Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Mechanism and Machine Theory 124 (2018) 197–210

Contents lists available at ScienceDirect

Mechanism and Machine Theory


journal homepage: www.elsevier.com/locate/mechmachtheory

AnyClimb-II: Dry-adhesive linkage-type climbing robot for


uneven vertical surfaces
Yanheng Liu a, TaeWon Seo b,∗
a
School of Mechanical Engineering, Yeungnam University, Gyeongsan 38541, Republic of Korea
b
School of Mechanical Engineering, Hanyang University, Seoul 04768, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: Vertical wall surfaces with obstacles present a serious challenge for wall-climbing robots.
Received 17 July 2017 Owing to their limitations in overcoming obstacles, these types of robots have not been
Revised 16 January 2018
commercialized yet. Several ideas on novel designs and precise control have been sug-
Accepted 28 February 2018
gested; however, further research is required to achieve enhanced robot capabilities in
overcoming obstacles. Specifically, the use of dry adhesive methods by wall-climbing
Keywords: robots to climb over obstacles present tremendous challenges. This study introduces the
Wall-climbing robot design of a new linkage-type, wall-climbing robot, based on dry adhesion, for uneven ver-
Linkage-type design tical surfaces. Based on a four-bar mechanism, repeated walking is achieved via a single
Compliant mechanism actuator. The robot’s most important feature is the linkage design used for climbing over
Four-bar mechanism obstacles, which has been adopted from rover running patterns. The symmetric linkage
Obstacle-overcoming
design renders the robot adaptable to uneven surfaces with a compliant motion. Addi-
tionally, flat dry adhesives were used for the attachment mechanism. The design param-
eters were determined based on kinematic and static analyses, and certain important is-
sues in linkage-type wall-climbing robot designs were addressed. The robot’s performance
was verified using experiments, whereby it was able to climb up and go down stairs with
maximum stair heights of 15 mm (equal to 0.6% of the robot’s height) during open-loop
vertical walking. We expect that the linkage design can extend the accessible area of the
wall-climbing robot.
This study introduces the design of a new linkage-type, wall-climbing robot, based on
dry adhesion, for uneven vertical surfaces. Based on a four-bar mechanism, repeated walk-
ing is achieved via a single actuator. The robot’s most important feature is the linkage
design used for climbing over obstacles, which has been adopted from rover running pat-
terns. The symmetric linkage design renders the robot adaptable to uneven surfaces with
a compliant motion. Additionally, flat dry adhesives were used for the attachment mech-
anism. The design parameters were determined based on kinematic and static analyses,
and certain important issues in linkage-type wall-climbing robot designs were addressed.
The robot’s performance was verified using experiments, whereby it was able to climb up
and go down stairs with maximum stair heights of 15 mm (equal to 0.6% of the robot’s
height) during open-loop vertical walking. We expect that the linkage design can extend
the accessible area of the wall-climbing robot.
© 2018 Elsevier Ltd. All rights reserved.


Corresponding author.
E-mail address: taewonseo@hanyang.ac.kr (T. Seo).

https://doi.org/10.1016/j.mechmachtheory.2018.02.010
0094-114X/© 2018 Elsevier Ltd. All rights reserved.
198 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

1. Introduction

Obstacles on a vertical wall present a challenge for wall-climbing robots (WCRs). Since there is no support on the oppo-
site side of the vertical wall, even a slight loss of contact causes WCRs to fall on the ground. Correspondingly, many WCRs
are used only on flat vertical walls. This characteristic limits the potential areas of application of WCRs. To extend the ar-
eas of operation of WCRs for searching, inspecting, and exploring, it is necessary that they possess capabilities for climbing
flat vertical walls and overcoming obstacles. Some reliable solutions have been proposed, such as magnetic climbing (e.g.,
wheel-type [1] or track-type [2]), and vacuum suction-type climbing (e.g., bipedal walking [3] and multilink [4]), however,
the obstacle-overcoming abilities of dry-adhesive-based WCRs have not been successfully developed yet [5–8].
The pioneers of dry-adhesive WCRs used quadrupedal or hexapedal walking locomotion. Stickybot [5] used a quadrupedal
locomotion that mimics the gecko lizard locomotion, and its directional attachment mechanism. Geckobot [9,10] employed a
similar quadrupedal locomotion, but its mechanism was simpler than that used by Stickybot, as it operated the moving and
preloading parts independently. Recently, the same researcher who developed Geckobot developed Abigaille-III [11], which
uses hexapedal locomotion with four front feet and two rear feet. The Abigaille-III has an advanced obstacle-overcoming
ability with one foot used to advance, and the other five remaining feet to adhere. The legged walking locomotion is a
very reliable solution for a flat surface, but it has a limited obstacle-overcoming ability at a relatively low speed. Moreover,
the walking locomotion required precise feedback-control [11] to determine the position of the feet during locomotion—the
slightest failure in its attachment invariably causes the robot to fall.
To resolve the problems of low-speed and high-complexity in the control of walking WCRs, wheel–leg mechanisms were
proposed. The most famous example is that of Waalbot [12], which had three footpads attached on the rotating wheel,
whereby the use of the two wheel-legs (Whegs) allowed the robot to climb and steer. The Whegs mechanism [13,14] con-
stitutes a successful example of wheel-leg mechanisms in WCRs. Originally, Whegs with triangular spoke wheels were devel-
oped for ground locomotion, i.e., for running on uneven surfaces. The mechanism was extended to triangular and rectangular
spoke wheels with dry adhesives on the spokes. The Whegs could climb a vertical wall and perform wall-to-wall transitions
by using the wheel-leg mechanism. The wheels’ legs overcame the drawbacks of walking locomotion satisfactorily, however,
their ability to climb over steps and grooves was limited.
Track mechanisms can be a solution for fast and reliable wall-climbing. Tankbot [15] presented very high speed wall
climbing with an enhanced ability to overcome obstacles. Seo and Sitti [16] extended the Tankbot mechanism by adding
a directional compliance with a tail, and the results yielded enhanced abilities for transitioning and for overcoming ob-
stacles. Furthermore, the robot achieved increased speeds and payloads as it climbed on vertical walls. Fabrication of the
track can be a problem for the robot, but there are technical reports describing methods to fabricate the track with fiber
structures [17,18]. Additionally, Lee et al. [19] extended the MultiTrack for high payload, high-rise building cleaning by using
active joints to achieve wall-to-to transition performance. Track mechanisms with a compliant design can achieve high-speed
climbing with a high obstacle overcoming ability, including steps and gaps. However, fundamentally, the track mechanism
has an increased potential of failure compared with the walking mechanism as even a slight failure of the track propagates
the entire track and ultimately causes the robot to fall.
In this study, we extended our previous WCR research on AnyClimb [20] for climbing curved surfaces with a single plat-
form, and we developed a new WCR with an enhanced ability to climb over obstacles. There are many arguments that
favor the necessity of a minimal robot WCR design. By reducing the number of actuators, which is the main factor for
extra weight, the safety margin of the WCR could be increased. If we can design a mechanism to adapt uneven surfaces
autonomously without any feedback control, this would effectively reduce the cost of computation. According to the de-
scriptions listed earlier, to-this-date, there has been no WCR with climbing abilities over obstacles using linkage-type and
dry-adhesive mechanisms, as shown in Table 1. Therefore, in this research a climbing robot has been developed that can
cross large obstacles (ridges and grooves). Additionally, we have reduced the robot’s weight to improve the climbing per-
formance without any controllers. Via a symmetric compliant linkage mechanism design, the requirements of the proposed

Table 1
Design requirements and existing wall climbing robot performance.

ROBUG II [27] Stickybot AnyClimb Tank-Like robot Waalbot Proposed robot

Size Lager Small Miniature Miniature Miniature Miniature


Weight(g) 17,0 0 0 400 142 180 70 138
Type Legged Legged Linkage Track Wheel-leg Linkage
Speed (mm/s) 16.7 40 12.5 60 60 ≤ 20
Adhesion Vacuum suction Dry adhesion Dry adhesion Dry adhesion Dry adhesion Dry adhesion
Controller Yes Yes No Yes Yes No
Large obstacles Yes No No Yes(50 mm) No Yes
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 199

Fig. 1. Configuration of the mechanism of AnyClimb-II. a: Outer frame, b: inner frame, c: timing belt, pulley, and motor, d: connecting rod, e: parallelogram
four-bar mechanism, f: symmetric compliant linkage mechanism, g: footpads, and h: tail.

robot can adapt on uneven surfaces with large obstacles even without the use of any controllers. To reduce the number
of controllers, a four-bar minimal walking mechanism [21] was used for climbing locomotion. The basic conceptual design,
analysis, parametric design, and experimental results, will be discussed in the present work. Concurrently, there are many
references on the use of minimal designs for steering [23], jumping [24], stair-climbing [25], and vertical wall-climbing [26],
which present the effectiveness of a minimal design approach.
The rest of the study is organized as follows. Section 2 describes the configuration of the mechanism for wall-climbing
and for adapting on uneven walls. Section 3 presents the kinematic and the static analysis, and the elicited results are
used to determine the design parameters to perform safety calculations. Section 4 presents the prototype assembly and
the experiments on different cases of wall-climbing. Concluding remarks and the proposed future work are proposed in
Section 5.

2. Mechanism configuration

2.1. Overall configuration

As shown in Fig. 1, the robot has been redesigned to achieve excellent performance by reducing its weight, and by
allowing it to adapt to uneven surfaces. This was achieved based on the findings of our previous research [20], whereby the
torque transfer system used one motor, one belt, two gears, and parallelogram four-bar mechanisms, to connect two body
frames on the robot’s platform, thereby reducing the robot’s weight. The most important design attribute to allow the robot
to climb on uneven surfaces is the employment of the symmetric compliant linkage mechanism.
The configuration of the mechanism of AnyClimb-II mainly consists of a rectangular body frame and dry adhesion foot-
pads. The rectangular body frame is divided into two parts, namely the inner frame and the outer frame that are connected
by the parallel four-bar mechanism. Owing to the rotation of the parallel four-bar mechanism, which is located between the
inner and the outer body frames, the robot can achieve forward motion. On the inner body frame, we assembled the belt.
Gears and one DC motor were used to provide the rotation force for the parallel four-bar mechanism. To adapt on uneven
surfaces, the robot should be able to cross the obstacles at the same time, and perform the climbing motion in the vertical
direction. Therefore, we designed the symmetric compliant linkage legs to allow it to adapt to uneven surfaces. The compli-
ant linkage mechanism can rotate in a plane around the horizontal direction that lies along the body frame. Furthermore,
the connecting rods at the linkage leg top were used to ensure that the footpads of each body frame can instantly detach
from or attach to the surface. To reduce the robot’s pitch-back moment on the outer frame, a tail was also assembled.
Moreover, owing to the adhesion of the footpads, the robot could climb on the vertical wall surface and not fall off. The
dry-adhesive system will be summarized in Section II-B.
200 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

 $ %


 DGKHVLRQ
DUHD



D E F
Fig. 2. VytaFlex-10 elastomer and its fabrication methods. (a) Fabrication process, (b) manufactured V-10 footpad, and (c) assembled WCR footpad.

,QQHUIUDPH
/RFRPRWLRQ
2XWHUIUDPH
GLUHFWLRQ

D E F G

Fig. 3. Vertical wall-climbing locomotion. (a) Outer body attaches and inner body moves forward, (b) inner body hits the surface and both attach on the
wall, (c) inner body attaches and outer body moves forward, and (d) outer body hits the surface and attaches onto it, thereby completing one cycle of
walking locomotion.

2.2. V-10 dry elastomer adhesive

In the present work, we used the dry elastomer VytaFlex-10 (V-10, Smooth-on Inc.) to create the adhesive footpads. This
adhesion method has already been validated successfully in previous research studies [21,22], and can attach on different
materials (wood, glass, acrylic). Uses of different adhesive thicknesses lead to different adhesive capacities. In other words,
the elastomer is a sticky material and its adhesion properties are used by the wall-climbing robot. All the V-10 footpads
are well cleaned by alcohol before the experiment, and the contamination of the footpads degrade the adhesion force. The
fabrication method of the V-10 elastomer is shown in Fig. 2. The mixing components comprise two parts. To produce the
V-10 elastomer, we mixed equal amounts of parts a and b at a mix ratio of 1:1 (Fig. 2(a–1) and (a–2)). We then let it rest for
one day to harden. However, since bubbles are expected to emerge as a result of mixing, an exhaust process should be added
using the vacuum pump and chamber (shown in Fig. 2(a–3)) to prevent their presence and the elastomer from becoming
excessively stiff. Finally a laser-cutting machine was used to cut the footpads to appropriate sizes (shown in Fig. 2(a–4) and
(b)) for the climbing robot design. Finally, Fig. 2(c) illustrates the elastomer’s footpad assembled on the proposed WCR.

2.3. Vertical climbing locomotion

As shown in Fig. 3, the parallelogram four-bar linkages are located between the inner body frame and the outer body
frame with one degree-of-freedom (1-DOF). When the drive shaft that connects the parallel linkages is actuated, the linkages
rotate and the robot achieves walking locomotion. Fig. 3(a) illustrates the initial state, where only the outer body is attached
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 201

Fig. 4. (A) Overall process of the symmetric compliant linkage mechanism of one body frame (inner or outer) adapted on an obstacle. (1),(3) views at
which the compliant linkage exists above the obstacle along with the body frame (inner or outer) so that only one foot can withstand the force from the
obstacle, thereby causing the symmetric compliant linkage to be rotated, as shown in (2) and (4). (B) Compliant adaptation on uneven surfaces based on
compliant link rotation. Views from (b) to (a) indicate how the robot climbs up the step, while views (d) to (c) show how the robot climbs down the step.

on the vertical surface. As shown in Fig. 3(b), the outer body frame is attached on the surface. As the link rotates, the inner
body hits the surface and adheres to it. The outer body frame then detaches from the surface, and moves forward with
the link rotating, as seen in Fig. 3(c). Until the instant at which the outer frame contacts the surface, the two bodies are
pressed against the wall to completely become attached from the surface, as shown in Fig. 3(d). The robot then completes
one cycle of locomotion, with the parallel four-bar linkage rotated by 360°. It then proceeds to the next cycle to climb along
the direction indicated by the red arrow. This mechanism is activated using only one DC motor to reduce robot weight, and
without the use of any controllers, as shown in Fig. 11.

2.4. Adaptation on obstacle surfaces

To ensure locomotion on uneven surfaces, the symmetric compliant linkage mechanism was designed and used that
guaranteed that all the footpads of the leg structure attach on the surface, and that the obstacles are overcome, as shown
in Fig. 4. The links PM and QN are 1-DOF joints and can be fixed at their midpoints on the robot’s body frame. When the
symmetric compliant linkage mechanism belongs to the body frame (inner or outer), the generated forces under the front
and rear foot are not balanced owing to the obstacle, as shown in Fig. 4(A) (1) and (3). The linkage will rotate clockwise
or counterclockwise at the link midpoints shown in Fig. 4(A) (2) and (4), so that the front and rear feet can adapt to the
obstacle.
Furthermore, the robot’s motion follows the direction indicated by the arrow, climbing up or down a vertical surface, as
shown in Fig. 4(B). The symmetric compliant linkage changes from a rectangular to a rhomboid parallelogram configuration
owing to their rotation with respect to the joints to adapt to the uneven surface. Fig. 4(B) ((b) to (a)) shows how this robot
goes up a step based on the rotation of links in the clockwise direction, while Fig. 4(B) ((d) to (c)) shows how this robot
goes down a step based on the rotation of the links in counterclockwise direction to adapt to the steps.

3. Analyses

3.1. Kinematic analyses to avoid interferences

To overcome obstacles on vertical surfaces, the symmetric compliant-linkage kinematic lengths constitute important pa-
rameters for the WCR. As shown in Fig. 5, when the robot climbs up the obstacle, this symmetric linkage should rotate to
transform from a rectangular (Fig. 5(a)) to a rhomboid parallelogram (Fig. 5(b)). The goal of the present work was the design
of a robot to be able to climb the maximum obstacle height of 10 mm. In other words, if the robot overcomes the purposed
obstacle safely, the minimum height difference between the front and the rear foot should be equal to 15 mm. Therefore,
    
in Fig. 5(b), OPP OMM we set the following condition for the distances PP and MM pertaining to the triangles OPP and

OMM :

LPP + LMM ≥ h, LP P  = LM M  (1)


202 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

D E

2 0
2 0 ′ ′
3
3
1
4 ′ ′
1
VWRSSHU 4

Fig. 5. Kinematic model used to determine the interfering postures.

D E
IURQWIRRW UHDUIRRW

9 9

3UHORDG 3UHORDG

Fig. 6. Force distribution on the V-10 elastomer footpads. (a) Front elastomeric footpad, (b) rear elastomeric footpad.Ffmg and Frmg are the effects of the robot
weight on the front and rear footpads, respectively. Correspondingly, Ffm is the normal force on the front footpad and Frm is the normal force on the rear
footpad from the motor’s side. Ffs and Frs are the front and the rear normal forces owing to the torsion spring. Fft and Frt are the normal force distributions
on the front and the rear footpads owing to the tail. Ff p and Fr p are the peeling forces on the front and the rear footpads.

since the obstacle height h was set to 15 mm. Therefore, LPP and LMM could not be less than 5 mm. The rotation angle θ is
expressed as follows:
2LP P 
θ = tan−1 (2)
L1
where h is the height of the obstacle, θ is the rotation angle of the symmetric link, and LPP and LMM are the distances that
vary with link rotation. In addition, L1 and L2 are the lengths of the links of the symmetric linkage mechanism, as depicted
in Fig. 5(a).
In addition, the shafts of the parallel linkage mechanism have a different role compared to the shafts connected to the
body, and assist the stoppers in the symmetric compliant linkage mechanism, as shown in Fig. 5. When the footpads detach
from the surface of the wall, the stoppers can prevent the symmetric linkage from rotating excessively, and from causing the
detachment to failure. The distance between the two stoppers depends on the length of the belt in the inner body frame
because the motor should be located at the center of the body. Therefore, the minimum length of link L1 could not be less
than the fixed distance between the two stoppers. Furthermore, the links of the compliant mechanism were set to a certain
width, with the length of link L2 equal to 20 mm, and the length of link L1 equal to 60 mm.

3.2. Static analysis

In the present work, V-10 was selected as the proposed climbing robot adhesive method, as previously mentioned. The
physical characteristics of V-10 have been presented in a previous research publication [22]. As an elastomer, it has adhesive
properties and a hysteretic response that is different from the preload. Correspondingly, based on its characteristics, it has
been used as a climbing robot that operates in accordance to the dry adhesive method. Otherwise, the relationship of the
normal force preload and adhesion is defined as follows [22]:
Pa = A(Pp )B (3)
where Pa and Pp are the adhesion pressure and the preload pressure effect on the V-10 footpads, respectively. In addition,
from the experimental data, the constant A is 9.86 and B is 0.269.
During the vertical-climbing process, the adhesive force plays a significant role in the robot’s climbing stability. To de-
termine the adhesive force, the preload on the elastomer footpads should be analyzed, as shown in Fig. 6. Based on the
distribution of the forces on the elastomeric footpads, there are five main forces affecting the preload, namely, the weight
of the robot (Ffmg , Frmg ), the tail preload force (Fft , Frt ), the motor torque (Ffm , Frm ), the peeling force (Ff p , Fr p ), and the com-
pliant joint spring torque (Ffs , Frs ). Therefore, the front (Ff ) and the rear (Fr ) footpad preload forces can be expressed as
follows:
Ff = Ffmg + Ffm + Fft + Ffs + Ff p
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 203

Fr = Frmg + Frm + Frt + Frs + Fr p (4)


Hence, the front and the rear normal forces acting on the elastomeric footpads owing to the robot weight are calculated
as
(mog + mi g) cos α
(moho + mi hi )g sin α
Ffmg = −
2 L1
(mog + mi g) cos α (moho + mi hi )g sin α
Frmg = + (5)
2 L1
where Ffmg and Frmg are the effective forces of the robot’s weight on the front and the rear footpads, respectively. Moreover,
mo and mi are the outer and inner frame masses, ho is the center height of the outer frame weight, hi is the center height
from the surface of the inner frame weight, and α is the slope of the wall’s surface.
The force distribution on the elastomeric footpads is affected by the motor’s torque that acts on the front and the rear
footpads, and can be expressed as
Tm
Ff m =
L1
Tm
Frm =− (6)
L1
where Ffm is the normal force on the front footpad, Frm is the normal force on the rear footpad, Tm is the motor torque, and
L1 is the symmetric compliant linkage length.
Based on the force transfer mechanism, the tail presses onto the surface and transfers the force to the front footpad to
improve the robot’s climbing ability. Therefore, owing to the tail force, the normal forces acting on the front and the rear
footpads can be calculated as
Lt
Fft = Ft
L1
(L1 + Lt )
Frt = −Ft (7)
L1
where Fft and Frt are the front and the rear normal forces generated owing to the tail pressing onto the surface. Moreover,
Ft is the force with which the tail pushes the surface, and Lt is the distance between the tail and the rear footpad.
According to the above, there is a compliant joint that was designed using a torsion spring. Thus, the torsion spring gives
rise to the force distribution on the footpads, which can be calculated as
k θ
Ff s = −
L1
k θ
Frs = (8)
L1
where Ffs and Frs are the front and the rear normal forces owing to the torsion spring, respectively. In addition, k is the
stiffness coefficient of the spring, and θ is the rotation angle.
The robot has two sets of feet, namely, the inner-frame and the outer-frame feet. During walking locomotion on a ver-
tical surface, if the outer-frame feet are the ones that perform the attachment, then the inner-frame feet will perform the
detachment at the same time. Therefore, the peeling force of the inner-frame feet acts as the preload to the outer-frame
feet to achieve attachment. Therefore, the peeling force Fp can be calculated as follows [28]:
 F 2 1 F 
(1 − cos θ p ) − R = 0
p p
+ (9)
w 2tE w
where Fp is the peeling force, w is the elastomeric footpad width, t is the elastomeric thickness, θ p is the angle between
the surface and the pulling direction of the elastomer, E is the Young’s modulus of the V-10 that is experimentally derived
to equal 110 kPa, and R is the adhesive energy that equals 70 N/m.

3.3. Failure analysis

To analyze the robot’s climbing performance on flat and uneven surfaces, the static normal forces at the front and the
rear footpads were calculated. Fig. 7 shows the free-body diagram for the static analysis of the present WCR on a flat surface.
The normal force under the footpads can be calculated using Eqs. (10)–(11).
 L  
Fr = −mog sin α ha − mi g sin α hd − (mo + mi )g cos α
1
+ Ft (Lt + L1 ) /L1 (10)
2

Ff = (mo + mi )g cos α − Ft + Fr (11)


204 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

/RFRPRWLRQ
GLUHFWLRQ

0
1 1

D E

Fig. 7. Free-body diagram for static analysis on a flat surface. (a) Attachment of outer-frame footpads, and (b) attachment of inner-frame footpads.

where Fr is the rear footpad normal force, Ff is the normal force under the front footpad, L1 is the distance between the
front footpad and the rear footpad, and Lt is the distance of the tail from the rear footpad. In addition, mi is the inner-frame
mass, mo is the outer-frame mass, and α is the slope of the wall surface. Moreover, ha and hd are the distances of the center
of gravity of the attached and the detached bodies from the wall surface, respectively, as shown in Fig. 7.
However, when the inner-frame footpads attach on the surface, the normal force under the rear footpad change in accor-
dance to (11):
 L  
Fr = −mog sin α hd − mi g sin α ha − (mo + mi )g cos α
1
+ Ft (Lt + L1 ) /L1 (12)
2
As shown in Fig. 8, when the symmetric linkages rotate to overcome the step, the distance of the robot’s center of gravity
from the reference surface changes. Therefore, the static equations for climbing up or down an obstacle are presented in
Eqs. (13)–(14).
 L  
Fr = −(mog + mi g) cos α cos θ − mog sin α (h a + hs ) − mi g sin α (h d + hs ) + Ft (L1 cos θ + Lt ) /(L1 cos θ ) (13)
1
2

Ff = (mog + mi g) cos α − Ft + Fr (14)

where ha = ha − ( 21 ) sin θ and hd = hd − ( 21 ) sin θ are the distances of the outer-body and the inner-body center of gravity
L L

from the reference surface when the robot climbs up the step, as shown in Fig. 8(a) and (b), respectively. In addition, hs is
the obstacle height, θ is the rotation angle of the symmetric compliant linkage, and θ = asin( Lhs ).
1
Similarly, when the robot inner-frame footpads attach to surfaces, the rear-footpad force is expressed as follows:
 L  
Fr = −(mog + mi g) cos α cos θ − mog sin α (h d + hs ) − mi g sin α (h a + hs ) + Ft (L1 cos θ + Lt ) / (L1 cos θ )
1
2
(15)

Fig. 9 shows the analysis of the normal force under the robot front and rear footpads when the robot climbs up flat
and uneven surfaces, respectively. From the results, it can be observed that the normal forces under the front and the rear
footpads are higher than zero, thereby leading to an increase in the surface slope. In other words, this WCR can climb on
flat vertical and on uneven surfaces safely, irrespective of whether the outer-body or the inner-body frame attaches onto the
surface. Furthermore, in this study, a passive tail has been designed. As shown in Fig. 9, the outer body frame attach on the
flat surface or uneven surface without tail, normal force on rear foot are less than zero cause the robot falling down from
surface. Therefore, in the analysis, we set the force generated by the tail to be equal to 1 N.
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 205

/RFRPRWLRQ
GLUHFWLRQ

0
0 1
1

D E

Fig. 8. Free-body diagram for static analysis while inclining. (a) Attachment of outer-frame footpads, while the front footpad climbs on a step. (b) Attach-
ment of inner-frame footpads, while the front footpad climbs on a step.

4. Prototype and experiments

4.1. Prototype description

Shown in Fig. 10, is the assembled prototype of the proposed robot. The main body frame and the symmetric linkage
mechanism were manufactured via a 3D printer, with a thickness of 3 mm. At the center of the inner body frame, a single DC
motor (298:1 Mini Metal Gear Motor, Solarbotics) was installed. Without the tail, the robot’s prototype size is 140 (w) × 127
(l) × 49 (h) mm and the weight is 138 g. There are two connecting carbon rods in the inner and the outer frames for the
connection of the symmetric linkage mechanism to ensure that the front and the rear footpads can detach and attach at the
same time to reduce the robot’s rotation in the vertical direction, while performing a walking locomotion, respectively. In
the symmetric compliance mechanism, a torsion spring of 5 × 10−4 N m/rad was assembled to achieve a compliant effect.
External DC power was provided to the robot to perform the climbing locomotion. Furthermore, Table 2 lists a summary of
the design parameters. As listed in Table 2, we can observe that the inner-frame mass (mi ) is larger than the outer frame
mass (mo ) because the inner frame exceeded the motor, the gear, the pulley, and the belt. There are four feet on every
robot’s body frame, made of VytaFlex-10 (V-10), as described in Section II-B, and a passive tail was installed on the outer
frame.

4.2. Flat vertical wall climbing

The experimental apparatus had been assembled prior to the conducted experiments, whereby acrylic was used as the
material for vertical surfaces. As shown in Fig. 11, in this experiment, the robot climbs up a flat surface before overcoming
the obstacle. From Fig. 11(a) to (e) the robot completes one cycle of movement by rotating in parallel a four-bar linkage
on a flat acrylic surface, as described in Section II. C. As Fig. 11(a) and (b) shows, the footpads of the outer frame in as-
sociation with the symmetric compliant linkages always maintain a rectangular configuration irrespective of whether they
are attached to (Fig. 11(a)), or detached from the surface (Fig. 11(b)). The same applies to the footpads of the inner frame
(Fig. 11(c)–(d)). Moreover, Fig. 11(f)–(j) shows top views of the robot’s walking locomotion on a flat vertical surface.

4.3. Inclining on a step

Fig. 12 shows the wall-climbing robot walking from the flat surface onto the step, based on the inclination of the sym-
metric compliant linkages. The obstacle height is 10 mm and is made of acrylic. The initial state is shown in Fig. 12(a), when
206 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

Outer Frame Attach on Flat Surface Outer Frame Attach on Flat Surface Inner Frame Attach on Flat Surface Inner Frame Attach on Flat Surface
3 3 3 3

y=0 y=0 y=0 y=0


without tail without tail withour tail without tail
2 2 2 2
with tail with tail with tail with tail
Normal force on front feet (N)

Normal force on front feet (N)


Normal force on rear feet (N)

Normal force on rear feet (N)


1 1 1 1

0 0 0 0

-1 -1 -1 -1

0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80

Slope angle of wall surface (degree) Slope angle of wall surface (degree) Slope angle of wall surface (degree) Slope angle of wall surface (degree)

Outer Frame Attach on Uneven Surface Outer Frame Attach on Uneven Surface Inner Frame Attach on Uneven Surface Inner Frame Attach on Uneven Surface
3 3 3 3

y=0 y=0 y=0 y=0


without tail without tail without tail without tail
2 2 2 2
with tail with tail with tail with tail
Normal force on front feet (N)

Normal force on front feet (N)


Normal force on rear feet (N)

Normal force on rear feet (N)


1 1 1 1

0 0 0 0

-1 -1 -1 -1

0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80

Slope angle of wall surface (degree) Slope angle of wall surface (degree) Slope angle of wall surface (degree) Slope angle of wall surface (degree)

Fig. 9. Normal forces on the robot’s footpads while climbing a flat surface and on a step. Figures in first row show robot while climbing on flat surface
without tail and with tail, and in second row show the robot while climbing on a step.

0RWRU 3DUDOOHO
,QQHU %HYHOJHDUV IRXUEDU
IUDPH %HOW OLQNDJH
&RQQHFWLQJ
URG

7DLO

2XWHU
IUDPH 6\PPHWULF
FRPSOLDQW
OLQNDJH
)RRWSDG

Fig. 10. Image of the AnyClimb-II prototype.

Table 2
Specifications of the prototype.

Parameters Values Parameters Values

mo 65 g mi 83 g
ho 35 mm hi 50 mm
L1 60 mm Lt 175 mm

the inner footpads climb on the step and cause the inner symmetric linkages to incline. However, the outer symmetric link-
ages retain their original rectangular configuration. From Fig. 12(b) to (c), the footpads of the outer body frame are in the
detachment state, whereas the inner body frame completes the detachment-to-attachment process, and the outer symmet-
ric linkages deform from the original rectangular to the rhomboid parallelogram geometry. Fig. 12(d) illustrates the robot’s
front footpads that are destined to step on an obstacle in the next locomotion cycle, and all the symmetric linkages that are
inclining to the parallelogram configuration. However, when the robot climbs up the step, the symmetric linkages return to
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 207

Fig. 11. Snapshots of the robot climbing a flat vertical wall. (a)–(e) Robot completes one cycle of motion by rotating in parallel a four-bar linkage on a flat
surface. (f)–(j) Top views of each posture. See multimedia extension for details.

Fig. 12. Snapshots of the robot climbing up the step. (a)–(d) Robot walking from the flat surface onto the step based on the inclination of the symmetric
compliant linkages. (e)–(h) Top views of each posture. See multimedia extension for details.

their original rectangular configuration, as shown in Fig. 12(d), which is the same as the case when the robot climbs up a
flat surface. Fig. 12(e)–(h) depict the top views of the footpads attaching to and detaching from the surface when the robot
climbs up the step.

4.4. Declining from a step

In the declining experiment, the robot goes down the step, as shown in Fig. 13. The initial position of the robot is on the
step, and the outer-frame footpads detach from the surface, and proceed to walk forward. The symmetric compliant linkages
on both frames attain their original rectangular geometries, as shown in Fig. 13(a). When the outer frame footpads attach
on the surface, the front footpads go down the step, whereas the rear footpads remain on the step, as shown in Fig. 13(b).
Subsequently, the outer-frame symmetric linkages begin to deform from a rectangular to a rhomboid parallelogram configu-
ration. Fig. 13(c) illustrates that the outer-frame symmetric linkages maintain the rhomboid parallelogram geometry, and the
inner-frame footpad moves forward to attach on the surface, thereby changing the linkage geometry to that corresponding to
a rhomboid parallelogram. Fig. 13(d) shows the robot’s configuration when it has completely gone down the step, whereby
the symmetric linkages maintain their rectangular geometries until the robot meets the next obstacle. Fig. 13(e)–(h) depicts
the top views of the robot going down the step.
208 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

Fig. 13. Snapshots of the robot climbing down the step. (a)–(d) the robot climbs down the step, based on the decline of the symmetric compliant linkage.
(e)–(h) Top views of each posture. See multimedia extension for details.

Fig. 14. Robot performance for climbing over different terrains. (A) Step with the same width but different height at which the robot steps on, or goes
down from it. Each of the two experiments was repeated 10 times. (B) Grooves and ridges with the same height but different widths at which the robot
crossed the groove and ridge. The two experiments were repeated 10 times each.

4.5. Robot performance when climbing over obstacles

In order to evaluate the robot’s performance for climbing over obstacles, we let the robot climb over vertical surfaces by
assigning four different tasks, namely, stepping on stairs, going down stairs, and crossing grooves and ridges, respectively.
As shown in Fig. 14(A), when the robot is stepping on stairs and going down stairs, nine levels of different step heights
have been set, and in each level, the robot repeated the same experiment 10 times. Additionally, Fig. 14(B) shows that the
robot performs experiments by crossing grooves and ridges on vertical surfaces. Similarly, different groove widths and ridge
widths were used with the same height, set at 10 levels. At each terrain level, we let the robot conduct the same experiment
10 times. From the analysis of the experimental results in Fig. 14(A), the robot can step on and go down the maximum
height of 15 mm successfully, and the results are consistent with the presented theory in Section 3.1. Fig. 14(b) shows the
performance of crossing grooves and ridges, whereby the crossing groove stability is higher than that for the crossing ridge.
Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210 209

Fig. 15. Snapshots of the robot overcoming a randomly structured surface. (a)–(c) Robot overcoming a flat surface towards the first step which is 30 mm
wide. (d)–(f) Robot overcoming the 90 mm step. (g) Robot overcoming the last step, which is 20 mm wide. See multimedia extension for details.

Moreover, when the ridge widths are 14 mm and 20 mm, the robot elicits lower success rates both for crossing grooves and
ridges. However, the robot has the worst performance when the ridge widths are within the range of 8–12 mm, because the
interface between the robot’s feet and the ridge does not allow the generation of adequate adhesive force. Otherwise, when
the ridge width is less than 8 mm, the robot can cross directly, and when the ridge width is larger than 12 mm, the ridge
area on which the robot establishes contact can provide enough adhesive force for attachment.

4.6. Climbing on a randomly structured surface

In this experiment, different step sizes were set, namely step widths of 30 mm, 90 mm, and 20 mm. Fig. 15 shows the
robot climbing up an unstructured surface from a flat surface by overcoming obstacles of various sizes. Fig. 15(a) shows
the robot climbing up a flat surface, and Fig. 15(b)–(c) depicts the process of the robot overcoming the first step, which
is 30 mm wide. The procedure of overcoming the second step with a height of 90 mm is shown in Fig. 15(d)–(f). Finally,
Fig. 15(g) shows the robot overcome the last step with a width of 20 mm. Throughout the climb, the robot’s outer and
inner frames attach and detach in an alternate manner on the surface by using the parallel four-bar linkage. The symmetric
compliant linkages of either the front or rear footpads deformed from a rectangular to the rhomboid parallelogram geometry
in order to adapt to the uneven surface and improve the robot’s climbing ability.

5. Conclusions

In this study, a new linkage-type, wall-climbing robot, with a capability to overcome obstacles has been designed for
vertical surfaces using dry adhesive. To adapt to obstacle heights, a symmetric compliant linkage mechanism was employed,
and a walking mechanism based on four-bars was used for walking locomotion. Therefore, based on the linkage-type de-
sign, the robot can step on, and move down stairs with a maximum height of 15 mm—which is 30.6% of the robot’s height—
successfully, without the use of any controllers. Additionally, the performance was verified with experiments. The assembled
prototype was based on design parameters from analytical results, and it proved that a minimal and compliant robot design
could successfully perform flat and uneven surface climbing. Even though the objective of this research was accomplished,
there are many problems to be resolved before the robot is launched into a real environment, including its steering, and pos-
sible contamination. Once these problems are solved, the dry-adhesive robots are expected to be used in specific operations,
including space missions.

Acknowledgment

This study was supported by the Basic Science Research Program through the National Research Foundation of Korea
(NRF) funded by the Ministry of Science, ICT, and Future Planning [grant number NRF-2017R1A2B4002123].

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.mechmachtheory.
2018.02.010.
210 Y. Liu, T. Seo / Mechanism and Machine Theory 124 (2018) 197–210

References

[1] F. Tache, W. Fischer, G. Caprari, R. Siegwart, R. Moser, F. Mondada, Magnebike: a magnetic wheeled robot with high mobility for inspecting com-
plex-shaped structures, J.f Field Robot. 26 (5) (2009) 453–476.
[2] G. Lee, J. Woo, J. Kim, T. Seo, High-payload climbing and transitioning by compliant locomotion with magnet adhesion, Robot. Autonom. Syst. 60 (10)
(2012) 1308–1316.
[3] H. Zhu, Y. Guan, W. Wu, L. Zhang, X. Zhou, H. Zhang, Autonomous pose detection and alignment of suction modules of a biped wall-climbing robot,
IEEE-ASME Trans. Mechatron. 20 (2) (2015) 653–662.
[4] G. Lee, H. Kim, K. Seo, J. Kim, M. Sitti, T. Seo, Series of multilinked caterpillar track-type climbing robots, J. Field Robot. 3 (6) (2016) 737–750.
[5] S. Kim, M. Spenko, S. Trujillo, B. Heyneman, D. Santos, M.R. Cutkosky, Smooth vertical surface climbing with directional adhesive, IEEE Trans. Robot.
24 (1) (2008) 65–74.
[6] P. Birkmeyer, A.G. Gillies, R.S. Fearing, Dynamic climbing of near-vertical smooth surfaces, in: Proceedings of the IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS), Algarve, 2012, pp. 286–292.
[7] E.W. Hawkes, D.L Chrsitensen, M.R. Cutkosky, Vertical dry adhesive climbing with 100X bodyweight payload, in: Proceedings of the IEEE International
Conference on Robotics and Automation (ICRA), Seattle, Washington 2015, May 26-30, pp. 3762–3769.
[8] K. Koh, M. Sreekumar, S. Ponnambalam, Hybrid electrostatic and elasromer adhesion mechanism for wall climbing robot, Mechatronics vol. 35 (2016)
122–135.
[9] O. Unver, A. Uneri, A. Aydemir, M. Sitti, Geckobot: a gecko inspired climbing robot using elastomer adhesives, in: Proceedings of the IEEE International
Conference on Robotics and Automation, Orlando, Florida, 2006, pp. 2329–2335.
[10] C. Menon, M. Sitti, Biologically inspired adhesion based surface climbing robot, in: Proceedings of the IEEE International Conference on Robotics and
Automation, Barcelona, Spain, 2005, pp. 2715–2720.
[11] M. Henrey, A. Ahmed, P. Boscariol, L. Shannon, C. Menon, Abigaille-III: a versatile, bioinspired hexapod for scaling smooth vertical surface, J. Bionic
Eng. 11 (1) (2014) 1–17.
[12] M.P. Murphy, M. Sitti, Waalbot: an agile small-scale wall-climbing robot utilizing dry elastomer adhesives, IEEE/ASME Trans. Mechatron. 12 (3) (2007)
330–338.
[13] K.A. Daltorio, T.E. Wei, A.D. Horchler, L. Southard, G.D. Wile, R.D. Quinn, S.N. Gorb, R.E. Ritzmann, Mini-Whegs TM clmbs steep surfaces using insec-
t-inspired attachment mechanisms, Int. J. Robot. Res. 28 (2) (2009) 285–302.
[14] W.A. Breckwoldt, K.A. Daltorio, L. Heepe, A.D. Horchler, S.N. Gorb, R.D. Quinn, Walking inverted on ceiling with wheel-legs and micro-structured
adhesives, in: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Hamburg, 2015, pp. 3308–3313.
[15] O. Unver, M. Sitti, Tankbot: a palm-size, tank-like climbing robot using soft elastomer adhesive treads, Int. J. Robot. Res. 29 (14) (2010) 1761–1777.
[16] T. Seo, M. Sitti, Tank-like module-based climbing robot using passive compliant joints, IEEE/ASME Trans. Mechatron. 18 (1) (2013) 397–408.
[17] M. Greuter, G. Shah, G. Caprari, F. Tache, R. Siegwart, M. Sitti, Toward micro wall-climbing robots using biomimetic fibrillary adhesives, in: Proceedings
of the International Symposium on Autonomous Minirobots for Research and Education (AMiRE), 2005, pp. 39–46.
[18] J. Krahn, Y. Liu, A. Sadeghi, C. Menon, A tailless timing belt climbing platform utilizing dry adhesives with mushroom caps, Smart Mater. Struct. 20
(11) (2011) 1–11.
[19] G. Lee, H. Kim, K. Seo, J Kim, H.S. Kim, MultiTrack: A multi-linked track robot with suction adhesion for climbing and transition, Robot. Autonom. Syst.
72 (2015) 207–216.
[20] Y. Liu, H. Kim, T. Seo, AnyClimb: a new wall-climbing robotic platform on various curvatures, IEEE-ASME Trans. Mechatron. 21 (4) (2016) 1812–1821.
[21] O. Unver, M. Sitti, A miniature ceiling walking robot with flat tacky elastomeric footpads, in: Proceedings of the IEEE International Conference on
Robotics and Automation, Kobe, Japan, 2009, pp. 2276–2281.
[22] O. Unver, M. Sitti, Flat dry elastomer adhesives as attachment materials for climbing robots, IEEE Trans. Robot. 26 (1) (2010) 131–141.
[23] D. Zarrouk, R.S. Fearing, Controlled in-plane locomotion of a hexapod using a single actuator, IEEE Trans. Robot. 31 (1) (2015) 157–167.
[24] J. Choi, K. Jeong, T. Seo, Pol-E: large-obstacle overcoming by energy conversion method using an elastic link, Int. J. Control Autom. Syst. 15 (4) (2017)
1835–1843.
[25] N. Morozovsky, T. Bewley, Stair climbing via successive perching, IEEE/ASME Trans. Mechatron. 20 (6) (2015) 2973–2983.
[26] M. Osswald, F. Iida, Design and control of a climbing robot based on hot melt adhesion, Robot. Autonom. Syst. 61 (issue 6) (2013) 616–625.
[27] B.L. Luk, A.A. Collie, J. Billingsley, ROBUG II: an intelligent wall climbing robot, in: Proceedings of the International Conference on Robotics and Au-
tomation, Sacramento, California, April 1991, pp. 2342–2347.
[28] K. Jendal, Thin-film peeling-the elastic term, J. Phys. D Appl. Phys. 8 (13) (1975) 1449–1452.

You might also like