Miniature Endoscopic Capsule Robot using

Biomimetic Micro-Patterned Adhesives

Mustafa Emre Karagozler1 , Eugene Cheung2 , Jiwoon Kwon3 , and Metin Sitti1 ,2 ,4
of Electrical and Computer Eng., 2 Dept. of Mechanical Eng., 4 Robotics Institute,
1 Dept.

Carnegie Mellon University, Pittsburgh, PA 15213, USA

3 Intelligent Microsystem Center, KIST, Seoul, Korea

{mkaragoz,eccheung,msitti}@andrew.cmu.edu, [email protected]

Abstract— This paper presents a stopping and a locomotion of the microcapsule, enabling advanced applications such
mechanism to be used with an endoscopic microcapsule robot. as biopsy, localized drug delivery and surgery.
In the diagnosis of gastrointestinal diseases, microcapsules Previous studies [4]–[6] have been made on the develop-
have been developed recently as alternatives to conventional
endoscopy. However, they have less accuracy and functionality ment of such stopping mechanisms and mechanical inter-
in diagnosis as they lack the ability to control their position. locking based attachment mechanism such as micro-hooks.
We propose mechanisms to be used with such microcapsules In this paper, we propose a novel stopping mechanism with
that would enable them to anchor and crawl in any position a biomimetic non-invasive attachment and friction method
inside the small intestines. The stopping mechanism, actuated that would enable an endoscopic microcapsule to anchor
by coil type shape memory alloys, makes use of dry and
wet elastomer (PDMS) micro-patterned adhesives inspired itself in one position within the small intestines. Further-
by beetles to attach to the intestinal tract. The locomotion more, we discuss how the usage of such a mechanism
mechanism, inspired by the locomotion principles of inch- could be expanded in order to be used as a crawling based
worms, is a modular expansion of the stopping mechanism. locomotion mechanism.
Both the stopping and the crawling locomotion mechanisms
have been built and successfully tested inside a flexible vinyl II. P ROBLEM D EFINITION
tube. Results showed stopping with high repeatability and 0.5
mm/sec locomotion speed. The stopping mechanism was also The main challenges in the development of an attach-
integrated to a tethered camera for testing. ment and locomotion mechanism for a microcapsule robot
Index Terms— Biomimetic, Locomotion, Endoscopic, Mi- are:
crocapsule. • Since the size of the capsule is very important (very
large pills are uncomfortable), the mechanism must
I. I NTRODUCTION be sufficiently small as to not drastically increase the
Traditional wired endoscopes are primarily used in the size of the system.
• In situ, the microcapsule is subject to the peristaltic
investigation of the gastrointestinal tract for diagnosis of
diseases. Research is being conducted on the use of semi- motion of the digestive tract at all times, so the
autonomous structures in order to avoid the inconveniences attachment mechanism should be able to conform to
introduced by traditional endoscopes [1]. In 2001, with the intestinal environment.
• The attachment mechanism must be supplied with the
the invention of microcapsules, a much more convenient
alternative was introduced to the market [2]. Microcapsules same power that is already onboard the microcapsule.
offer virtually non-invasive diagnoses and require less com- As these power systems are the limiting factor in the
plicated operation procedures, resulting in overall higher endurance of most capsule systems, the attachment
convenience. In addition, the use of microcapsules enables mechanism must be low power.
• When operating within the human body, safety is a
screening of the small intestines, which is otherwise im-
possible with current traditional endoscopy. major concern; any attachment mechanism must be
Despite their advantages over traditional endoscopes, chosen with safety and biocompatibility as a priority.
microcapsules have a low accuracy and functionality in With these constraints, many traditional attachment
diagnosis mainly due to the lack of control over their mechanisms can be discarded. Mechanical interlocking
position, orientation and speed. The peristaltic motion mechanisms must not be used to avoid damaging the
provided by the contraction of intestinal muscles pushes ev- digestive tract. Suction based systems require too much
erything inside; anything that does not deploy a mechanism power to be useful. Therefore, bio-inspired repeatable and
of attachment will be forcibly moved. The microcapsule non-invasive attachment mechanisms are proposed as a
developed by RF Norika [3] has an external magnetically solution in this work.
actuated system that controls the orientation of the capsule,
but it lacks the ability to stop inside the intestines and A. Application Scenario
further investigate a region of interest. The need for such Since the proposed capsule robot would be used only for
a mechanism is great, as it will improve the accuracy of a limited duration due to the power consumption issues and
the diagnosis and provide a level of control to the motion it is not economical to consider a medical doctor waiting all
the time during the all endoscopy process which can take effect is also possible [12]. Due to its simplicity of fab-
around 5-20 hours, the following scenario is proposed for rication and integration, these elastomer micro-pillars are
the application of the robotic endoscopic capsule. At first, chosen in this work.
the standard commercial capsule camera is swallowed by Using the photoresist SU-8, optical lithography tech-
the patient with the standard out of hospital process and nique is used to pattern micro-holes inside a spin coated
get a map and approximate position of all intestine regions thick SU-8 layer in the order of 50 − 300 μm. Then,
with potential problems. Here, peristaltis of intestines move polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning)
the capsule camera passively. Next, the robotic capsule is is molded inside these micro-holes and peeled off to
swallowed by the patient. When there is no intervention, it fabricate PDMS micro-pillars with different density, aspect
would move with the peristaltis of intestines like the passive ratio and diameter. From adhesion experiments, we found
capsule camera. Only at the predetermined problematic that 140 μm pillars with less than 1 : 2 aspect ratio with
locations of intestine, a doctor teleoperates the robotic a spacing of 105 μm between pillars would be the most
capsule to stop it at the exact location needed to conduct optimal pillar for adhering to small intestines and also
a detailed imaging, biopsy, or localized drug delivery. wide range of other surfaces such as glass, aluminum, and
Locomotion capability of the robotic capsule here will be polystyrene. Molded micro-pillars are shown in Figure 1.
used to adjust the forward or backward position of the Besides bare dry micro-pillars, we also used silicone oil
capsule for short distances only; thus, it will be used for coated pillars as a wet micro-patterned adhesive similar
short durations and distances. These operations can be also to beetles and crickets. For oil coating, silicone oil with
realized remotely by the doctor using internet. 10000 cSt is spun on a flat glass surface and PDMS micro-
pillars are pressed on to this thin oil layer and oil layer is
III. S TOPPING M ECHANISM transferred to the pillar tips.

Several research groups have proposed different minia- B. Design Principle

ture mechanisms that make use of setae–like micro legs The design of the stopping mechanism was geared
employed by earthworms [4], [6]. The design idea of the towards the use of synthetic biomimetic micro-patterned
stopping mechanism presented in this paper relies on a adhesives. This requires a mechanism that will push the
biologically inspired wet or dry polymer micro-patterned adhesive material onto the intestinal walls, the simplest of
adhesive material that would be pushed against the intesti- which is a set of adhesive-tipped legs that can open to
nal walls, creating an adhesion [8]. stabilize the capsule.
Many factors must be considered in the selection of
A. Bio-Inspired Adhesives the actuator for this purpose. The main concern is that
it must be biocompatible. Also, high output force and
Nature has evolved to develop sub-optimal repeatable low power consumption are required. Several actuators,
attachment mechanisms for robust and agile locomotion including piezoelectric materials, polymer actuators and
on wide range of smooth and rough surfaces. These shape memory alloys were considered in the design stage.
mechanisms mainly use dry or liquid coated microme- While the piezoelectric materials have high force output
ter or nanometer scale foot-hairs using capillary and/or and low power consumption, they typically need very high
van der Waals type of intermolecular forces, claws using driving voltages. This, in addition to the limited strain
mechanical interlocking, and suction pads using vacuum capabilities of the actuator, make it an inappropriate choice
suction. For a robust and non-invasive attachment mech- for our purpose. Polymer actuators have much better strain
anism inside intestine where the intestine wall is very output, and they are biocompatible. However, they are slow,
rough (covered with micro-vilia), compliant, and slippery they require high power and they are incapable of high
(mucus secretion), using micro/nano-hairs would be the output force.
bestmost robust and non-invasive solution. Beetles, flies, Shape memory alloy (SMA) wires are selected here
crickets, and cockroaches use few micron diameter foot- like other several research groups that proposed these
hairs covered with a carbon based hydrophobic oil [13],
[14] while geckos and spiders use hierarchical dry fibers
with diameters starting from micron scale and branching
down to 100s of nanometer scale hairs with saucer type
endings (spatulae) [9]–[11]. In all of these fibrillar adhesive
mechanisms, major principle is to increase the contact area
with a given rough surface by having as small as possible
diameter compliant fibers with high density to enhance the
adhesion and friction. These high aspect ratio and high
density hairs necessitate relatively rigid hair biomaterials
Fig. 1. 140 μm diameter, 1 : 1 aspect ratio, and 105 μm spaced PDMS
such as beta-keratin in order to have minimal hair matting. micro-pillars: top-view optical microscope image of the dry pillars (left
However, using soft elastomer materials with very low image), and top-view image of the silicone oil (10000 cSt) coated pillars
aspect ratio micro-pillars, a similar adhesion enhancement (right image).
actuators for microcapsule applications [4], [6] since they
are compact, biocompatible, and have 3 − 5% strain with
large force output. A specific shape memory alloy wire, the
coil type SMA wire [7], is selected in this work for micro-
actuation due to their high output force and large strain up
to 50%, in comparison to 3 − 5% of standard SMA wires.
The actuation is through heating; a voltage applied across Fig. 3. Diagram showing the dimensions used in the mathematical model.
the wire causes current flow, and the power is dissipated The mechanism is drawn in the unactuated position.
as heat. This also means that they have a very high power
consumption and low efficiency that limits their untethered
usage. Therefore, as mentioned in the application scenario,
these actuators would be used for only short durations when
stopping and locomotion are required.
Figure 2 shows the design concept of the stopping
mechanism. The mechanism is built on a hollow cylindrical
casing (D). Three legs (B) are each attached on a cylindrical (a) (b)
pulley (F) that is free to rotate. A rubber spring (E) is Fig. 4. A schematic of the two contact modes and the forces
attached to the pulley on one end, and to the casing on that act on the leg: a) In the flat mode, the footpad completely
the other end. An adhesive pad (A) is attached to the contacts the substrate. b) In the angled mode, only the distal end
of the footpad is in contact with the substrate.
footpad, which is attached to the leg with a polymer hinge
(not shown). The SMA coiled wire (C) connects the upper
surface of the leg to the casing.
the pad starts to touch the intestinal wall, the polymer
The initial state of the leg is closed (SMA coiled wire hinge bends, applying both shear and normal preload.
not actuated). When the SMA wire is heated by passing Consequently, when the leg closes, this structure creates
current through it, it pulls the leg, creating a torque. This a peeling motion, greatly reducing the force necessary to
torque causes the pulley to turn, thereby rotating the leg. detach. This compliant structure is formed using a stiff
The rubber spring is stretched as the pulley turns, creating a polymer film which connects the two separate parts of the
torque in the other direction. The leg stops opening when leg.
it reaches an equilibrium point where the torque created
by the SMA wire equals the torque created by the rubber C. Modeling
spring. Once the SMA wire is shut off, the leg is pulled A mathematical model of a single leg mechanism was
back by the rubber spring since the SMA wire no longer created to predict the forces between the foot and substrate
creates a strong resisting torque. when current is applied to the SMA actuator. Figure 3
The foot carries an adhesive pad that will stick and shows the relevant dimensions of the mechanism used in
generate high adhesion forces when pushed on the intesti- this model. The leg and foot were treated as rigid beams.
nal walls. As the adhesive pad reaches the walls and is The force required to bend the ankle joint (polymer hinge)
preloaded, the leg does not open further, but it applies a was calculated using large beam deflection theory. The
preload as the SMA is continuously actuated. Three legs PDMS spring was treated as a linear spring whose spring
are placed symmetrically on the casing, so that a strong grip constant could be calculated through simple axial beam
from three points is formed. One critical mechanism is the loading equations. These assumptions were experimentally
polymer hinge through which the adhesive pad is connected validated with simple force measurements using a load cell.
to the legs. The purpose of the compliance created by The current-force and strain-force relations for the SMA
the polymer hinge is to aid the adhesive pad in sticking wire were also found experimentally and incorporated
to and detaching from the walls of the intestine. When into the model. For the the 100 μm coil-type SMA, the
empirical output force FSM A in mN was found to be
reasonably approximated by
FSM A = 0.0341c2 − 0.0966c + (772.72 − 17.053c) ε (1)
where c is the applied current in mA and ε is the strain.
This equation is only valid above a threshold current of 60
mA and below a burnout current of 180 mA.
If enough current is applied to the SMA, the leg will
rotate open and push the footpad against the substrate.
There are two modes of contact as displayed in Figure 4:
Fig. 2. CAD drawing of the conceptual design (right image), and labelled
the “flat mode” where the footpad is pressed flat against the
details: A: Adhesive Pad, B: Leg, C: SMA coiled wire, D: Casing, E: substrate and the “angled mode” where only the distal end
PDMS Spring, F: Pulley of the footpad is in contact with the substrate. The shear
force Fshear is the friction force on the footpad, calculated D. Fabrication
with Fshear = μFpreload , assuming high preloads. By bal- Two prototype stopping mechanisms were built and
ancing moments about the pulley, the leg angle θ, preload successfully tested. The casings, cylindrical pulleys and
force, shear force, and contact mode can be calculated legs were machined from Delrin® which is a hard but easy
for any given input current to the SMA wire. Assuming to machine polystyrene. The casing is a hollow cylinder
a uniform pressure on the footpad in the flat mode, the which is 9 mm long and 8 mm in diameter, having 0.8 mm
preload force for the flat and angled mode were found to thick walls. The actuators are 100 μm diameter coil type
be: SMA manufactured by Toki Corporation. The adhesive pad
(r + t ) FSM A − rFspring is a 3 mm by 1 mm PDMS pad.
Fpreload,f lat = (2)
Df lat After having all three legs and the casing machined,
Df lat = (L + μr) cos θ + (μL − r) sin θ the parts were assembled using Loctite 495 Super Glue.
+ (L /2 + μtf ) The electrical connections for the SMA wires and the
external connectors were done using clamping. The SMAs
were connected in series for simplicity, requiring only two
(r + t ) FSM A − rFspring
Fpreload,angled = (3) connectors per three actuated legs.
Dangled
Dangled = (L + μr) cos θ + (μL − r) sin θ E. Experiments
+ (L + μtf ) cos (θ − α) + (μL − tf ) sin (θ − α) A series of tests were conducted to gain a measure of
the viability of the fabricated stopping prototype. Measure-
where r, L , Lf , t and tf are mechanism dimensions
ments were made of the preload and shear forces produced
defined in Figure 3, θ is the opening angle of the leg (see
by a single leg mechanism. In addition, in vitro tests of the
Figure 4), and α is the bend angle of the ankle joint.
complete mechanism were performed inside a vinyl tube.
The factor that determines which mode of contact the
1) Single Leg Test: A single leg mechanism was actu-
mechanism will achieve is the ankle joint. In order to
ated with a current of 150 mA while the distance between
reach and maintain the flat mode, sufficient force must
the unactuated footpad position and the substrate was
be applied to the ankle to maintain the bend. Given a
varied. The applied preload force was measured using a
contact mode assumption, the resulting preload force can
load cell (GSO-10, Transducer Techniques).
be calculated. This force can then be used to determine the
The initial preload test was done with a rigid glass sub-
force on the ankle joint, which will then validate or refute
strate. To more accurately represent the working environ-
the assumption.
ment of the capsule, two deformable substrates were also
The previous equations are easily adapted to include the
tested. These substrates were in the form of a membrane;
effects of a deformable membrane in place of a rigid wall.
mounted only on the edges, the leg could push well past
The preload force causes a deformation x in the wall given
the undisturbed substrate surface as is expected inside the
by a simple spring equation x = Fpreload /kmem , where
human gastrointestinal tract. The first membrane was 0.75
kmem is a measure of the flexural rigidity of the membrane.
mm thick PDMS rubber, and the second was 0.1 mm thick
For a built-in circular membrane of radius a and thickness t
nitrile rubber. The results of these characterization tests are
pressed at the center, kmem = (16π/a2 )(Et3 /(12(1−ν 2 )))
shown in Figure 6. When compared to the model prediction
where E and ν are the material properties of the membrane.
(see Figure 5) the characterization results show the same
This effectively increases the distance d by the deformation
three distinct contact modes. Indeed, visual observations
amount x. Since increasing d will necessarily change the
confirmed the type of contact at each point corresponded to
preload force applied, a simple iterative method is used to
determine the equilibrium position.
A graph of the calculated preload forces for various
distances d from both a rigid substrate and deformable
PDMS membrane can be seen in Figure 5. There are two
clear transition points on the graph that indicate a switch
in contact modes. On the left side of the graph, where
the preload force decreases as the distance increases, the
mechanism maintains a flat mode of contact. At d = 6.7
mm for the rigid substrate and 5.8 mm for the membrane,
the contact changes to angled mode; after this transition,
the preload force stays relatively constant (even increasing
slightly). Finally, at d = 9.2 mm for both configurations,
the foot no longer reaches the substrate, and the forces drop
to zero. The PDMS membrane appears to result in a simple
Fig. 5. Simulated preload forces on a rigid substrate and thin PDMS
left-ward shift of the rigid substrate graph, in agreement membrane using the developed mathematical model. For these calcula-
with the earlier assertion that the membrane deformation tions, L = 9.5 mm, t = 0.9 mm, Lf = 4.1 mm, tf = 1.7 mm,
causes an increase in d. r = 1.35 mm, c = 150 mA, and μ = 0.75.
theory. In addition, Figure 5 was generated with parameters
that matched the fabricated prototype so it can be seen that
the rigid glass substrate results compare favorably to the
model. The contact modes transition at approximately the
same distances and the initial slopes of the curves match.
The characterization results appear to be simply shifted to a
lower force from the model predictions, a phenomenon that
might be explained by fabrication imperfections and fric- Fig. 7. Single robot leg friction measurement setup using a guided almost
tion. The PDMS membrane results match up similarly, with frictionless sliding of robot leg body attached to a load cell through a
the exception of the angled contact mode section between tendon. Fresh pig intestine is stretched between two manual stages, and
the SMA actuated leg foot-pad applies a preload to the adhesive and
d = 6 and 9 mm. The model suggests that this should sliding of the foot-pad gives the static and dynamic friction data.
have the same values as the rigid substrate. The smaller
values seen in the experiment can be attributed to oversim-
plifications in the calculation of the membrane deflection. The stopping mechanism was then tested in vitro. By the
These results show that the developed mathematical model use of fishing line attached to the casing, the mechanism
provides a reasonable estimate of the performance of the was lowered into a 19 mm inner diameter vinyl tube that
leg mechanism. In the future, this may be used to further was set up vertically on the test bench. As current was
refine the design via the optimization of various parameters. applied to the actuators, the legs opened, stopping the
In addition to measuring the preload force on a given capsule within the tube (see Figure 10). When the current
compliant or rigid surface, using this preload, a custom was removed, the legs retracted to their original position,
setup illustrated in Figure 7 is used to measure the friction peeling the adhesive pads off of the wall per the design
force generated by the foot-pad of the single leg consisting and detaching the capsule from the walls of the tube. To
of flat, dry, and silicone oil coated PDMS micro-pillars. ensure that it was not the tether that was supporting the
In this setup, friction of flat, dry and silicone oil coated capsule but the adhesive pads themselves, the tether was
140 μm diameter micro-pillars are tested on a very fresh wiggled, released and even pulled; the prototype always
pig intestine surface with preserved mucus layer. Friction held its position within the tube. In addition, the capsule
results are shown in Figure 8. As a result, dry micro-pillars prototype was tested within a larger vinyl tube with an
enhance static friction around 4 times with compared to inner diameter of 25 mm. The capsule performance was
the flat PDMS material. Our oil coated pillars are expected comparable to that in the smaller tube. These tests showed
to show enhancement around 7 times with compared to that this prototype is capable of basic stopping performance
the flat PDMS material. Thus, oil coated and dry micro- in accordance with the design parameters. It works robustly
patterns can apply sufficient friction and attachment force and promises to cover the basic requirement; it appears to
to resist against the weight of the capsule (in the vertical provide enough preload on the adhesive pads.
configuration especially) and intestine peristaltis pushing During tests, it has been observed that the intestinal
lateral force. Therefore, this bioinspired adhesive is very tissue does not allow sharp-edged structures to pass through
promising to stop the microcapsule in a desired location easily, as they pierce in the tissue. Another issue is that the
in any location in small intestines and crawl the capsule intestinal tissue is in collapsed form; which collapses onto
forward or backward directions.
2) Whole Mechanism Test: The successful tests of the
single leg led to whole mechanism tests. Figure 9 shows the
opening and closing of the 3 legs assembled on the casing.


'LASS
 0$-3 MEMBRANE
.ITRILE MEMBRANE






&ORCE M.










          
$ISTANCE MM
Fig. 8. Experimental friction force data of flat PDMS and dry PDMS
micro-pillars with 140 μm diameter, 125 μm long, and 105 μm spacing
Fig. 6. Single leg experimental preload characterization results. The on an almost in vivo pig small intestine wall. Adhesive area is 5 ×
prototype single leg mechanism was actuated with 150 mA and 2.30 V 1.5 mm2 . Adhesive preload is around 6 mN and constant motion speed
for all data points. is 0.2 mm/s.
Fig. 9. The stopping mechanism: Legs closed initially (left image), and
legs opened (right image).

Fig. 11. The CAD drawing of the proposed inchworm locomotion

mechanism (exploded view, left image), and the inchworm locomotion
principle (right image). A: Compression Spring, B: Hollow Cylinder, C:
Coil Type SMA.

any time, three of the legs are always open, anchoring the
capsule to the intestine walls, while the other module either
pushed or pulled to achieve the desired movement. The
Fig. 10. Snapshots of the stopping mechanism inside a vinyl tube: Free to
move (legs closed, leftmost image), anchored (legs open, middle image), mechanism works as follows: first, the front legs are closed,
and detached, free to move (legs closed, rightmost image). and the piston is extended; then the front legs are opened
again and this time the rear legs are closed; after that, the
rear legs are pulled by shrinking the piston; once pulled,
the adhesive pads, causing stiction even when not desired. they are again opened, returning to the initial state.
So, it is important to keep the adhesive pads away from the The first locomotion prototype is built. Figure 12 shows
intestinal walls, when they are not opened. Considering the stills of a movie in which the prototype is operated
above mentioned issue, the design is modified to have as desired. The capsule prototype was also tested in a 19
round edges and slits in which the legs will be placed when mm diameter vinyl tube. The mechanism was lowered into
they are not opened. the tube and the lower legs were opened. The control of
the three parts (lower legs, upper legs and the piston) was
IV. L OCOMOTION M ECHANISM
performed manually by switches. Figure 13 shows four
The stopping mechanism is designed to be hollow, states of the crawling locomotion taking a step. Tests have
allowing the space inside to be used for a camera and shown that the capsule could achieve robust motion inside
other modules to extend the capabilities of the capsule. the tube with an average velocity of 0.5 mm/sec. These
The stopping capsule is an advantageous mechanism for experiments showed that sharp edges are not desired and
applications like monitoring the gastrointestinal tract; a legs and adhesive foot-pads need to be hidden inside before
locomotion mechanism would be even more advantageous, legs are opened. Using these results, an improved design
as it would allow forward and backward position control. is proposed in Figure 14.
The current design enables us to extend the usage of the
stopping mechanism and use it as a locomotion mechanism.
Two stopping modules can be integrated to perform loco-
motion using an inchworm locomotion principle. The CAD
drawing of the conceptual design is shown in Figure 11.
The two stopping mechanisms are connected with a com-
pression spring (A) and a hollow cylinder(B), forming a
piston. There is also a coil type SMA wire (C) connected
to the two casings on the ends. The SMA wire and the
Fig. 12. Photo of the locomotion mechanism: Initial state (left image)
spring work in an antagonist fashion. When the SMA wire and legs open and body shrunk (right image).
is not actuated, the spring pushes the casings apart so the
capsule is in its expanded state. When the SMA wire is
actuated, it works against the compression spring, pulling V. C AMERA I NTEGRATED C APSULE
the casings together. To further test the stopping mechanism, it was decided
Locomotion is performed by sequentially opening and to integrate the stopping mechanism to a tethered pill size
closing the legs and actuating the piston. Figure 11 gives analog camera that was available. A new stopping module
a graphical representation of the inchworm movement. At that has the proper dimensions has been build. The figure
 (a) (b) Fig. 15. Photo of the tethered camera integrated to the stopping
mechanism.

that are currently being used. It allowed a miniature camera

to be integrated successfully. The design continues to be
optimized using the developed model in conjunction with
in vitro tests. Possible further works include the research of
actuator power minimization, additional modules, improved
(c) (d) fabrication techniques, and an improved testing environ-
ment.
Fig. 13. Snapshots of the capsule crawling: a) initially only lower legs ACKNOWLEDGMENT
are open, the capsule is anchored, b) body is shrunk, c) upper legs are
opened, d) lower legs are closed and body extended. The authors would like to thank to the Intelligent Mi-
crosystem Center, Korea for funding this research through
the 21st Century Frontier Program.
R EFERENCES
[1] A. Menciassi, Jong H. Park, S. Lee, S. Gorini, P. Dario, and Jong-Oh
Park, ”Robotic Solutions and Mechanisms for a Semi - Autonomous
Endoscope”,Proc. of the IEEE/RSJ Int. Conf. on Intelligent Robots and
Systems, Lausanne, Switzerland, 2002.
[2] http://www.givenimaging.com
[3] http://www.rfnorika.com/
[4] Young Pyo Lee, Byungkyu Kim, Moon Gu Lee, and Jong-Oh Park,
”Locomotive Mechanism Design and Fabrication of Biomimetic Micro
Robot Using Shape Memory Alloy”, Proc. of the IEEE Int. Conf. on
Fig. 14. Improved locomotion mechanism prototype drawing with no Robotics and Automation, vol.5, pp. 5007–5012, New Orleans, LA,
sharp corners and hidden legs and foot-pads. USA, 2004.
[5] Byungkyu Kim, Sunghak Lee, Jong Heong Park, and Jong-Oh Park,
”Design and Fabrication of a Locomotive Mechanism for Capsule-
Type Endoscopes Using Shape Memory Alloys (SMAs), Proc. of the
15 shows the picture of the built stopping mechanism IEEE Int. Conf. on Robotics and Automation”, Barcelona, Spain, April
integrated to the pill camera. 2005.
The camera is placed in the hollow cylindrical casing. [6] A. Menciassi, S. Gorini, G. Pernorio, Liu Weiting, F. Valvo, and P.
Dario, ”Design, Fabrication and Performances of a Biomimetic Robotic
The objective of the camera is facing out. The tethers of the Earthworm”, Proc. of the IEEE Int. Conf. on Robotics and Automation,
camera come out of the cylindrical casing through a hole Barcelona, Spain, April 2005.
that is drilled on the back. The camera - integrated stopping [7] BioMetal, Toki Corporation, Tokyo, Japan. http://www.toki.co.jp/
[8] Eugene Cheung, Mustafa Emre Karagozler, Sukho Park, Byungkyu
mechanism is 33 mm long and 15 mm in diameter. The Kim, and Metin Sitti, ”A new Endoscopic Microcapsule Robot using
mechanism is identical to the previously explained stopping Beetle Inspired Microfibrillar Adhesives”, Proc. of the IEEE/ASME Int.
mechanism. The camera -integrated stopping mechanism Conf. on Advanced Intelligent Mechatronics, Monterey , CA, 2005.
[9] A. K. Geim, S. V. Dubonos, I. V. Grigorieva, K. S. Novoselov, A. A.
has been tested in a 25 mm diameter vinyl tube. The tests Zhukov, and S. Yu. Shapoval, ”Microfabricated adhesive mimicking
shown that stopping could be achieved successfully with gecko foot-hair”, Nature Materials, vol. 2, pp. 461-463, July 2003.
the integrated camera. [10] M. Sitti and R.S. Fearing, ”Synthetic Gecko Foot-Hair Micro/Nano-
Structures for Future Wall-Climbing Robots”, Proc. of the 2003 IEEE
Int. Conf. on Robotics and Automation, Taipei, Taiwan, 2003.
VI. C ONCLUSION [11] M. Sitti and R.S. Fearing, ”Synthetic gecko foot-hair micro/nano-
structures as dry adhesives”, Journal of Adhesion Science and Tech-
This paper demonstrates progress toward the devel- nology, vol. 17–8, pp. 1055–1073, 2003.
opment of a complete stopping and locomotion mecha- [12] A.J. Crosby, M. Hageman, and A. Duncan, ”Controlling polymer
adhesion with ”pancakes””, Langmuir, vol. 21(25), pp. 11738–43,
nism for endoscopic microcapsules. Characterization and 2005.
optimization of important design parameters through the [13] T. Eisner and D. Aneshansley, ”Defense by foot adhesion in a beetle
modelling of the mechanisms and measurements have been (Hemisphaerota cyanea)”, PNAS, 97(12), pp. 6568–6573, 2000.
[14] W. Federle, M. Riehle, A.S.G. Curtis, and R.J. Full, ”An integrative
achieved. The designs for stopping and locomotion mech- study of insect adhesion: mechanics and wet adhesion of pretarsal pads
anisms for an endoscopic microcapsule robot perform ro- in ants”, Integrative and Comparative Biology, vol. 42–6, pp. 1100–
bustly. The modularity of the stopping mechanism enables 1106, 2002.
it to be used as an expansion module for microcapsules