Soft Flexible Gripper Design, Characterization and Application

Abstract

This paper presents a design concept of soft flexible gripper dedicated for delicate objects manipulation. Traditional grippers are composed of rigid components and consist of finite number of discrete joints. The more joints they have the better they can adapt for the specific object. However, manipulating with fragile objects still requires precise control and some kind of measurement as well. In this paper the Authors propose soft flexible gripper that is able to adapt for the manipulated object shape without any additional computational effort nor any sensors. The gripper is made only of flexible materials such as rubber and silicone. Since the gripper lacks of any discrete joints and is actuated through smooth deformation of its body, it can take very complex shapes and thus easily adapt to the surface of grasped object. The mechanism is based on soft pneumatic actuators developed by the authors and its configuration can be easily redesigned in order to extend its application for special purposes. In this paper the design and experimental characterization of the gripper prototypes is presented. Possible applications are discussed.

Full text

Soft flexible gripper design, characterization and
application
Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
Industrial Research Institute for Automation and Measurements
al.Jerozolimskie 202, 02-486 Warsaw, Poland
jfras, mmacias, fczubaczynski, psalek, jglowka@piap.pl
http://www.piap.pl
Abstract. This paper presents a design concept of soft flexible grip-
per dedicated for delicate objects manipulation. Traditional grippers are
composed of rigid components and consist of finite number of discrete
joints. The more joints they have the better they can adapt for the spe-
cific object. However, manipulating with fragile objects still requires pre-
cise control and some kind of measurement as well. In this paper the
Authors propose soft flexible gripper that is able to adapt for the ma-
nipulated object shape without any additional computational effort nor
any sensors. The gripper is made only of flexible materials such as rubber
and silicone. Since the gripper lacks of any discrete joints and is actuated
through smooth deformation of its body, it can take very complex shapes
and thus easily adapt to the surface of grasped object. The mechanism
is based on soft pneumatic actuators developed by the authors and its
configuration can be easily redesigned in order to extend its application
for special purposes. In this paper the design and experimental charac-
terization of the gripper prototypes is presented. Possible applications
are discussed.
Keywords: robotics, soft manipulator, gripper, grasping, artificial mus-
cle
1 Introduction
One of the most fundamental challenges in robotics is object grasping. Conven-
tional grippers and graspers are expensive and sometimes unsuitable for tasks
where the manipulated object weight, its shape or the environment are uncer-
tain. Moreover, manipulating delicate objects with a traditional rigid gripper
requires sophisticated sensing and high precision that might not always be eas-
ily provided. Rigid grippers designed to handle fragile objects are often complex
or can handle only specified type of items. One of the possible solutions is re-
placement of the gripper by a soft device that can passively adapt to the object
and to the environment. For that reason, soft robotics has been extensively in-
vestigated by scientists, and many different kinds of soft mechanical structures
has been proposed [1][2][3]. Soft robots offer high flexibility, adapt to the exter-
nal conditions and interact more safely with human than any rigid machine. In

2 Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
this paper we present a soft flexible gripper that is able to manipulate with frag-
ile objects and adapt to unstructured shapes. In the following sections design,
manufacturing process and the gripper capabilities are presented. Then, possible
applications are discussed. In general two issues are discussed when character-
izing robotic grippers: ability to fixture object (often called grasping), and task
of manipulating object with fingers [4]. In this paper we will focus on grasping
only.
2 Design of the gripper
2.1 Finger construction
A crucial element of the gripper are fingers. The fingers are derived from ar-
tificial muscles technology developed first in late 1950 [5][6], principle of their
operation is, however, significantly different from the initial muscles technology.
An actuator based on the similar principle our fingers are, was first described in
1991 and called flexible micro-actuator [2]. The most important difference be-
tween artificial muscle and the actuator described in this paper is that the force
exerted by actuator and the one generated by muscle have opposite directions.
In our design each finger is a cylinder made of silicone with a pressure actuated
chamber aligned with its central axis, Fig. 1. Two kinds of silicone are used -
EcoFlex 0050 for softer and SmoothSil 950 for stiffer parts. The actuation cham-
ber (c) is made of the softer material and reinforced with tight helix (e) (made
of polyester thread) in order to limit radial expansion of the finger [7]. The inner
fingers’ side (f) is made of the stiffer silicone and formed in a bellow shape (g)
in order to increase grasping capabilities. For the outer side of the finger (a) the
softer silicone is used, so that the pressure application results in bending (j) as
the inner part of the finger elongates less than the outer one. The tip (d) and the
bottom (h) of the finger are sealed with the stiffer silicone. An actuation fluid is
provided via small rubber pipes connected to the actuation chamber at the base
of each finger.
a
b
c
d
i
g
e
f
h j
Fig. 1: Single finger design. a - outer silicone layer, b - actuation chamber wall, c
- actuation chamber, d - finger’s tip, e - reinforcement, f - inner layer, g - bellow
shaped surface, h - finger’s base, i - pressure cable, j - actuated finger. Light and
dark grey colour corresponds with soft and stiff silicone, respectively.

Soft flexible gripper design, characterization and application 3
2.2 Gripper construction
The gripper consists of a number of fingers. In particular, every single finger can
be considered as a gripper as it can coil around the object and grasp it. The
fingers can be arranged symmetrically or asymmetrically, actuated in groups or
working independently each other. Fingers are bound to the gripper body with
the same silicon type that is used for the inner finger side. In this paper we
present few patterns of the fingers arrangements, however, due to simple design
the setup can be very easily rearranged and adapted for a specific application.
Exemplary arrangements are presented in Fig. 2.
(a) (b) (c)
Fig. 2: Example grippers with different fingers arrangements: (a) with axial and
(b) linear symmetry, (c) human hand inspired.
3 Manufacturing
The manufacturing process consists of several steps. The fingers are prepared
first. For each finger special cylindrical chamber rod with removable core is pre-
pared. Next, the rods are tightly braided using a low-diameter polyester thread,
Fig. 3a. Then, the chamber reinforcement prepared in this way is covered with
silicone layer separately - the soft outer part and the inner harder one. After
the silicone cures, the rods are removed by first removing their cores, Fig 3b.
Removing whole rod at once would cause high friction between the rod and the
finger structure and result in finger damage. Next the actuation chamber wall is
created by pouring the soft silicone inside the finger reinforcement and inserting
dedicated rod of smaller diameter than the previous one, Fig. 3c. Last step of
finger manufacturing is sealing its tip with the hard silicone. Once the fingers
are ready, they are bound together in the final mould that forms the gripper
body, Fig. 3d.

4 Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
(a) (b) (c) (d)
Fig. 3: Manufacturing steps: (a) chamber reinforcement preparation, (b) rod re-
moval (core goes first), (c) inner chamber wall moulding, (d) final moulding.
4 Characterization
The described gripper has been tested in terms of geometry and generated forces,
partially covering possible characterization [8][9]. The force generated by the sin-
gle finger depending on its length and the overall gripper capabilities were exam-
ined. The bending angle of the finger as a function of pressure was determined
and compared with proposed mathematical model.
4.1 Single finger bending
Mathematical description The pressure applied into the actuation chamber
results in a force acting in a cross-section of the finger. Considering the cross-
section perpendicular to the finger neutral axis, the force is perpendicular to the
cross-section plane. Its value is proportional to the pressure and the chamber
cross-section area. As the Young modulus of the materials used in the finger
construction differs at the cross-section, the bending neutral axis is shifted to-
wards the inner finger part, Fig. 4.
Since the activation camber is aligned with the finger geometrical centre axis,
the pressure applied results in the force that is not aligned with the bending neu-
tral axis. That leads to a bending moment in the cross-section. The cross-section
geometry does not change along the finger and due to the Pascal’s law, force act-
ing in all the cross-sections has the same value. Thus the bending moment along
the finger is constant. The bending moment around neutral axis resulting from
pressure can be expressed as (1). The Pstates for pressure value, dcorresponds
with the distance of the neutral bending axis from cross-section’s geometrical
centre that is a point of internal force acting, Astates for the actuation chamber
cross-section area and equals π(r3)2.
M=P dA (1)
The neutral axis position can be obtained from the assumption that tensions
for pure bending compensates on both sides of that axis. In such case the tension
at the point (x, y)is expressed by (2),

Soft flexible gripper design, characterization and application 5
r1
r3
r2
Bending neutral axis
d
Geometrical center axis
x
y
Fig. 4: Finger cross-section. Light grey and dark grey colours correspond with
soft and stiff silicone, respectively.
σ(x, y) = E(x, y)x−d
ρ(2)
where E(x, y)states for the Young’s modulus of the material used at (x, y)
and ρstates for the curvature radius of the neutral plane at the cross-section .
Hence, the neutral axis position dcan be obtained from equilibrium condition
(3) [10].
ZZ
xy
σ(x, y)dxdy = 0 (3)
Since the ρvalue is a constant the equation can be rewritten as (4).
ZZ
xy
E(x, y)(x−d)dxdy = 0 (4)
Then the curvature radius ρfor a certain bending moment Mcan be derived
from (5),
ρM=PnEnIn
M(5)
where Enand Instates for Young modulus and moment of inertia in respect
to neutral bending axis of the cross-section nthcomponent, respectively. Once
the position of the neutral bending axis is determined the moments of inertia
can be obtained using parallel axis theorem.

6 Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
The pressure applied into the actuation chamber causes not only bending,
but elongation as well. The change of length can be derived from Hooke’s law
(6),
∆l =l0F
PnEnAn
(6)
where l0represents rest finger length, Enand Anstates for the area and
Young modulus of the relevant cross-section component, respectively. Fis stretch-
ing force that equals to pressure value multiplied by an activation chamber cross-
section area. Hence, the overall bending angle can be expressed by the (7).
α=l0+∆l
ρM
(7)
Thus the bending angle for a specific finger length is a quadratic function of
the pressure.
Experiment The finger bending angle as a function of pressure has been mea-
sured. Due to limited data regarding material’s Young modulus and not very
precise manufacturing process (3D printed moulds, manual moulding, etc.) ob-
taining parameter values to compare the results with theoretical model is not
easy. However, the second order polynomial fit presents good approximation of
the gathered data, thus the principle of the model seems to be in force. The
length of an active finger part was 20mm and the pressure range was 0-0.6 bars.
The empirical data is presented in Fig. 5.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
45 mm
35 mm
25 mm
15 mm
Activation pressure [bar]
Generated force [N]
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
0
50
100
150
200
250
Pressure [bar]
Bending angle [deg]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
20
40
60
80
100
120
140
160
180
200
Experimental data
Polyniminal fit
Pressure [bar]
Bending angle [deg]
Fig. 5: Bending as a function of pressure.
4.2 Force generated by a single finger
The finger capabilities in terms of generated force values has been examined. The
finger was mounted horizontally and the force has been measured as a function of

Soft flexible gripper design, characterization and application 7
pressure. The point at which the force has been measured was constant in space.
For the measurement precise scale was used. The weight exceeded the finger
capabilities, thus its position on the scales was stable during the experiment.
The measurement setup and the results of the experiment are presented in Fig.
6
(a)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
45 mm
35 mm
25 mm
15 mm
Activation pressure [bar]
Generated force [N]
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
0
50
100
150
200
250
Pressure [bar]
Bending angle [deg]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
45 mm
35 mm
25 mm
15 mm
Activation pressure [bar]
Generated force [N]
(b)
Fig. 6: Finger capabilities examination: (a) experimental setup, (b) force gener-
ated by fingers measured for different length fingers.
4.3 Grasping capabilities
Single finger gripper would have very limited application capabilities. In this
section the performance of a grasper composed of five fingers is presented. Two
manipulated objects of different kinds were tested. The force of the grasp was
obtained for each object for various actuation pressures. The experiment was
performed by grasping the object of known weight with the maximal available
pressure and reducing the pressure until the object was dropped. Due to high
variation of the results the experiment was repeated 6 times for each weight
value. The tested use case is presented in Fig 8a. The performance for irregular
soft object and for spherical rigid one is presented in Fig. 8b
5 Usage scenario discussion
There are many potential applications of flexible graspers in various scenarios.
One of the examples is dealing with soft, flexible or delicate materials, in indus-
tries like apparel or shoe manufacturing [11]. Soft robots can enable us to make
exoskeletons or wearable protetics, and provide more unobtrusive way to inter-
face with human body [12]. Another intensively investigated application in recent
years is friut picking that require caution due to objects fragility [13][14]. Higly
demanding potential usage scenario is forensic evidences collection [15]. Since

8 Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
(a)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
45 mm
35 mm
25 mm
15 mm
Activation pressure [bar]
Generated force [N]
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
Generated force
Linear fit
Activation pressu re [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressu re [bar]
Force [N]
0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
Generated force
Linear fit
Activation pressure [bar]
Force [N]
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
0
50
100
150
200
250
Pressure [bar]
Bendin g angle [deg]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
20
40
60
80
100
120
140
160
180
200
Finger 25mm
long
Pressure [bar]
Bendin g angle [deg]
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2Irregular flexible object
Linear fit (irregular object)
Regular spherical rigid object
Linear fit (regular object)
Activation pressu re [bar]
Grasping force [N]
(b)
Fig. 7: Grasper capabilities examination: (a) experimental setup, (b) grasping
force comparison.
the evidences may be fragile and of irregular or unexpected shapes they may
be difficult to grasp using conventional gripper. Experience of PIAP’s experts
gained from working in this field in cooperation with crime scene investigators
concludes that such a soft grippers may be very useful for remote physical crime
evidence gathering. One of the PIAP’s robots equipped with the soft gripper
is presented in Fig. 8. Generally handling fragile objects (fruits, human body,
forensic evidences), in very different situations and usages can become domain
dominated by soft robots [16], especially soft grippers such as the one described
in this paper.
In the Fig. 9 few example handled objects are presented.
(a) (b)
Fig. 8: Mobile security robot equipped with the soft gripper: (a) the robot, (b)
the device designed to actuate the soft gripper and to be carried by robot.

Soft flexible gripper design, characterization and application 9
Fig. 9: Examples of grasped objects.
6 Summary
Traditional grippers are very widely utilized devices. However in some cases,
the usage of rigid tools is problematic due to their stiffness. Soft and flexible
graspers and grippers demonstrate great potential in such cases. Their adapta-
tion capabilities and soft contact make them safe and adequate for fragile object
manipulation. Due to their safety properties, soft grippers and grasper are ex-
tensively considered as a substitution for rigid devices in the human-present
environments. Moreover, the passive adapting skills make it possible to use soft
devices in uncertain environments and for unstructured object manipulation
without any additional control effort. In many cases there is no need to use any
sensing to realise a task, since the gripper flexibility successfully compensates all
the object irregularities.
References
1. R. Maruyama, T. Watanabe, and M. Uchida. Delicate grasping by robotic gripper
with incompressible fluid-based deformable fingertips. In 2013 IEEE/RSJ Interna-
tional Conference on Intelligent Robots and Systems, pages 5469–5474, Nov 2013.
2. Koichi Suzumori, Shoichi Iikura, and Hirohisa Tanaka. Development of flexible mi-
croactuator and its applications to robotic mechanisms. In Robotics and Automa-
tion, 1991. Proceedings., 1991 IEEE International Conference on, pages 1622–1627.
IEEE, 1991.
3. T. Hassan, M. Manti, G. Passetti, N. d’Elia, M. Cianchetti, and C. Laschi. Design
and development of a bio-inspired, under-actuated soft gripper. In 2015 37th An-
nual International Conference of the IEEE Engineering in Medicine and Biology
Society (EMBC), pages 3619–3622, Aug 2015.
4. Mark R Cutkosky. Human grasp choice and robotic grasp analysis. Dextrous robot
hands, 1:5–31.

10 Fraś J., Maciaś M., Czubaczyński F., Sałek P., and Główka J.
5. R.H. Gaylord. Fluid actuated motor system and stroking device, July 22 1958. US
Patent 2,844,126.
6. Glenn K Klute, Joseph M Czerniecki, and Blake Hannaford. Artificial muscles:
Actuators for biorobotic systems. The International Journal of Robotics Research,
21(4):295–309, 2002.
7. J Fraś, J Czarnowski, M Maciaś, J Główka, M Cianchetti, and A Menciassi. New
stiff-flop module construction idea for improved actuation and sensing. In Robotics
and Automation (ICRA), 2015 IEEE International Conference on, pages 2901–
2906. IEEE, 2015.
8. Joe Falco, Jeremy Marvel, and Elena Messina. A roadmap to progress measurement
science in robot dexterity and manipulation. US Dept. Commer., Nat. Inst. Std.
Technol., Gaithersburg, MD, USA. Tech. Rep. NISTIR, 7993, 2014.
9. Glenn K Klute, Joseph M Czerniecki, and Blake Hannaford. Mckibben artificial
muscles: pneumatic actuators with biomechanical intelligence. In Advanced Intelli-
gent Mechatronics, 1999. Proceedings. 1999 IEEE/ASME International Conference
on, pages 221–226. IEEE, 1999.
10. Zdzisław Dyląg, Antoni Stefan Jakubowicz, Zbigniew Orłoś, and Wydawnictwa
Naukowo-Techniczne. Wytrzymałość materiałów. Wydawnictwa Naukowo-
Techniczne, 2013.
11. Robotics 2020 multi-annual roadmap for robotics in europe call 2 ict24 (2015) –
horizon 2020. 2nd ed. Report, euRobotics, 2015.
12. Robert J. Wood and Conor J Walsh. Smaller, softer, safer, smarter robots. Science
Translational Medicine, 5(210):210ed19, 2013.
13. Shaoming Qian, L. Zhang, Q. Yang, G. Bao, Zhiheng Wang, and Liyong Qi. Re-
search on adaptive multi-contract grasping model of flexible pneumatic finger. In
2009 International Conference on Mechatronics and Automation, pages 2823–2828,
Aug 2009.
14. G. Bao, P. Yao, S. Cai, S. Ying, and Q. Yang. Flexible pneumatic end-effector for
agricultural robot: Design amp; experiment. In 2015 IEEE International Confer-
ence on Robotics and Biomimetics (ROBIO), pages 2175–2180, Dec 2015.
15. Grzegorz Kowalski, Mateusz Maciaś, and Adam Wołoszczuk. Recent Advances in
Automation, Robotics and Measuring Techniques, chapter Selected Issues of Col-
lecting Forensic Evidence with a Mobile Robot, pages 431–440. Springer Interna-
tional Publishing, Cham, 2014.
16. Filip Ilievski, Aaron D Mazzeo, Robert F Shepherd, Xin Chen, and George M
Whitesides. Soft robotics for chemists. Angewandte Chemie International Edition,
50(8):1890–1895, 2011.