Você está na página 1de 6

Proceedings of 2018 IEEE

International Conference on Mechatronics and Automation


August 5 - 8, Changchun, China

A Transient FEA-based Methodology for Designing Soft


Surgical Manipulators
Mohamed Elkeran1 and Mohamed Fanni1,2
1
Department of Mechatronics and Robotics Engineering, Egypt-Japan University of Science and Technology, Egypt.
2
on leave: Department of Production Engineering and Mechanical Design, Faculty of Engineering, Mansoura University, Egypt.
{mohamed.elkeran & mohamed.fanni}@ejust.edu.eg

Abstract - Traditionally, any new development in soft robotics surgery. Despite the many gains of MIS for the patients, some
has to come first through many trial-and-error experiments. Many surgeons prefer to proceed with traditional open surgery
researchers tried to develop analytical models either to study techniques due to minimal access. The surgeon hand can
certain phenomena in an already existed deigns or to use them to accomplish agile tasks freely rather than the MIS tools that
enhance soft robots performance. However, there is no general
lack dexterity and maneuverability. The traditional rigid tools
methodology that can be used to design any arbitrary soft robot
from scratch. Such methodology should have the ability to test applied for MIS have limited number of degrees of freedom
different operation scenarios for checking certain control (DOFs) besides being not easy to move and control. This in
approaches or certain types of sensors and actuators. This paper turn requires more body interventions and more than one
introduces a general tool for developing soft surgical manipulators, surgeon to operate synchronously, which needs more practice
which is based on transient finite element analysis (FEA). To verify and steep learning curve. These drawbacks of MIS have
the feasibility of the proposed approach, experiments have been motivated new emergence of robotic platforms in surgery
conducted on a fabricated soft manipulator. The test setup [4, 5].
incorporates pressure sensors and vision tracking system to record
Conventional rigid robots have been widely applied for
the manipulator trajectory. Air pressure measurements are used as
input to FEA for transient simulation. The experimental results
MIS [5]. They offer high accuracy, however they have certain
demonstrate high level of agreement with FEA results and prove limitations due to lack of dexterity which cause some surgical
that the suggested approach is promising to be used in designing tasks impossible to carry out. Moreover, if they were not well-
new soft surgical robots. controlled by the surgeon, they would cause injuries to the
patient soft tissues. To overcome the lack of dexterity issue,
Index Terms – Soft robot, fiber-reinforced actuator, minimally some researchers used hyper-redundant manipulators and
invasive surgery, finite element analysis.
continuum-like robots for providing large number of DOFs
[6]. In spite of increased DOFs, hyper-redundant manipulators
I. INTRODUCTION
are not straightforward to control, yet they poses drawbacks of
Soft robotics has become recently rich and prosperous unsafe rigid robots. Soft-bodied robots can provide high
research field with many applications and designs [1]. flexibility, dexterity, and safety when applied in MIS [3, 7].
Safety, compliance, maneuverability, and squishiness are They are inherently compliant and when they come to contact,
prominent features of soft robotics. Traditional robots are they introduce gentle force as they are more likely to deform
comprised of rigid links and joints, whereas soft robots are and absorb energy. This in turn decreases the possible harm of
made from highly elastic material like silicone rubber [2]. approaching or even touching delicate structures such as
When it comes to interact with a human being, traditional organs and tissues.
rigid robots have to be kept in isolated or fenced areas like in Throughout different designs and actuation techniques for soft
industrial environments. Safety precautions must be taken for robots, flexible fluidic actuators (FFAs) have been widely employed
safe interaction with such rigid robots. Despite their as a source of actuation, especially in MIS [8, 9]. They allow
accuracy, speed, and productivity, rigid robots are not fully different patterns of motion depending on their design, arrangement,
safe in human interaction situations. Soft robots are paving a and actuation sequence. The primary motions of FFAs involve
new and safe way to cope with human centered interaction elongation and omnidirectional bending. Air or water pressure are
applications. They are ideal for personal robots that interact the main input source to FFAs, which make them safe and
without causing harm to people, service robots, robots convenient for medical application. This advantage of FFAs qualifies
working in unstructured environments, and medical robots, them over traditional electrically energized actuators such as motors,
especially for use in surgery [3]. shape memory alloys, and piezo-electric actuators. FFAs are easy to
Minimally Invasive Surgery (MIS) is the most recently fabricate, have very simple design, and encompass few parts to
applied strategy in surgical operations. It has many advantages manufacture and assemble. Among different available FFAs, the
over the traditional open surgery, which involves reduced three chamber FFA [3, 7, 8-10] seems to be the most vastly applied
patient pain, low cost, less bleeding and infection, and fast actuator for soft robot actuation in MIS application. Since the soft
convalescing. During the operation of MIS, only tiny skin robot is composed of highly deformable elastomeric material, it
incisions are required to access the patient body to conduct the tends to expand upon actuation and may undergo ballooning [9].

978-1-5386-6075-1/18/$31.00 ©2018 IEEE 568


The design presented in Ref. [10] is adopted throughout this work
due to its advantage of embedded fibers around each chamber to
hamper the radial distortion and prevent ballooning.
Ref. [11] thoroughly reviewed the use of constant curvature
kinematics for continuum manipulators. Constant curvature
assumption simplifies the modeling of continuum robots and can
provide good approximations for its movements. However, this
approach exhibited large errors in cases of applied forces and
dynamic response. Researchers in [12, 13] utilized static FEA
for design and optimization of soft surgical manipulators. They
studied the shape of pneumatic chambers and its effect on
ballooning behavior and bending of the manipulator. Despite
that, they did not investigate the application of fiber-
reinforcement in their FEA studies. On the other hand, static
FEA cannot be used to check certain control approach or certain
operation scenario, which limit its use as a general design
methodology. Ref. [14] introduced a lumped model using
Lagrangian dynamics which represents the soft manipulator as a using a stiffer material at the top is to prevent top surface from
series of infinitesimally rigid parts combined with springs and bulging as the air is pressurized inside the fluidic chambers.
dampers. That approach is promising and suitable for control The top surface has to be flat during the application of the
applications, however, it requires experimental estimation of pressurized air. Since the top surface center will be considered
varying soft manipulator’s stiffness and damping parameters. as the origin of the frame attached to the manipulator end-
This experimental estimation is dependent on the fabricated soft effector, its location w.r.t. the top surface must remain fixed.
manipulator and cannot be assessed beforehand. Research work This issue is of great importance because the sensory feedback
by [15] presented a geometry deformation model for soft requires no deformation at the top surface upon actuation.
continuum manipulators with fiber-reinforcement. The The manipulator has a cylindrical shape with a diameter of
developed model is comprehensive yet complex including many 25 mm and a total length of 70 mm as depicted in Fig. 1. The
assumptions in order to realize a better accordance to the three fiber-reinforced FFAs have outer diameter of 8 mm and
experimental work. Each of the previous approaches has its inner diameter of 4 mm. The normal distance between each
certain assumptions and cannot be easily adapted to FFA’s outer wall and the nearest manipulator’s outer surface is
accommodate new design modifications. The aim of this study is 1.75 mm. A helical single fiber of 0.36 mm diameter is wound
to employ transient finite element analysis (FEA) as a general around each FFA at a tight pitch, whereas each turn touches its
tool for designing and inspecting the dynamic performance of predecessor. The reinforcement fibers are typical nylon knitting
soft surgical manipulators. This will be demonstrated through threads. These fibers prevent each FFA from radial expansion
validation of transient FEA results against experimental work for and the result is elongation upon air actuation. The elongation of
good agreement. one FFA causes a bending motion of the whole manipulator.
This paper is structured as follows. Section II presents the Since the manipulator has three FFAs, it can accomplish
soft manipulator design mentioning all dimensions and general omnidirectional bending and/or elongation. The fixed base
layout. Section III describes the used materials and their coordinate system is selected whereas the x-axis passes through
properties along with the fabrication process. FEA is illustrated the center point of FFA1 as shown in Fig. 1.
in section IV with elaboration of meshing and boundary
conditions. Experimental results are discussed versus FEA III. MATERIAL CHARACTERIZATION AND FABRICATION
results in section V and conclusion is outlined in section VI.
A. Material Modeling
II. MANIPULATOR DESIGN The soft surgical manipulator is fabricated using
Ecoflex 00-30 silicone rubber and Dragonskin 30. These
The main structure for the surgical manipulator throughout materials undergo stress-strain behavior of hyperelastic
this work comprises three fiber-reinforced FFAs contained in a materials. Assessing the mechanical characteristics of these
soft silicone rubber enclosure as shown in Fig. 1. These materials has been previously carried out in [13]. The third-
actuators are equally distributed at an angle of 120° between order Yeoh model was proved to be the best constitutive
each other. The manipulator has two parts, namely part I and model for Ecoflex 00-30, while the second-order Yeoh model
part II. Part I is the main body of the manipulator and is made of was the best constitutive model for Dragonskin 30. The nth
Ecoflex 00-30 silicone rubber [16] which is a very soft and order Yeoh material model is represented by (1), where U is
biocompatible material. Part II establishes the top of the the strain energy potential, I1 is the first deviatoric strain
manipulator and is made of Dragonskin 30 silicone rubber [17] invariant and Ci0 is a material specific parameter. For Ecoflex
which has higher stiffness than Ecoflex 00-30. The role of 00-30, the material model parameters were determined to be:

569
C10 = 5072 J·m-3, C20 = –331 J·m-3, C30 = –15 J·m-3, and demolded, sealed at bottom, and thin latex tubes are
N = 3. For Dragonskin 30, the material model parameters were introduced to manipulator base (Fig. 2f). These tubes are the
determined to be: C10 = 1190 J·m-3, C20 = 23028 J·m-3, and means through which air is pressurized into each FFA.
N = 2.
(1)
V. FINITE ELEMENT ANALYSIS
B. Soft Robot Fabrication This section describes the finite element analysis (FEA)
The fabrication process of the soft manipulator is procedure using ANSYS software. Fig. 3 shows the manipulator
conducted through subsequent steps of casting as shown in environment. The manipulator assembly (Fig. 3a) is composed
Fig. 2. Both Ecoflex 00-30 and Dragonskin 30 come into two of eight parts (Fig. 3b to Fig. 3e). The inner tubes (Fig. 3b) are
parts A and B [16, 17]. Each material’s part A and part B are fitted inside the manipulator’s body (Fig. 3d) with helical
to be mixed together at a volumetric ratio of 1:1. After mixing fibers (Fig. 3c) in between. After that, the manipulator’s top
the two parts, air bubbles are formed inside the mixture. These part is mated over the manipulator’s body. Inner tubes,
air bubbles have to be removed in order to attain good manipulator’s body, and manipulator’s top part are modeled as
mechanical properties of the silicone rubber. The mixture solid bodies. They are meshed with SOLID186 element type
should be introduced into a vacuum chamber at around –0.9 which has quadratic displacement behavior and mixed u-p
bar to dispose all the entrapped air bubbles. After that, the formulation for simulating incompressible hyperelastic
mixture is poured carefully to the mold and left to be cured. materials. Helical fibers are modeled as beams with
Curing times for Ecoflex 00-30 and Dragonsking 30 are 4 and BEAM188 element type.
16 hours, respectively. The mesh sizes are globally selected as 2 mm for
The fabrication process begins with wrapping a helical manipulator’s body and top part, 1.5 mm for inner tubes, and
fiber around three 8 mm rods at a close pitch. The three rods 1 mm for helical fibers. The solid bodies are meshed with a
are then inserted in the mold (Fig. 2a) for the first silicone congruent mesh, whereas all neighboring nodes from different
casting process. A mixture of Ecoflex 00-30 parts A and B is parts are merged. This results in enhanced performance, fast
prepared in a cup, degassed in a vacuum, and then poured convergence, and lower computation times. Displacement
gently into the mold. After pouring is completed, the mold is constrictions are induced between the helical fibers and the
put in a vacuum again to remove any air bubbles that may have outer surfaces of inner tubes. These constrictions are
been formed during the pouring step. The cast silicone is then introduced by using constraint equation interfaces between
left for full curing (Fig. 2b) at room temperature. Up to this pairs of nearby helical fibers nodes and inner tubes meshing
stage, the main body of the soft manipulator enclosing fibers is elements.
formed. Once curing is completed, the three 8 mm rods are The whole manipulator assembly is set fixed at its bottom
taken away very gently so as not to deteriorate the cured surface. Air pressure is applied to the inner surface of inner
silicone, nor the fibers. Then, three 4 mm rods are inserted in tubes and the bottom surface of manipulator’s top part that faces
the mold retaining the cured silicone rubber (Fig. 2c). An the inner tubes. The transient structural module is used for
additional Ecoflex 00-30 layer is injected between the rods and the dynamic simulation in ANSYS workbench. Material densities
cured silicone (Fig. 2d) and left for curing. After that, the three rod are 1070 kg/m3 for Ecoflex 00-30, 1080 kg/m3 for Dragonskin
are pushed down slowly until their top surfaces touch the cured 30 and 1150 kg/m3 for nylon fibers. Helical fibers are modeled
silicone top surface. A thick layer of Dragonskin 30 is then poured using linear isotropic elasticity with E = 3.5 GPa and ν = 0.25.
to the mold top (Fig. 2e). Eventually, the soft manipulator is

570
drive the solenoid valves accordingly.
In order to track the manipulator’s motion, a fiducial marker
is glued to its top. This marker is then detected from a captured
image taken by a camera at a resolution of 1280×960. The
marker detection is carried out using Aruco library [18] which is
implemented in C++ based on OpenCV library. After detecting
the fiducial marker, its pose is estimated accordingly. This
requires camera calibration procedure beforehand. Both pressure
sensors readings, marker detection, and pose estimation are
synchronously acquired to the host PC. As pressure control
signals are sent from the PC to the Arduino board, the solenoids
are actuated and pressure sensors record the instantaneous air
pressure inside each FFA. The pressure sensors measured data
are saved in order to be used later in finite element simulations.
Two coordinate frames are attached to the manipulator as
V. EXPERIMENTAL WORK depicted in Fig. 4. The first one is fixed to the center of the
manipulator’s base and is denoted by {X0, Y0, Z0}. The second
This section establishes for the experimental assessment of one is congruent to the center of the manipulator’s top face and
the soft manipulator dynamic performance versus the is denoted by {X1, Y1, Z1}. During experiments, the coordinate
numerical simulations using transient FEA. After completing frame {0} remains fixed, while the coordinate frame {1} moves
the fabrication process, the soft manipulator is connected to a with the manipulator’s top face.
fixed based and held in a setup for conducting experiments.
B. Transient Response of One Actuated FFA
A. System Setup In this experiment, air is actuated though FFA1 using
The experimental setup is shown in Fig. 4. It consists of five solenoids 1 and 2. Fig. 5 shows the input pressure as a function
main subsystems: air compressor, solenoid valves, pressure of time. Fig. 6 gives X, Y, and Z components as well as the total
sensors, control circuitry, and visual motion tracking. The displacement of experimental and FEA responses. That figure
Arduino Mega 2560 board is used for interfacing different shows good agreement between FEA and experimental data for
subsystems to the host PC through USB communication. Three the soft manipulator. The minimum difference of total
pairs of normally closed 2/2 solenoid valves are employed for displacement is 2 mm. This represents 3% of the manipulator’s
controlling the air pressure inside each FFA. Each pair is original length and occurs when the instantaneous pressure is
targeted at one FFA, whereas one valve delivers the pressurized almost constant (between 1.1 s and 1.6 s). The maximum
air and the other evacuates the entrapped air upon actuation. difference of total displacement is 10 mm (14%), which happens
Pressure control signals are sent from the host PC through at rapid pressure changes. Fig. 7 shows the shapes of the soft
Arduino board to the MOSFET amplifier circuit to manipulator at two different static pressures, namely 0.5 and 1
bar, for both experimental study and finite element simulation.
That figure proves shape similarity of the soft manipulator for
both experimental and finite element results.

Fig. 5 Input pressure as a function of time.

C. Transient Response of Two Actuated FFAs


In this experiment, air is actuated though FFA1 and FFA2
using solenoids 1, 2, 3, and 4. The air pressure is the same as
shown in Fig. 5. Fig. 8 depicts X, Y, and Z components as well
as the total displacement of the experimental and FEA

571
Fig. 6 Transient response for one actuated FFA. Fig. 8 Transient response for two actuated FFAs.

P = 0.5 bar P = 0.5 bar

P = 1.0 bar P = 1.0 bar

Fig. 7 Shape of one actuated FFA at two pressures (0.5 and 1.0 bar) versus Fig. 9 Shape of two actuated FFAs at two pressures (0.5 and 1.0 bar) versus
finite element simulation. finite element simulation.

572
responses. The minimum difference of total displacement is [9] M. Cianchetti, et al, “STIFF-FLOP surgical manipulator: mechanical
design and experimental characterization of the single module,” in
3 mm (4%), while the maximum difference is 13 mm (19%).
Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International
Fig. 9 demonstrates close resemblance between shapes of the Conference pp. 3576–3581, 2013.
soft manipulator for both experimental and finite element [10] J. Fraś, et al, “New STIFF-FLOP module construction idea for improved
results at two different pressures. actuation and sensing,” in Robotics and Automation (ICRA), 2015
IEEE International Conference, pp. 2901–2906, 2015.
[11] R. J. Webster III and B. A. Jones, “Design and kinematic modeling of
VI. CONCLUSION constant curvature continuum robots: A review,” The International
Journal of Robotics Research, vol. 29, no. 13, pp.1661–1683, 2010.
This paper introduces a transient FEA based tool for [12] Y. Elsayed, C. Lekakou, T. Geng, and C. M. Saaj, “Design optimisation
designing soft surgical manipulator. The soft manipulator of soft silicone pneumatic actuators using finite element analysis,” in
design, fabrication process, and finite element modeling are Advanced Intelligent Mechatronics (AIM), 2014 IEEE/ASME
presented. An experimental setup has been built to test the International Conference, pp. 44–49, 2014.
[13] Y. Elsayed, et al, “Finite element analysis and design optimization of a
proposed methodology. Pressure sensors are used to measure pneumatically actuating silicone module for robotic surgery
the pressure inside flexible fluidic actuators and transfer them applications,” Soft Robotics, vol. 1, no. 4, pp.255–262, 2014.
to FEA as input loads. Vision tracking system is employed to [14] I. S. Godage, W. Raul, I. D. Walker, and R. J. Webster III, “Accurate
record the manipulator trajectory. Comparing finite element and efficient dynamics for variable-length continuum arms: A center of
gravity approach,” Soft Robotics vol. 2, no. 3, pp.96-106, 2015.
results with experimental results shows that the estimated
[15] S. M. Sadati, et al, “A geometry deformation model for braided
difference between them in the total displacement ranges from continuum manipulators,” Frontiers in Robotics and AI 4, 2017.
4%, at nearly constant pressure, to 19% at high transient [16] https://www.smooth-on.com/products/ecoflex-00-30/
pressure conditions. The agreement between transient FEA [17] https://www.smooth-on.com/products/dragon-skin-30/
response and experimental system response indicates that the [18] S. Garrido-Jurado, R. Muñoz-Salinas, F .J. Madrid-Cuevas, and M. J.
Marín-Jiménez, “Automatic generation and detection of highly reliable
suggested methodology can predict well the system response. fiducial markers under occlusion,” Pattern Recognition, vol. 47, no. 6,
Hence, it can be used for checking the design of new soft pp. 2280–2292, 2014.
robots without the need to build them in advance. Future work
will focus on using the proposed methodology in designing
and controlling of new soft robots to show its effectiveness in
reducing the development period.

ACKNOWLEDGMENT
The first author is supported by a scholarship from the
Mission Department, Ministry of Higher Education of the
Government of Egypt which is appreciatively acknowledged.
The authors would like to thank the Egypt-Japan University of
Science and Technology (E-JUST) for providing the required
materials and equipment to carry out this research work.

REFERENCES
[1] A. Verl, A. Albu-Schäffer, O. Brock, and A. Raatz, Soft Robotics.
Springer, Heidelberg; 2015.
[2] D. Trivedi, C. D. Rahn, W. M. Kier, and I. D. Walker, “Soft robotics:
Biological inspiration, state of the art, and future research,” Applied
bionics and biomechanics, vol. 5, no. 3, pp. 99–117, 2008.
[3] M. Cianchetti, et al, “Soft robotics technologies to address shortcomings
in today's minimally invasive surgery: the STIFF-FLOP approach,” Soft
robotics, vol. 1, no. 2, pp.122–31, 2014.
[4] H. M. Le, T. N. Do, and S. J. Phee, “A survey on actuators-driven
surgical robots,” Sensors and Actuators A: Physical, vol. 247, no. 15,
pp. 323–354, 2016.
[5] V. Vitiello, S. L. Lee, T. P. Cundy, and G. Z. Yang, “Emerging robotic
platforms for minimally invasive surgery,” IEEE reviews in biomedical
engineering, vol. 6, pp.111–126, 2013.
[6] J. Burgner-Kahrs, D. C. Rucker, and H. Choset, “Continuum robots for
medical applications: A survey,” IEEE Transactions on Robotics, vol.
31, no. 6, pp. 1261–1280, 2015.
[7] T. Ranzani, G. Gerboni, M. Cianchetti, and A. Menciassi, “A
bioinspired soft manipulator for minimally invasive surgery,”
Bioinspiration & biomimetics, vol. 10, no. 3, 2015.
[8] A. De Greef, P. Lambert, and A. Delchambre, “Towards flexible
medical instruments: Review of flexible fluidic actuators,” Precision
engineering, vol. 33, no. 4, pp.311–21, 2009.

573

Você também pode gostar