veal high accuracy and repeat accuracy, which are key elements
for robotic instruments in MIS. Further, the prototype is able
to exert sufficient force and it is possible to perform a simulated
needle insertion task using the manipulator, demonstrating the Fig. 1. EndoWrist needle driver from Intuitive Surgical with a diameter
feasibility of the proposed mechanism. of 8.0 mm (top). Prototype of the proposed mechanism with a diameter of
3.5 mm (bottom). Both instruments have four DOF at the tip.

I. INTRODUCTION
In recent years, surgical robots have been widely adopted Surgical robots have been widely studied to overcome
in clinics, and the range of their applications is expanding. the aforementioned limitations in MIS. The introduction
Minimally Invasive Surgery (MIS) is one of the most suc- of robotic technology has brought a number of benefits
cessful applications of surgical robots. In MIS, thin and tube- such as: 1) miniaturized multi-DOF instruments to increase
shaped surgical instruments are inserted into the patient’s dexterity, 2) inherent high accuracy, enabling hand tremor to
body through incisions on the skin surface together with an be reduced, and 3) further integration with software-based
endoscope. The surgeon manipulates the instruments while assistance such as navigation systems. The most represen-
monitoring the endoscopic view to perform the surgery. Al- tative and successful example of surgical robots in MIS is
though there are great benefits to MIS such as the reduction the da Vinci Surgical System [2], [3]. Although the range
of pain, duration of hospital stay, and medical cost, MIS of applications of da Vinci is expanding, there are still
is known to present technical barriers that surgeons must limitations of the instruments that mean they cannot reach
overcome. The three primary problems of MIS are stated in deep and narrow areas nor perform dexterous manipulations.
[1]. First, there is a discrepancy between the endoscopic view These limitations originate from the size and bending radius
and the surgeon’s hand, which is also known as the “hand-eye of the instruments. A small diameter is crucial for enabling
coordination.” Second, MIS leads to poor dexterity because the instrument to reach the surgical area through a narrow
of the lack of degrees-of-freedom (DOF) on the instruments. and confined path. To properly handle tissues in a confined
Third, no force or tactile feedback is provided for the surgeon cavity, a small bending radius greatly facilitates work in a
through the long handled instruments. Further, as surgical limited space (i.e., a large bending radius means that the
procedures in MIS are substantially and rapidly advancing, instrument requires a large amount of space to execute a
surgeons are required to have up-to-date and proficient skills. bending motion, thus potentially damage the surrounding
tissues).
1 Jumpei Arata, Yosuke Fujisawa and Kazuo Kiguchi are
with Department of Mechanical Engineering, Kyushu University,
One example that requires further advancement in robotic
Japan, 744 Motooka, Nishi-ku, Fukuoka-city, 819-0395, Japan instruments is endoscopic transsphenoidal surgery [4], [5].
jumpei@mech.kyushu-u.ac.jp This surgery is performed to remove a tumor in the pituitary
2 Ryu Nakadate is with Center for Advanced Medical Innovation, Kyushu
University, 3-1-1, Maidashi, Higashi-ku, Fukuoka-city, Fukuoka, 812-8582,
by inserting the instruments through the nasal cavity and
Japan. sphenoidal sinus. The most skill-demanding procedure in
3 Kanako Harada and Mamoru Mitsuishi are with Department of Mechan-
endoscopic transsphenoidal surgery is the suture of the dura
ical Engineering, the University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, matter, which is performed at the end of surgery to avoid
113-8654, Japan.
4 Makoto Hashizume is with Kitakyushu Chuo Hospital, 1-13-1, the leakage of cerebrospinal fluid. The main workspace for
Kawaraguchi, Kita-ku, Kitakyushu-city, Fukuoka, 802-0084, Japan. the surgery is the sphenoidal sinus, which is approximately

978-1-5386-6027-0/19/$31.00 ©2019 IEEE 2663

 (a) 1 DOF bending motion
!"#$ !-#$$2'()*$34)5061$!7# !8#$$2'()*$34)5061$!9#
+,-./0$!9#
7(<** !%#$
+,.-":$
;,-./0 6(<)
**
:=">$;,-./0
!"'()**
!'(6**
+,-./0$!1#
+?">5 !"#$%&'"()$,+.- !"#$%&'"()$*+,-
2./$34./5 !&#$
7(8** !"#$$%&'()*$+,-./01* !(#$:6;801)($!-#$-1($!"#

* 6('**
76(')*

+,-./0$!9#

!"#$%&'"()$/+0- !"#$%&'"()$,+,-

Fig. 2. (a) Proposed mechanism, consisting of four main mechanical parts: spring Fig. 3. FEA was performed to optimize the mechanism. The aim
A, a pair of spring Bs, a shaft, and a pin joint. Springs A and B have partly thinner of the design was to balance the tradeoff between the range of
structures that locally deform ((b) and (c), respectively). motion and rigidity/output.

a sphere of 25 mm in diameter [6]. Although access to the mechanisms, the transformation from linear to rotational
pituitary from the nose is straightforward, the instruments motion is generally done using a mechanical chain of pulleys,
must be dexterously manipulated in an extremely confined links, or gears that increase the number of mechanical
area to perform the suture. Therefore, further improvements elements thus the complexity. An elastic structure enables the
such as reduced size and bending radius are challenging but motion transformation with a fewer number of mechanical
beneficial for the field of surgical robots. elements and without backlash. This characteristic greatly
contributes to the simplicity of the structure, thus enabling
II. RELATED WORK it to be more easily miniaturized. A mechanical structure
The development of robotic surgical instruments (e.g., that uses elastic deformation is also known as a compliant
robotic forceps with multiple DOF) has been widely studied. mechanism [17].
Representative examples can be found in the review papers While these mechanisms with elastic structures provide
[7]–[9]. Challenges in the development of surgical robotic the above advantages, the design process requires particular
instruments are diverse, but there are three primary problems. consideration because of the large deformation that occurs
First, the size of the surgical tool is key to minimizing in the structure. Deformation should be allowed only within
the invasiveness so that it is within the required range the material’s yield strain to prevent breakage or fatigue from
of motion and output. The bending radius must also be repeated deformation. Compared to a thick structure, a thin
considered in the design. Second, the surgical tools must structure enables a larger deformation; however, the rigidity
be bio-compatible, sterilizable, and robust against electrical and output force are lower. Therefore, design is needed
noise. Third, multi-DOF articulation at the tip is preferable in to balance the tradeoff between the range of motion and
most cases. Further, the actuators should be remotely located rigidity/output. In addition, the bending radius is inherently
and detachable for sterilization. larger than that of conventional mechanisms such as pulleys
Recently, the field of soft robotics has become actively because the deformation of the structure must be distributed
explored [10], [11]. There are two soft robotics studies to prevent concentrated strain. In [13], the minimum bending
related to this research. Simaan presented surgical robotic radius was reported as 12.7 mm for the proximal segment
forceps using flexibly deforming tubes connected in multiple and 8.13 mm for the distal segment. Although the bending
stages [12]. The elastic tubes are actuated in push-pull modes radii were not reported in [14] and [15], they are estimated to
to provide a multi-DOF together with graspers implemented be rather large because the deformation is widely distributed
at the tip. The forceps were further implemented in a surgical in these structures.
robotic system [13]. Webster and Dupont presented a multi- In addition to the above related work, the author previ-
DOF surgical robotic tool using a multi-layer pre-curved tube ously presented a 2-DOF manipulator in which four elastic
[14], [15]. Each tube is driven in rotation and translation elements are partly introduced into a rigid body [18]. The
along the long axis, thus providing multi-DOF motion at current work was inspired by this author’s previous work and
the tip. Mechanical design and modeling are further intro- extends the concept to a robotic instrument in the form of
duced and discussed in [16]. Although different kinematic forceps, which are widely used in MIS.
designs were proposed, a notable common feature of these Miniaturization and the reduction of bending radii in
mechanisms is an elastic structure that deforms when it is in robotic instruments are key factors for further advancements
motion to transform the linear input motion from the remote in MIS. In this paper, we propose a mechanism for a
actuators to rotational motion at the tip. In conventional surgical manipulator that consists of multiple elastic elements

2664
that deform when moved. The manipulator consists of four %&' ()*#+),*-&./0/12#!3"400
springs that were designed to deform locally to achieve
further miniaturization with a smaller bending radius than
mechanisms proposed in previous studies.
III. PROPOSED MECHANISM !"#$
A. Mechanical structure !"#"$"%&'()*'+,--
An overview and the structure of the proposed mechanism %5' ()*#+),*-&./0/12#"36700
are illustrated in Figs. 1 and 2, respectively. The mechanism
consists of four main mechanical parts: spring A, a pair of !"#$
spring Bs, a shaft, and a pin joint. Spring A (Fig. 2(b))
consists of a pair of flat springs on their sides. There is a hole
in the center along the long axis where the two spring Bs !"#"$"%&'./)0'+,--
(Fig. 2(c)) are connected through a pin joint and face each
other. As illustrated in Fig. 2(b), when one part of spring A is
pulled and the other is pushed, the tip bends bi-directionally Fig. 4. Rigidity was analyzed using FEA by applying a load at the tip
from different angles (perpendicular to spring A and B).
(α axis in rotation). For one spring B, the proximal spring
(the flat spring in Fig. 2) is bent according to the deformation
of spring A to transmit linear movement to the distal part. for the modeling. As a result, we obtained the optimized
When the proximal end of one spring B is actuated (push design shown in Fig. 2, which satisfies the required range
or pulled) along the long axis, the spiral spring part (the of motion of 120 deg in both 2-DOF bending motions (the
distal part of the spring shown in Fig. 2) deforms around the maximum strain is 4.2% in Fig. 3(a) and is 6.4% in Fig. 3(b))
pin joint. This enables the linear input displacement to be and a range of motion of 120 deg for the grasping angle (the
translated into a rotational motion that can be used for the maximum strain is 5.3% in Fig. 3(c)) within the range of
bi-directional rotational motion of each gripper. Driving the spring recovery strain (7%).
two spring Bs in different directions results in a bending To verify the design, the rigidity was tested using FEA
motion (β axis in rotation), whereas driving them in the by monitoring the displacement when 10 N of load was
same direction results in a gripper open/close motion (γ applied at the tip with two different algles (perpendicular
axis in rotation). Rotational motion along the long axis is to A and B), as shown in Fig. 4. These results reveal a
performed by actuating the shaft with a remotely located high rigidity relative to conventional wire or link mechanisms
actuator. Therefore, the manipulator has four DOF at the tip. that is a result of the relatively simple and dense structure,
The design concept employed in this study is to modify demonstrating the feasibility of this design.
the spring deformations by changing the thickness of the
material. Although the structures of springs A and B are C. Kinematic Model
monolithic, a part of the material is designed to be thinner
The kinematic model of the proposed design is derived
to concentrate the deformation locally, thus reducing the
by defining the curvature of the springs as a circular arc, as
bending radius. As mentioned in Section II, design is needed
shown in Fig. 5.
to balance the tradeoff between the range of motion and
1) α axis by spring (A): As shown in Fig. 5(a), let the
rigidity/output while taking into consideration the material’s
bending angle of spring A be α , bending radius be D,
yield strain to prevent breakage or fatigue. To design the
distance between spring A and the center axis be R, and
mechanism, we conducted a series of finite element analyses
the thickness of flat spring B be t. The two distal ends of
(FEAs).
spring A are actuated by rigid links 1 and 2 in the long axis
B. FEA while both spring Bs are actuated by rigid links 2 and 3. The
A series of FEAs was conducted to optimize the structure displacement of the rigid links from the initial position for
given the following requirements: motion around the α axis are denoted by lα 1 , lα 2 , lα 3 , and
• Ni-Ti as the material for springs A and B (Young’s
lα 4 for rigid links 1, 2, 3, and 4, respectively. Because we
modulus: 5,000 kgf/mm2 , yield strength ∼ 237 MPa) model the curvature as a circular arc, we obtain
and stainless steel for the other parts; lα 1 = Rα (1)
• an external diameter of 3.5 mm and diameter of 3.0 mm
lα 2 = −Rα (2)
for the Ni-Ti material;
• a range of motion of ± 60 deg for the 2-DOF bending We assume that both spring Bs also deform as a circular arc
and 1-DOF grasping motion; according to the motion around the α axis.
• a maximum strain below 7%
t
FEA was performed using mechanism analysis software lα 3 = α (3)
2
(DAFUL 6.1, Virtual Motion Inc., Korea) together with t
meshing software (FEMAP 10.0.2, Siemens Inc., Germany) lα 4 = − α (4)
2

2665
 !(#5%%

!"#$%%&
6%8
1

'((#'%%& )((#*%%&
0
# $ %&' 312/2,1-/4&%1210
!(#5%%
# !(#$%%
-
" !
+(,#"-'. 5#5%%&
+(,#"-'/
)$#(5%%& +,-./0&%1210
!"#$#%&'()*#$#)"

2#-#3'&#"4*
Fig. 6. Prototype, consisting of a tool unit and an actuator unit. The
springs are made of Ni-Ti, using electro discharge machining while the other
())*))+)), mechanical parts were fabricated from stainless steel, including the pin and
shaft. The tool unit is detachable from the actuator unit for sterilization.
678 1 698 1
%./ 2 %.0 %30 %3/
Note that the proposed mechanism needs four linear
5 4 actuators. The proposed mechanism thus has an extra DOF
0 0
along the long axis (i.e., when all four linear actuators move
1 1 in the same direction, the tip of the manipulator, without the
shaft, moves along the long axis). Although this extra DOF
%./ !"#$#%&' %3/ 5 %30 is not currently taken into consideration in the kinematic
%.0 ()*#$#)"
model, it can be used to change the bending radius in real
time in a future implementation.

Fig. 5. (a) Overview of the kinematic structure bending with respect to the
IV. PROTOTYPE IMPLEMENTATION
α axis, while (b) and (c) show bending with respect to β and γ , respectively. Figure 6 shows a prototype of the proposed design. The
prototype consists of a tool unit and an actuator unit. The
2) β and γ axis kinematics for spring B: Let the move- springs are made of Ni-Ti, mainly fabricated by electro
ment around the β axis for rigid links 3 and 4 be lβ 3 and discharge machining. The other mechanical parts, including
lβ 4 . The motion around the β axis is illustrated in Fig. 5(b). the pin and shaft, were fabricated from stainless steel. The
Then, lβ 3 and lβ 4 can be derived as pin was rigidly welded by laser to spring A, and the tool
unit is detachable from the actuator unit for sterilization.
lβ 3 = rβ (5) The actuator unit consists of four linear motors (Maxon
lβ 4 = −rβ (6) Spindle Drive RE8 MR TypeS GP8S, Switzerland) and a
rotation motor (Maxon Motor DCX10L ENX10 GPX10,
In addition, we determine lγ 3 and lγ 4 as the displacements Switzerland). The four linear motors are aligned in parallel in
due to movement around the γ axis, as shown in Fig. 5(c). a cylinder which can be rotated by a rotational motor with
γ a timing belt. The control system was implemented using
lγ 3 = r (7)
2 MATLAB (Mathworks, US). A PID position control was
γ
lγ 4 = r (8) implemented and follows position commands given based
2 on the kinematic model described in the previous section.
3) Overall kinematic model: We finally obtain the overall The control frequency was configured at 1 kHz.
kinematic model as the sum of the displacements for the α ,
β , and γ axes. V. EVALUATIONS
= Rα l1 (9) To mechanically verify the feasibility of the proposed
= −Rα l2 (10) mechanism, a series of tests were conducted on accuracy
t γ and grasping force. In addition, a simulated suturing task
l3 = α + rβ + r (11) trial was performed to verify the overall performance.
2 2
t γ
l4 = − α − rβ + r (12) A. Accuracy
2 2
As shown in the equations, the α , β , and γ axes are To evaluate the accuracy, a periodic motion was performed
coupled and not independent from each other. This is because on the prototype and recorded by a camera (α 99II, Sony
spring B should be bent along with the bending of spring A, Corporation, Japan) with a macro lens (SAL100M28, Sony
as expressed in Equations (3) and (4). Corporation, Japan). The obtained image was processed by

2666
 :5+2 ;<=19>8-3/,+?.>/ ;.?@19>8-3/,+?.>/
in the result. Figure 8 shows the experimental results for the
α and β axes, both with the compensation. The accuracy
(#
was 0.94 deg on average (α : 1.00 deg, β : 0.94 deg, and γ :
)*+,-./01+/023145306 '#
0.89 deg). The test results demonstrate that the prototype
&#
%#
has a fairly high accuracy and can be used for surgical
$#
procedures. In addition, the repeat accuracy was found to
"# be 0.04 deg on average (α : 0.03 deg, β : 0.04 deg, and γ :
# 0.05 deg). The high repeat accuracy was likely achieved by
!"# the backlash-free structure thanks to the elastic elements.
# ' "# "' $# $'
7.8314,396
In surgical procedures in MIS, the surgeon manipulates the
instruments through a master device while monitoring the
Fig. 7. Experimental results of the accuracy test for the γ axis (grasping). endoscopic view. Thus, the repeat accuracy is more important
The compensation substantially reduces the error caused by the discrepancy than the absolute accuracy. Overall, the experimental results
between the kinematic model and actual porotype. positively showed the feasibility of the design concept.
942*1 :2+;1< By further processing the recorded images, it was found
"# that the bending radius around the α axis was approximately
$# 3.5 mm. Because spring B is fixed at the pin joint, it
()*+,-./0*./120342/5

%#
behaves as a pulley. The bending radii for β and γ are thus
&#
#
1.35 mm. This suggests that the proposed mechanism has a
!&# significant advantage over previously presented instruments.
!%# The recorded motion is also shown in the attached media
!$# file.
!"#
# ' &# &'
6-7203+285 B. Grasping force measurement

:53+2 ;3,<2=
The experiment was performed to evaluate the output force
of grasping. The grasping motion was performed on a force
(#
sensitive register sensor (FlexiForce A201-1, Tekscan Inc.,
'#
)*+,-./01+/023145306

US) while the prototype and the sensor were fixed on a

&#
%#
table. A thin metal plate was fixed on the force sensitive
$#
register sensor to stably convert the grasping force into the
"# pressure measured by the sensor, as shown in Fig. 9. We set
# the grasping angle to 0 deg when the grasper was contacting
!"# the sensor without load. The input grasping angle was then
# ' "# "' $# $'
7.8314,396
gradually decreased to exert more grasping force. Note that
this extra grasping angle is not usual in a rigid system, but is
Fig. 8. Experimental results of accuracy test for the α axis (upper) and β possible only because of the inherent flexibility of an elastic
axis (lower), both with the error compensation method. structure.
Figure 10 shows the experimental result. It was found that
the grasping force gradually increased as the input angle was
software to track the motion (VW-H2MA, Keyence Corpo- increased from 0 deg to 50 deg, then saturated at around 60
ration, Japan). The resolution of the image tracking was deg. The output force was stable at around 5.0 N for input
13.5 µ m/pixel for a 2,000 × 2,000 image. Note that the grasping angles greater than 60 deg. The result shows the
evaluation was performed on the motion in each DOF and prototype is capable of 5.0 N of output force at maximum.
not on the combined motion because of the limitations of The result also suggests that the grasping force can be
our measurement system. This issue is further discussed in modulated by adjusting the extra grasping angle thanks to
the last section. the elastic structure.
As a representative result, the grasping motion is shown in
C. Suturing task trial
Fig. 7. In the initial experiment, there a maximum of approx-
imately 9.0 deg of error was found. It was assumed that the A suturing task was performed to verify the feasibility of
error was caused by the discrepancy between the kinematic the prototype. A silicone rubber sheet (0.1 mm in thickness)
model and actual deformation of the elastic elements. We was fixed and a simulated suture task was performed using
thus introduced a simple error compensation method based the prototype. The bending angle of prototype was fixed so
on the spline function to reduce the error. The parameter was that it grasped a needle and was manually held at the actuator
extracted from the error without compensation and applied unit to perform the task. Note that the surface of the graspers
to the kinematic model in a feedforward manner. The error was processed to prevent slips using a method commonly
was significantly reduced after the compensation, as shown used for surgical instruments.

2667
 &
()*+"&,"-,)*

*+,-./01234+5622789
%

"

!
! '$! ')! '(! '"#! '"&!
!"#$%&'%$#" :0.;<21+,-./012,01=627>619

Fig. 10. Experimental results showing that the grasping force gradually
Fig. 9. Grasping force measured by a force sensitive register sensor with
increases from an input angle of 0 deg to 50 deg, then saturates at a value
the prototype and sensor fixed to a table.
of approximately 5.0 N at around 60 deg.

mechanism is the small number of mechanical parts at the

!"# $"# %"# tip. This allows the mechanism to be further miniaturized,
and we are currently working on a new prototype with further
miniaturization. The simplicity will be also useful for making
&"# '"# #"# the structure sterilizable.
A limitation of the proposed mechanism is possible break-
age or fatigue due to the repeated deformation. Because the
deformation occurs locally, extra consideration is required
for fatigue prevention. Our preliminary test revealed that the
Fig. 11. The pictures show a simulated needle insertion task performed by prototype could bend 8,000 times (on the β axis, where
the prototype in time series. The numbers on the top right corner in each
picture show the time ordering. Prototype successfully performing the task the angle varied from -30 to 30 deg at 1,200 times/hour).
without encountering difficulties. This suggests that the mechanism is fairly feasible, but
further improvement in design optimization will strengthen
its advantages. In addition, the accuracy evaluation was
Figure 11 shows the experimental result: the prototype conducted only on each DOF and not on the combined
successfully performed the needle insertion without encoun- motion, although combined motion is feasible on the current
tering difficulties. Needle handling is one of the hardest tasks prototype, as shown in the attached multimedia file. As it
in surgery because of the required output force and rigidity; was revealed in FEA, combined motion will increase the
hence, the result suggests that the prototype is suitable for strain compared with motion in an individual DOF. We
general procedures in MIS. plan to further evaluate combined motion by introducing an
experimental setup including multiple cameras that will allow
VI. DISCUSSIONS AND CONCLUSIONS us to measure accurately in three dimensions.
In this paper, we proposed a four DOF manipulator that The contribution of this paper lies in the mechanical
has articulations driven by elastic structures. We described design concept of using elastic and monolithic elements
the design strategy, optimization method using FEA, pro- that contain thinner parts to locally induce deformation. The
totype implementation, and evaluation. The evaluations re- motion conversion using the elastic elements was found to
vealed a fairly high accuracy and high repeat accuracy, which be an effective way to simplify the mechanism, which helps
are key elements for robotic instruments in MIS. Further, the when reducing the size of the instrument. Further, local
prototype is able to exert sufficient force and it was feasible deformation was found to be an effective way to reduce
to perform a simulated needle insertion task. the bending radius, which is a key parameter for robotic
A notable feature of the proposed manipulator is that a part instruments in MIS. The presented design strategy can be
of the structure is thinner than other parts and the thinner part applied to other robotic applications that require a simple
deforms effectively to transform the input motion into the structure with multi-DOF motion.
desirable output motion. Because the springs are designed Currently, our collaborators are developing a surgical
and optimized to deform locally, the bending radius, which robotic system that is equipped with a master device and
is a crucial parameter of the mechanism, could be reduced. a slave robot that includes the presented device.
An analysis of test motion images revealed that the bending
ACKNOWLEDGEMENT
radius is 3.5 mm for α and 1.35 mm for β and γ . This
suggests that the prototype has a significant advantage over This work was funded by ImPACT Program of Council
other mechanisms for manipulation in confined spaces such for Science, Technology, and Innovation (Cabinet Office,
as the sphenoidal sinus. Another advantage of the proposed Government of Japan).

2668
 R EFERENCES [14] R. J. Webster, J. M. Romano, N. J. Cowan, Mechanics of precurved-
tube continuum robots, IEEE Trans on Robotics and Automation,
[1] A. J. Madhani, G. Niemeyer, J. K. Salisbury Jr., The Black Falcon: Vol.25(1), pp.67-78,2009.
A Teleoperated Surgical Instrument for Minimally Invasive Surgery, [15] P. E. Dupont, J. Lock, B. Itkowitz, E. Butler Design and Control
Proc. IEEE Int. Conf. Intelligent Robots and Systems, pp.936-944, of Concentric-Tube Robots, IEEE Trans Robotics and Automation,
1998. Vol.26(2), pp.209-225, 2010.
[2] C. Freschi, V. Ferrari, F. Melfi, M. Ferrari, F. Mosca, A. Cuschieri, [16] R. J. Webster, B. A. Jones, Design and Kinematic Modeling of
Technical review of the da Vinci surgical telemanipulator, Int. J. Constant Curvature Continuum Robots: A Review, Int. J Robotic
Medical Robotics and Computer Assisted Surgery, Vol.9, pp.396-406, Research, Vol.29(13), pp.1661-1683, 2010.
2013. [17] L. L. Howell, Compliant mechanisms, Wiley-Interscience, ISBN 978-
[3] Intuitive Surgical, da Vinci, https://www.intuitive.com/ 0-471-38478-6, 2001.
(Sept. 2018) [18] J. Arata, Y. Saito, H. Fujimoto, Outer Shell Type 2 DOF Bending
[4] H. Jho, Endoscopic endonasal transsphenoidal surgery: experience Manipulator using Spring-link Mechanism for Medical Applications,
with 50 patients, J. Neurosurg, Vol.87, pp.44-51, 1997. Proc. Int. Conf. Robotics and Automation, pp.1041-1046, 2010.
[5] H. Jho, Endoscopic transsphenoidal surgery, J. Neuro-Oncology,
Vol.54, pp.187-195, 2001.
[6] D. Sareen, A. K. Agarwal, J. M. Kaul and A. Sethi, Study of Sphenoid
Sinus Anatomy in Relation to Endoscopic Surgery, Int. J. Morphol,
Vol.23(3), pp.261-266, 2005.
[7] G Dogangil, B. L. Davies, F. R. Baena, A review of medical robotics
for minimally invasive soft tissue surgery, Proc. IMechE, Vol.224 Part
H: J. Engineering in Medicine, pp.653-679, 2009.
[8] P. Kazanzides, G. Fichtinger, G. D. Hager, A. M. Okamura, L. L.
Whitcomb, R. H. Taylor, Surgical and Interventional Robotics, IEEE
Robotics and Automation Magazine, pp.122-130, June 2008.
[9] R. H. Taylor, D. Stoianovici, Medical Robotics in Computer-Integrated
Surgery, IEEE Trans Robotics and Automation, pp.765-781, 2003.
[10] D. Rus, M. T. Tolley, Design, fabrication and control of soft robots,
Nature, Vol.521, pp.467-475, 2015.
[11] M. Manti, V. Cacucciolo, M. Cianchetti, Stiffening in Soft Robotics:
A Review of the State of the Art, IEEE Robotics and Automation
Magazine, pp.93-106, Sept 2016.
[12] N. Simaan, R. Taylor, P. Flint, A Dexterous Systemfor Laryngeal
Surgery, Proc. Int. Conf. Robotics and Automation, pp.351-357, 2004.
[13] N. Simaan, K. Xu, A. Kapoor, W. Wei, P. Kazanzides, P. Flint, R. H.
Taylor, Design and Integration of a Telerobotic System for Minimally
Invasive Surgery of the Throat, Int. J. Robotic Research, Vol.28(9),
pp.1134-1153, 2009.

2669