This document summarizes the design optimization of a climbing robot that uses micro-fiber arrays for adhesion. It describes how the thickness of the elastomer adhesive and the wheel radius were optimized through testing. Models are presented for peeling forces from elastomer stretching and bending. Equations are developed to calculate peeling forces based on parameters like elastic modulus, thickness, wheel radius, and mass. Charts show the relationships between effective peeling force, thickness, and wheel radius for different ratios of wheel to natural peeling radius. The optimal values that maximize peeling force within a 90% range are presented.

1. Design, Optimization And Manufacturing Of Climbing Robots Utilizing
Micro-Fiber Arrays
Ahmet Çalışa
, Dr. Özgür Ünverb
Master of Science, Department of Mechanical Engineering, Hacettepe University, Turkey
February 2013
____________________________________________________________________________________________________________________________
Abstract
Few climbing robots have been built using flat dry elastomer adhesives until now, and none of them have looked deeply into the wheel
size and dry elastomer thickness. In this project, the effect of the adhesive thickness and the wheel radius is comprehensively observed.
Furthermore, the utilization of compliant mechanisms, as well as active tail control to achieve higher adhesion forces during climbing,
surface transition and overcoming obstacles are illustrated.
Key Words: Climbing robot, elastomer thickness optimization, wheel optimization.
____________________________________________________________________________________________________________________________
Introduction:1.
Robots have been used for many years in industrial
applications and daily life and their prevalence increases
every day. The main reasons people use robots are:
increasing public health and safety and finishing any given
duty without any flaw. Climbing and executing a given
duty can be done effortlessly and efficiently by climbing
robots.
In recent years, there have been many studies and
improvements in climbing robots. Thus, climbing robots
are used actively in many missions, such as in nuclear
facilities, on planes, buildings, and pipelines for inspection,
surveillance, maintenance and repair.
In this work, climbing robot design is optimized
dimensionally according to the ratio between adhesion and
robot size. Climbing tests are examined if performance
levels are met. Further studies are done to increase
climbing safety and performance.
So far, dimensional optimization has not been taken into
account while designing climbing robot.
Elastomer Film Peeling Models2.
This section shows two extreme cases of peeling of
elastic films analytically. These cases are; peeling only by
stretching and bending. In these models quasi-static and
plane strain conditions are considered. An external work
must be provided to peel the elastomer film from a surface.
Then, this work is utilized both internally as an elastic
deformation in the film and as an adhesive energy to keep
the adhering bodies in contact with the contact area.
Internal energy is also divided into two as axial and
bending elastic deformation energy. The total energy
balance can be shown as;
dE!"# = dE!"# + dE!"#$ (1)
Fig. 1. Peeling only by bending (a), peeling by pure stretching (b)
M
EI
dθ
R0
M
dL
a)
EA
F
dL
F
α
(1-cos α)dL
εdL
b)

2. External Work2.1.
As seen in Fig.1 an elastic thin film can be peeled off the
surface by bending (M) and pulling (F). These external
actions can be applied individually or simultaneously as in
the equations below;
𝑑𝐸!"# = 𝐹 𝑑𝑢 + 𝑀𝑑𝜃 (2)
It is assumed that both force and moment are applied in 2D
plane.
Internal Energy2.2.
Internal energy is related to the elastic deformations in
the elastomer film due to pulling and bending which can be
formulized as;
𝑑𝐸!"# =
!
!
𝐸𝐴𝜖!
𝑑𝐿 +
!
!
𝐸𝐼𝜅!
𝑑𝐿 (3)
Where 𝐸 is the modulus of elasticity, 𝐴 is the cross
section area of the elastic film (𝑤𝑡), 𝜖 is the strain, 𝐼 is the
second moment of area (𝑤𝑡!
/12), 𝜅 is the curvature, and
𝑑𝐿 is the unit peeling distance. In this equation 𝐸𝐴 and 𝐸𝐼
are related to the axial and bending stiffness, respectively.
Axial stress and bending moment can be calculated as 𝐸𝐴𝜀
and 𝐸𝐼𝜅 respectively.
Contact Energy2.3.
The contact energy is related to the adhesion between
elastomer and adhered surface. This term also indicates the
work of adhesion (𝑤!"!). Work of adhesion is the energy
needed to separate the surfaces from each other. The
contact energy can be calculated as;
𝑑𝐸!"#$ = 𝑤!"! 𝑑𝐴!"#$ (4)
where, 𝑑𝐴!"#$ is the unit peeling area.
Peeling Only by Bending2.4.
This is the extreme case of bending. As seen in Fig.1a,
elastomer film detaches from the surface only due to the
bending moment. In this case, external energy (because of
the moment) is used to bend the material and detach it from
the contact surface. This equality can be shown as;
𝑀𝑑𝜃 =
!
!
𝐸𝐼𝜅𝑑𝜃 + 𝑤!"! 𝑤𝑅! 𝑑𝜃 (5)
where, 𝜅 = 𝑅!
!!
, 𝑑𝐿 = 𝑅! 𝑑𝜃, 𝑀 = 𝐸𝐼𝜅 and 𝐼 = 𝑤𝑡!
/12.
By solving equation 5;
𝑀 =
!!!!!!!"!
!
(6)
𝑅! =
!!!
!"!!"!
(7)
If the adhesive material and the surface properties are
constant, then 𝑀 increases linearly with 𝑤 and non-linearly
with elastomer thickness by 𝑡!/!
. Likewise; 𝑅! increases
non-linearly with elastomer thickness by 𝑡!/!
. In addition,
when the elastomer width is kept constant, then 𝑀 and 𝑅!
become linearly dependent.
Peeling Only by Stretching2.5.
This is the extreme case of stretching. As seen in Fig.1b,
elastomer film detaches from the surface only due to the
pulling force. In this case, created external energy is
consumed to stretch the material and detach it from the
contact surface. This equality can be shown as;
𝐹! 1 − 𝑐𝑜𝑠𝜃 + 𝜖 𝑑𝐿 =
!
!
𝐸𝜖!
𝑑𝐿 + 𝑤!"! 𝑤𝑅! 𝑑𝐿 (8)
where, 𝐹! = 𝐸𝐴𝜖. By solving equation 8;
𝐹! = 𝐸𝑤𝑡
!!!"!
!"
+ (1 − 𝑐𝑜𝑠𝜃)! − (1 − 𝑐𝑜𝑠𝜃) (9)
Optimization3.
As seen in Fig.2 free body diagram of elastic thin film
peeling of by pure stretching where Fp is peeling force, FpN
is normal component of peeling force, Fshear
is shear force
of elastic thin film, Fmg
is weight on the wheel due to
gravity, FS
is slack force, Ft
is tail force FN
, Fm
is force due
to motor torque, Tm
is motor torque, Ffric
is friction force, Fx
is force on the wheel’s x-axis.
Fig. 2. FBD of peeling only by bending
Contact Surface
F
fric
F
F
p
F
pN
F
shear
F
S
F
mg
F
t
F
m
F
N
F
N
F
x
T
m

3. Fig. 3. FBD of peeling only by pure bending
In this project, legged robots are not preferred due to
their complex nature, instead wheeled or tracked robots are
considered. For tracked climbing robots, keeping the track
on the wheel is a big challenge, therefore, wheeled type
climbing robot becomes the focus of this work.
Elastomeric thin film is wrapped around the wheel and
fixed to the wheel hub by strings. The main challenges of
this task are the optimization of the elastomer thickness and
the wheel radius.
Elastomer Stretching Force3.1.
Using equation 9, peeling force can be found for each
corresponding peeling angle assuming no moment effect. In
reality, there will only one peeling angle and its
corresponding peeling force due to the balance of stretching
and peeling force. To find out these values, elastomer
should be stretched from initial length to the final length.
These lengths can be calculated as given below;
𝐿! = 2𝜋 𝑅! +
!
!
(10)
𝐿! = 𝑅! +
!
!
2𝜋 − 𝜃 + 2 𝑅! +
!
!
tan
!
!
(11)
Then the stretching force can be calculated as;
𝐹!"# = 𝐸𝑤𝑡
!!!!!
!!
(12)
When the wheel is stationary, elastomer is its relaxed
state and 𝜃 = 0. When the wheel starts to rotate, due to the
surface adhesion, elastomer does not come off the surface
at first and starts to elongate (stretch). Simultaneously, 𝜃
starts to increase. As the wheel rotates, 𝐹!"# and 𝜃 keep
increasing until 𝐹!"# becomes equal to 𝐹! as seen in Fig 2.
At this point, transition period finishes and the elastomer
detaches from the surface and this snapshot (𝜃, 𝐹!"# = 𝐹!) is
preserved throughout the operation assuming all parameters
are constant. 𝐸 = 110 𝑘𝑃𝑎, 𝑊! = 70 𝑁 𝑚 , 𝑤 =
10 𝑚𝑚, 𝑅! = 30 𝑚𝑚 and 𝑡 = 2 𝑚𝑚 are used to plot
Fig.2. According to the plot, peeling force is 0.45 𝑁 and
𝜃 = 118 degree.
Fig. 4. Plot of peeling and stretching forces
Peeling occurs when stretching force equals to the
peeling force. In this case, for every given θ and t there is
only one intersection point which satisfies the equality of
stretching and peeling force. This equation can be
generalized as;
𝐹!"# − 𝐹! = 0 (13)
The peeling force attained from Fig. 2 does not affect
the climbing performance directly. However, the normal
direction of this force is used to push the robot towards the
surface and to preload the elastomer. Therefore, utilizable
component of the peeling force is the force perpendicular to
the surface. This normal peeling force can be calculated as;
𝐹!
!
= 𝐹! 𝑠𝑖𝑛𝜃 (14)
Moreover, in Fig.2, it is assumed that the wheel and the
elastomer are massless. However, especially for inverted
climbing, mass pulls the robot away from the surface
towards the ground, therefore, masses should be taken into
account and should be subtracted from 𝐹!
!
. In this work,
elastomer film and wheel hub is considered as masses. The
available adhesion left from elastomer and wheel hub is
going to be used to carry the robot and its accessories such
as sensors and links. Mass of the elastomer can be
calculated as;
𝑀! = 𝜋𝑤𝜌 𝑅! + 𝑡 !
− 𝑅!
!
(15)
where, 𝜌 is the material density. Mass of the wheel is
calculated roughly utilizing the manufactured wheel masses
which fits nicely to the equation given below for our
application;
𝑀! = 0.001 + 0.2𝑅! (16)
Therefore, the total force which pulls the robot towards
the ground during inverted climbing can be calculated as
Contact Surface
F
fric
F
F
p
F
mg
F
t
F
rm
F
N
F
N
F
x
M
Tm

4. Table 1.
Values of 𝐹!
!
, 𝑡, and 𝑅! for different 𝑅!/𝑅! ratios.
𝑅!/𝑅! Maximum 𝐹!
!
[N] Optimum 𝑡 [mm] %90 𝑡 Range [mm] Optimum 𝑅! [mm] %90 𝑅! Range [mm]
1 0.59 10.0 5.2-15.4 8.2 3.0-15.0
3 0.51 6.5 3.6 - 10.0 12.5 5.3-24.1
5 0.47 5.5 2.9-7.9 15.2 6.4-28.9
7 0.44 4.6 2.5-6.9 17.6 7.2-32.0
10 0.41 3.9 2.2-5.7 20.5 8.5-34.8
20 0.35 2.8 1.6-4.1 24.0 10.7-41.6
50 0.27 1.8 1.1-2.5 30.8 13.8-49.2
100 0.21 1.2 0.7-1.6 33.5 16.3-51.2
𝐹! = 𝑀! + 𝑀! 𝑔, where 𝑔 is the gravitational
acceleration. As a result effective peeling force can be
expressed as;
𝐹!
!
= 𝐹!
!
− 𝐹! (17)
The equations above are valid if the bending moment is
negligible. However, in some cases bending stiffness is
dominant and the elastomer can peel off the surface without
any extra effort when running on a wheel due to the
bending moment. which does not add any adhesion to the
robot. If the diameter of the wheel is equal or less then 𝑅!
then the elastomer peels off the surface by itself due to the
high curvature. The wheel radius must be much larger than
𝑅! to minimize the effect of bending moment. There is a
relationship between tread thickness and minimum wheel
radius as given in equation 6. By using equations 6 and 7 it
can be seen that 𝑀 and 𝑅! are linearly dependent upon each
other when tread dimensions and material are defined. In
this project, wheel radius (𝑅!) is chosen to be 𝐾 times
larger than 𝑅!, so that the effect of peeling due to the
bending is diminished. However, note that the bending
moment’s effect cannot be fully removed.
Fig. 5. Plot of elastomer thickness vs. wheel radius for pure moment
detachment
Fig. 6. Plot of wheel radius vs effective peeling force.
Fig. 7. Plot of thickness vs effective peeling force.
𝐸 = 110 𝑘𝑃𝑎, 𝑊! = 70 𝑁 𝑚 , 𝑤 = 10 𝑚𝑚, 𝑅!/𝑅! =
20, and 𝜌 = 1000 𝑘𝑔/𝑚!
used to plot Fig.5, 6 and 7.
As seen in the Table 1., 𝐹!
!
makes a peak when
𝑅!/𝑅! = 1, however, in reality the elastomer peels off the
surface due to pure bending, in this case. In other words, it
would not improve the climbing performance due to self-
peeling. As 𝑅!/𝑅! ratio increases the effect of the bending
moment, effective peeling force and elastomer thickness
decreases and wheel radius increases. When 𝑅!/𝑅! = 20,
to get maximum 𝐹!
!
, 𝑡 and 𝑅! have to be set to 2.8 and 24
mm respectively. In addition, instead of getting maximum
𝐹!
!
, a range could be defined to get at least %90 of 𝐹!
!
. In
this case, the designer would be more flexible when
designing a robot and it would be easier to diminish the
effect of the bending moment. For example, when
𝑅!/𝑅! = 20, instead of having 0.35 𝑁 as an effective
peeling force, getting at least 0.315 𝑁 broadens the
possibilities of the elastomer thickness from 1.6 to 4.1 𝑚𝑚,
and the wheel radius from 10.7 to 41.6 𝑚𝑚. To minimize
the moment effect, it would be logical to choose the
minimum elastomer thickness with the maximum wheel
radius as long as the other design criteria allows.
Robot Design4.
Chassis Design4.1.
The parameters to take into account when designing
chassis for climbing robots are:
• Weight,
• Tail position and length,
• Hierarchal compliance,

5. • Position of the center of gravity.
Manufacturing constraints are taken into account when
designing the chassis. A 2-axis laser cutter (Gravograph
LS100Ex) is used for cutting a 3mm Plexiglas sheet. In
order to provide integrity and easy assembly, sheets are
designed taking into account 2-axis manufacturing.
The climbing robot is designed in a way that consists of
two mirrored modules connected to each other by a shaft.
Thus, even though the robot has four wheels it acts like a
tripod by means of compliance. Surface transitions will
ease, climbing safety will increase and it will be more agile.
Fig 8. Rocker Boogie
For stabile climbing the motors are assembled inside the
wheels. Also, in order to increase adhesion area and
compliance to surface discontinuities, foam inserts are used
between wheel hubs and elastomer pads.
With the help of flexible structures such as foam and
stitching of elastomer pads which hold them in place, a
configuration similar to climbing robots with legged
mechanism is created.
Adhesion Force/Weight-Dimensional Optimization4.2.
In climbing robots the area of elastomers which provide
adhesion is proportional to the robot’s length squared (𝐿!
).
Additionally, the weight of the robot is proportional to the
robot’s length cubed (𝐿!
). Due to necessary equipment to
be used on the robot such as the electronic board, battery,
wireless communication board, and sensors, the robot`s
weight has a minimum value of approximately 75 grams
without a chassis, motors, tires and tails etc.
Hence, there must be an optimum point where the
adhesion force is maximized proportionally to weight.
𝐿 → 0 𝑚!"#"$ ≅ 75𝑔𝑟 (18)
𝐿 → ∞
!!
!! → 0 (19)
Fig. 9. Chassis Length L!, Chassis Width L!, Chassis Thickness L!,
Tire Contact Region Length L!, Tire Contact Region Width L!, Tail
Length L!, Tail Thickness L!, Tail Width L!
The robot`s total weight 𝑚!, weight of the equipment to
be used on the robot (wireless communication, sensors,
etc.) 𝑚!, robot`s chassis weight 𝑚!, weight of the motors
𝑚!, weight of the robot`s tires 𝑚!, weight of the tail 𝑚!,
weight of the battery is 𝑚!,
m! = m! + m! + 3m! + 2m! + m! + 𝑚! (20)
All calculations are done according to one module of the
robot.Rotation force for each tire can be calculated as;
𝐹!"! = 𝐹! (21)
𝐹!, is determined by adhesion model.
Motor Selection4.3.
The motor torque, which is necessary to overcome
adhesion force on each wheel, can be determined as the
formula below. Motor power (𝑃!, motor torque 𝑇!, and
angular speed 𝜔, tire diameter 𝑟!,
𝑇! ≅ 0.93𝑟! 𝐹! (22)
𝑃! = 𝑇!. 𝜔 (23)
𝑃! = 0.93𝑟! 𝐹! . 𝜔 (24)
𝑚! = 𝐴! 𝑃!
!
+ 𝐵! 𝑃! + 𝐶! (26)
Motors which can be found easily for these kinds of
applications are shown in the graph according to their
power density.

6. Fig. 10. Motors Power Density Graph
Equation for motors can be written as below:
𝑚! = 0.13430𝑃!
!
+ 3.64426𝑃! + 7.44339 (27)
Battery Selection4.1.
Batteries which can be found easily for these kinds of
applications are shown in the graph according to their
power density.
Fig. 11. Battery Power Density Graph
𝑚! = 6.3382𝑃!
!
− 2.4266𝑃! + 10.647 (28)
Thus, all variables are parameterized by dimensional
variables. Equations related to weight, adhesion force and
robot`s dimensions will be used to optimize 𝐹!"!/𝑚!
relation.
Implementation5.
In order to achieve dimensional optimization, some
initial conditions should be determined. These initial
conditions are determined according to the abilities the
climbing robot should have as inner and outer surface
transition. In order to achieve an outer surface transition,
the distance between tire centers must be 2.3 times the
length of the tire diameter as shown in Figure 5. If it is
more than that value the chassis will touch the corner. If it
is smaller there is a risk of two elastomers sticking to each
other and causing the robot to fail.
𝐿! length is taken as equal to 𝐿!.
Fig. 12. Effect of relation between tire radius and 𝑳 𝟒 to surface transition.
Desired speed for climbing robot, no matter what the size
is, 0.5 Length/second is chosen. In order to achieve this
speed when length between tire centers is 2.3 times the
length of the tire diameter, angular speed should be 1.15
rad/s.
Fig. 13. Relation Between tire radius, r and 𝐿!
Adhesion area length 𝐿!, in order to provide safe adhesion
and prevent elastomers from front and rear tires sticking to
each other, is accepted as 1.1 times as much as the tire
radius.
In the optimization code, to keep the chassis rigidity set, as
chassis length 𝐿! increases proportionally to chassis
thickness 𝐿!, a relation of chassis deflection and chassis
length of 𝛿/𝐿! 0.05 is maintained and is recommended by
the manufacturer of Plexiglass.
𝛿 =
!!!
!
! !!"
(29)
𝑥 =
!!
!
(30)
!
!!
=
!!!!
!!
= 0.05 (31)
1.1r=
𝐿
r
2.1r2.8r 2.3r
𝐿! =

7. 𝐿! =
!".!∆!!!
!!!!!
!!!
!
(32)
Dimensional optimization code is written and run in Matlab
software.
Fig. 14. Adhesion Force/Weight (N/N), Adhesion Force-Weight (Kg)
Graph
As it can be seen in the figure, in optimization code the aim
was to maximize the ratio and difference of adhesion force
to weight. It physically means respectively how many times
its own weight a robot can adhere to the surface and how
much payload it can carry.
According to the optimization results, the robot`s length
must be between 7-12 cm for the highest possible ratio of
adhesion force to weight.
For maximized climbing safety, when the robot`s length is
10.5 cm adhesion force proportional to weight is
maximized.
If it is desired to maximize the payload capacity, the robot
must be between 27 and 33 cm. When the robot`s length is
28.3 cm, weight carrying capacity is maximized by 1.45 kg.
Fig. 15. Chassis design and assembly
Hierarchal compliance5.1.
In order to have high adhesion force climbing robot`s limbs
must comply with different factors. Surface discontinuities,
manufacturing defects and robot`s rigidity may lead to
adhesion difficulties. These effects can be faced on a wide
scale. In order to overcome these hindrances, a different
method can be employed. Moving linkages, softer materials
such as foam for the tires and compliance of elastomer pads
can be utilized to overcome these discontinuities.
Hierarchal compliances are utilized as cm-mm-µm in
chassis, tyre foam and elastomers to provide higher
adhesion forces.
Fig. 16. Hierarchal compliance
Force Analysis6.
In climbing robot design most important parameters are
robot`s weight, the height of center of gravity from surface
and elastomer`s dimensions.
As the center of gravity moves away from climbing surface
the moment that causes front wheels loose adhesion
increases. Force distribution and elastomer`s adhesion force
is a key factor in stable climbing.
Therefore the forces on elastomer and robot should be
analysed.
Forces on the robot and elastomer can be categorized as
below;
• Forces due to robot’s weight,
• Tail preload force,
• Forces due to motor torque,
• Elastomer peeling force.
Force analysis are similar to the Tankbot climbing robot .
Free body Diagram6.1.
Force distribution on front and rear tires due to gravity can
be seen below.
Sensors and Functions7.
Infrared Proximity Sensors7.1.
A QTR-1A sensor is used as a proximity sensor attached
to the bottom of the robot next to each tire. It is used to
determine tail actions according to the deviation between
Boy(m)
AdhesionForce-Weight(Kg)
Length (m)
AdhesionForce/Weight(N/N)
cmmmµm
Elastomer Foam Chassis

8. each modulus’s front and rear sensors. Also by measuring
the proximity to the climbing surface, climbing safety can
be measured experimentally and presented wirelessly on
the control panel in real time.
Active Tail Control7.2.
In order to take advantages of rocker-boogie mechanism
robot has to have two tails which can operate individually.
PID controller using Ziegler and Nichols’ method utilized
in Labview Software.
PID controller works injunction with two proximity
sensors attached to robot’s chassis next to the tires and
faced down to the climbing surface. Front and rear sensor
values differ according to the chassis angle to the climbing
surface. Error value is calculated via difference between
front, rear tire sensor values and set point proximity value.
Set point value can be adjusted from front panel.
Error=(Front Sensor Value-Rear Sensor Value)-Set Point
Value
PID controller act when front tire is getting away from
the surface comparing the rear tire. Kc, Ti, Td values are
controller constant, integral time (m), reaction time (m)
respectively. These values are being used in calculation of
motor output value. Transfer function can be seen below.
G!"#$%"&&'% s = K!(1 +
!
!!!
+ T!s)
PID values are found by Labview PID Auto Tune
Wizard using Ziegler and Nichols’ method. The wizard
calculates reaction time τ, time constant Tp, thus using
Ziegler and Nichols’ PID values can be found below in the
table.
Table 2
PID Constants and Values
Kc Ti Td
PID 1.1Tp/τ 2.0τ 0.5τ
PID Values 46.92377 0.033992 0.008158
Tests8.
Tests consist climbing on vertical surface, different
surface transitions such as from ground to vertical surface,
from vertical to ceiling, from vertical to horizontal surface,
from horizontal to vertical surface and overcoming
obstacles on climbing surface, payload carrying capacity.
Also different surfaces tested to simulate different surface
roughness such as concrete surface, wood surface and
metal surface.
For example, climbing from horizontal wooden surface
to vertical wooden surface and from horizontal wooden
surface to vertical concrete surface are tested as well as
vertical to ceiling transitions.
Table 3
Climbing Robot Parts and Equipment Weight
Part/Equipment Pcs Weight(gr) Sub Total(gr)
Chassis 2 18.80 37.60
Tire 4 6.26 25.04
Foam 4 2.52 10.08
Elastomer 4 10 40
Tail 2 4.42 8.84
Arduino Nano 1 6 6
Xbee Shield 1 14 14
Motor Shield 1 17 17
Accelerometer 1 1 1
Proximity Sensor 4 1 4
Li-Po Batarya 1 54 54
Tire Motor 4 10.48 41.92
Tail Motor 2 9.64 19.28
Total Weight(gr) 278.76
Climbing on Vertical Surface8.1.
Vertical surface climbing is tested with total weight of
278 gr at the speed of 5cm/s successfully.
Climbing on Ceiling8.2.
Inverse climbing and from vertical to ceiling transitions
are tested successfully.
Surface Transition8.3.
Surface transition can be categorised outer surface
transition and inner surface transitions. Inner surface
transition are climbing from horizontal to vertical and
vertical to ceiling transition. Outer surface transition is
climbing from vertical surface to the roof.
Payload Capacity8.4.
Payload capacity is measured if robot can carry a
payload of its own weight. It is tested with 280 gr of
payload at 3.8cm/s speed successfully.

9. Fig. 19. Vertical Climbing
Fig. 20. Outer Surface Transition
Fig. 21. Inner Surface Transition From Vertical to Ceiling
Fig. 22. Overcoming Obstacle on Vertical Surface 10mm Diameter Obstacle
Fig. 23. Overcoming Obstacle on Vertical Surface 12x23 mm Rectangular Obstacle
(a)
(b) (c) (d) (e)(a)
(b) (c) (d) (e)
(a) (b) (c) (d) (e) (f)
(a) (b) (c) (d) (e) (f)
(a) (b) (c) (d) (e) (f)

10. Results9.
In this project, we tested the effectiveness of elastomer-
using climbing robots with tires. The most important
criteria of climbing robots of inner and outer surface
transition, loitering on vertical surfaces and ceilings were
completed successfully.
Climbing speed is measured as 0.45L/s (Robot`s length
per second).
Glass, metal, acrylic, wooden etc. vertical surface
climbing and transition were accomplished.
Without an external power more than 10 minutes
operation is accepted as successful in test this value
measured up to 30 minutes.
So as to simulate obstacles and discontinuities on
climbing surface 10-20 mm diameter circular and 12x23
mm rectangular obstacles are put on the climbing surface
and successfully overcame by climbing robot.
Table 5
Criteria and Test Results
Criteria Results
Speed 0.45 L/s
Turning Radius 100 cm
Power Consumption (90˚ Climbing) 3 Watt
Loitering on Vertical Surface 2 Minutes
Loitering on Ceiling 10 Seconds
Payload Capacity 280 gram
Inner Surface Transition
Outer Surface Transition
Climbing on Ceiling
Climbing on Painted Wall
Climbing on Wooden Surface
Climbing on Metal Surface
Overcoming Obstacles on Climbing Surface
Cm, mm, µ degrees hierarchical compliances
Operating Time 20 Minutes
In comparison to the performance criteria of other
climbing robots using elastomer by means of adhesion
technique, our robots outperformed them thanks to stabile
climbing, the ability to climb a variety of surfaces, high
climbing safety on surface discontinuities with the help of
hierarchical compliance and the ability to perform surface
transitions with the help of an active tail.
Future Projects10.
The biggest challenge faced was the assembly of
elastomers to tires. We received the best results when
elastomers were stitched to the tires. Therefore, in future
work it is recommended that a more effective method to
attach the elastomer to foam that lets elastomers stretch,
avoid bending and sliding from the foam should be used.
Due to manufacturing limitations the chassis design is
provided for a 2-axis laser cutter and a 2-axis assembly. In
future projects it is recommended that the use of 3D
printers may achieve a lighter and more compact structure.
Acknowledgement:
This research is supported by Scientific and
Technological Research Council of Turkey (TUBITAK)
research grant number 110E186. The authors would like to
thank TUBITAK for the support.
References
[1] Unver, O. “Design and Optimization of Miniature Climbing
Robots using Flat Dry Elastomer Adhesives,” Carnegie Mellon University
422 Scaife Hall, 5000 Forbes Avenue Pittsburgh, PA 15213 2009.
[2] Roger A. Sauer “The peeling behaviour of thin films with finite
bending stiffness and the implications on gecko adhesion,” The Journal of
Adhesion, Aachen Institute for Advanced Study in Computational
Engineering Science (AICES), RWTH Aachen University, Templergaben
55, 52056 Aachen, Germany, 2010.
[3] Prof. Dr. habil. Jörg Roth, www.wireless-earth.org University of
Applied Sciences Nuremberg Keßlerplatz 12 D 90489 Nuremberg
Germany, 2012.
[4] National Instruments Labview PID Control Toolset User Manual,
November 2001 Edition, 11500 North Mopac Expressway Austin, Texas
78759-3504 USA Tel: 512 683 0100.
[5] Unver, O. and Sitti,M. “Tankbot: A Palm-size, Tank-like
Climbing Robot using Soft Elastomer Adhesive Treads,” The International
Journal of Robotics Research published online 24 September 2010.
[6] Unver, O. and Sitti,M. “A Miniature Ceiling Walking Robot with
Flat Tacky Elastomeric Footpads,” Department of Mechanical
Engineering Carnegie Mellon University 422 Scaife Hall, 5000 Forbes
Avenue Pittsburgh, PA 15213, 2009.