IEEE ROBOTICS AND AUTOMATION LETTERS, VOL. 4, NO.

1, JANUARY 2019 65

A Hybrid Gripper With Soft Material and

Rigid Structures
Wookeun Park , Seongmin Seo , and Joonbum Bae , Member, IEEE

Abstract—Various robotic grippers have been developed over SPAs with a variety of forms and operating pressures, in-
the past several decades for robotic manipulators. Especially, the cluding granular jamming, and fiber-reinforced SPAs have been
soft grippers based on the soft pneumatic actuator (SPA) have been suggested; however, their practical application as a robotic grip-
studied actively, since it offers more pliable bending motion, inher-
ent compliance, and a simple morphological structure. However, per has been limited by a small fingertip force and slow ac-
few studies have focused on simultaneously improving the fingertip tuation speed. A large pneumatic compressor can be used to
force and actuation speed within the specified design parameters. address these issues, but the size of the whole system would be
In this study, we developed a hybrid gripper that incorporates increased [2], [6]. The granular jamming gripper, for example,
both soft and rigid components to improve the fingertip force and uses granular components, which are jammed in the soft cham-
actuation speed simultaneously based on three design principles:
first, the degree of bending is proportional to the ratio of the rigid ber under vacuum pressure [7]. As a large vacuum pump is
structure; second, a concave chamber design is preferred for large required for granular jamming, the pneumatic system required
longitudinal strain; and third, a round shape between soft and would inevitably be bulky. General SPAs with operating pres-
rigid materials increases the fingertip force. The suggested princi- sures in the range of 0–80 kPa have a slow actuation speed and
ples were verified using the finite element methods. The improved low fingertip forces [5], [6] have difficulties with the limited
performance of the hybrid gripper was verified experimentally and
compared with the performance of a conventional SPAs. The abil- actuation speed and fingertip forces. To increase the actuation
ity of the hybrid gripper to grasp different objects was evaluated speed, a fast-pneumatic network (fast-PneuNet) comprising a
and was applied in a teleoperated system. series of soft chambers and channels has been suggested [2].
Index Terms—Soft material robotics, grippers and other end-
However, the resulting fingertip force is low (<1.5 N) [2]–[4].
effectors, flexible robots. Improved fingertip force actuation has been achieved with fiber
or fabric reinforcement [8]–[10]. Since a large pneumatic pump
I. INTRODUCTION with pressures of 350 kPa is needed, the entire system is in-
evitably bulky. Also, the actuation speed is relatively slow as
ECENTLY, the limitations of rigid gripper due to the ne-
R cessity of additional linkages, complicated controls, and
so on, have been overcome by the soft gripper [1]. Especially, the
large volume changes are required for the bending. Above all,
few studies have attempted to enhance the actuation speed and
fingertip force simultaneously.
soft grippers based on soft pneumatic actuators (SPAs), which
Integrating different actuating mechanisms [11]–[13] or ma-
offer better compliance, and a higher degree of freedom than
terials have also been tried as hybrid actuators [14], [15]. How-
the rigid robot, have been actively developed [2]–[9]. Due to
ever, we are not focused on the integrating different actuating
the inherent compliance of soft materials, complicated controls
source, but incorporating the soft material and rigid structures.
and additional structures are not needed with soft grippers, as
Rigid materials were usually used to increase the stiffness of
the proper selection of materials of various stiffness allows
the soft structures [14]. For example, rigid structures were used
for simple control. Furthermore, soft grippers with SPAs are
to reinforce the soft chambers, inspired by the lobster [15].
constructed from low-cost and readily-available elastomers.
However, the degrees of freedom may be limited depending on
Manuscript received June 27, 2018; accepted October 14, 2018. Date of the rigid structures, which may reduce the compliance as the
publication October 31, 2018; date of current version November 15, 2018. This main advantages of SPAs, and may not focus on enhancing the
letter was recommended for publication by Associate Editor C. Laschi and actuation speed and fingertip force simultaneously.
Editor Y. Sun upon evaluation of the reviewers’ comments. This work was
supported by the National Research Foundation of Korea (NRF) under Grant Thus, in this letter, we developed a hybrid gripper that in-
NRF-2016R1A5A1938472 funded by the Korean Government (MSIP), in part corporates both soft material and rigid parts to overcome the
by the Institute for Information and Communications Technology Promotion drawbacks of the conventional SPAs without reducing the com-
(IITP) under Grant 2017000910001100 (Development of multi-material 3D
printing technologies for flexible motion detection and control sensor modules) pliances (Fig. 1). The rigid structures are inserted into a fast
funded by the Korea government (MSIP), and in part by the Basic Science PneuNet, hereafter referred to as the hybrid PneuNet, assem-
Research Program through the National Research Foundation of Korea (NRF) bled as the grippers, which have improved fingertip force, and
under Grant NRF-2018R1A1A1A05077442 funded by the Ministry of Science
and ICT. (Corresponding author: Joonbum Bae.) increased actuation speed. The main contributions of this letter
The authors are with the Bio-Robotics and Control (BiRC) Laboratory, De- are as follows: 1) design methodology for a hybrid PneuNet,
partment of Mechanical Engineering, Ulsan National Institute of Science and 2) experimental verification of the hybrid gripper, 3) a practi-
Technology, Ulsan 689-805, South Korea (e-mail:,[email protected]; tjtjdals
[email protected]; [email protected]). cal application as a simple and compact pneumatic system for
Digital Object Identifier 10.1109/LRA.2018.2878972 teleoperations.

2377-3766 © 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution
requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
66 IEEE ROBOTICS AND AUTOMATION LETTERS, VOL. 4, NO. 1, JANUARY 2019

Fig. 1. The proposed hybrid gripper with soft and rigid structures.

II. DESIGN PRINCIPLES FOR THE HYBRID GRIPPER

The bending motion of a SPA is controlled by differential
strain effects between the inextensible layers and soft cham-
bers, as shown in Fig. 2(a). The nature of this motion is manipu-
lated by the geometries and material properties of the chambers.
The governing equations for bending motion of the SPA are as
follows:
L = (1 + εcham )L 0
L (1 + εcham )L 0
θspa = = (1)
+R h h + R
where L 0 and L is the initial and longitudinally elongated length
of the actuator, εcham is the longitudinal strain of the chamber,
θspa is the bending angle of the SPAs, and h  is the height where
maximum longitudinal strain occurs. R is the radius of the bent
SPAs, which decreases during the bending motion.
According to (1), the bending angle, θspa is proportional to
the the longitudinal strain of the chamber. Thus, the longitudinal
strain of the chamber, which induces actuation of longitudinally
elongated SPAs, should be increased for improved bending mo- Fig. 2. Working principles for the bending motion of (a) the general SPAs,
tion. The hybrid PneuNet, which has a rigid structure between (b) the proposed hybrid PneuNet (c) fingertip force of the hybrid PneuNet.
soft chambers, increases the elongated length of the SPAs. Fur-
thermore, the force transmission is improved since the damp- height between the rigid and soft materials, respectively. θh is
ing ratio is decreased with placing the rigid structures, which the bending angle of the hybrid PneuNet. The contact height
increases the force transmissibility [16]. h  , was assumed to be constant during inflation. From (2), the
The working principle for bending motion of hybrid Pne- bending motion is mainly induced by the longitudinal strain of
uNet were shown in Fig. 2(b). The basic concept of the bend- the soft chamber and the length of the rigid structures.
ing motion is similar to the previously developed SPAs, but The fingertip force of the hybrid PneuNet, which is exerted
the existence of rigid structures should be considered. The de- normal to the ground, is based on the simplified model shown
sign principles were derived from the concepts in Fig. 2(b) and in Fig. 2. Given that the soft chambers and rigid structures are
the governing equation of the hybrid PneuNet is expressed as positioned in an alternating sequence along the actuator length,
follows: and that the length of the strain limiting layer remains constant,
the simplified linkage forms a rotating joint. The fingertip force,
L = n rigid L rigid + n cham (L cham + L cham )
normal to the ground, is the sum of the bending moments of the
L n rigid L rigid + n cham (L cham + L cham ) individual linkages; the governing equation is as follows:
θh = = (2) n
h + R h + R Mi n
Mi
where n rigid and n cham are the numbers of specific rigid structure Ftip = i=1 = (3)
L0 l
i=1 i
cos θi
and the chamber, respectively, and L rigid represents the length
of the rigid structure. L cham and L cham represent the initial where n is the number of simplified linkage with the soft cham-
and longitudinally elongation of the chamber at the contact ber and rigid structure, Mi is the bending moment of the ith
PARK et al.: HYBRID GRIPPER WITH SOFT MATERIAL AND RIGID STRUCTURES 67

linkage, and L 0 is the initial longitudinal length of the hybrid

PneuNet. Fi is the force exerted on the ith linkage for the bend-
ing moment. The resultant force at the tip position (Ftotal ), which
can be decomposed into the fingertip force Fti p and the friction
force at the tip position F f , in equilibrium with the friction on
the ground. During inflation, the length of the moment arm, ex-
pressed in terms of the angle (θi ) and length for each linkage (li )
depends on the postures of the hybrid PneuNet. It was assumed
that the length of the moment arm does not change. From the
(3), a stronger fingertip force requires a larger total bending
moment of each linkage (Mi ).
Based on the analysis results described above, we focused
on the following design principles for the enhancements of the
hybrid PneuNet: 1) increasing the rigid structure length via rigid
structure ratio, 2) increasing the longitudinal strain of the soft
chamber by manipulating the chamber shape, and 3) applying
a round edge design between the soft and rigid materials to
enhance fingertip force. Finite element method (FEM) simula-
tions based on the hyperelastic Mooney-Rivlin model [17] were
carried out to verify these design factors under various linear
pressure loading conditions.

A. Rigid Structure’s Ratio

The first design principle is based on manipulating the length Fig. 3. Bending angle comparison based on the simulation with different rigid
of the rigid structures according to (2). The improved longitu- structure ratios (γrigid = 0.06, γrigid = 0.31, γrigid = 0.61).
dinal elongation of the hybrid PneuNet with increased length
of the rigid structures leads to a large bending motion. To dis- FEM simulations were conducted to verify the first design
cuss the length of the rigid structures quantitatively, the ratio of principle and the simulation was performed for a chamber with
the rigid structures, γrigid , with respect to the soft chambers is a plain shape. In Fig. 3, the bending angles of the hybrid PneuNet
defined as follows: were compared with the simple plain shape; up to 50 kPa, γrigid
n rigid L rigid L 0 − n cham L cham was set as 0.06, 0.31, and 0.61, respectively. The bending angle
γrigid = = (4) increased with γrigid in agreement with the expectation that the
L0 L0
hybrid PneuNet with a larger ratio should have more bending
where γrigid is constrained from the longitudinal length of the motion.
chambers at the contact height, L cham,i , which is limited by the
fabrication technique. Generally, the soft chambers, fabricated B. Chamber Shape
with three-dimensional (3D) printed molds, have at least a 2 mm
wall thickness to provide durability of the molds and soft cham- The cross-sectional shape of the chamber is another key
bers. As such, the minimum length of the soft chamber and design factor to enhance the bending angle under the same
the rigid structure were set as 6 mm and 1 mm, respectively. pressure. As shown in (2), manipulating the elongation of the
The size (138 × 23 × 21 mm3 ), and number of soft chambers soft chamber, which is also expressed as the longitudinal strain
(ncham = 9) of the hybrid PneuNet were determined considering (εcham ), increases the bending angle with increased L. The longi-
the conventional SPAs of the soft gripper [2]. Thus, the max- tudinal strain of the material is proportional to the strain energy
imum and minimum rigid structure ratio were determined as of the chamber [18]. The strain energy is affected by the energy
0.06 and 0.61, respectively. dissipation of the chamber, specifically the chamber’s shape. In
By rearranging (4) and adding the ratio of the rigid structures, a comparison of various chamber designs, three types of shape,
γrigid , the longitudinally elongated length of the actuator can be plain, convex, and concave, were evaluated in terms of resulting
expressed as follows: longitudinal strain. The cross-sectional shape of the chambers
are illustrated in Fig. 4(a).
γrigid L 0 + n cham (1 + εcham )L cham
θh = (5) The tension of the soft chamber around the center, describing
h + R the elastic behavior, is depicted by the red arrows in Fig. 4(a).
Note that the bending angle is increased by γrigid and L cham is The resultant vector at the center of the height, Tsoft (red arrows
inversely proportional to γrigid . However, the longitudinal strain in Fig. 4(a)) acts against chamber inflation. The applied pressure
is increased by γrigid under the boundary condition, since the (Pload ), depicted as black arrows in Fig. 4(a), exerts normal to
stiffness at top of the chamber is increased depending on the the chamber’s inner wall. The resultant vectors of Tsoft and the
reduced L cham . Consequently, the strain of the chamber occurs Pload , depicted as blue arrows, cause accumulation difference
more longitudinally, not in a radial direction. of the strain energy of the soft chamber at the mid-height. For
68 IEEE ROBOTICS AND AUTOMATION LETTERS, VOL. 4, NO. 1, JANUARY 2019

the plain and convex chambers up to 30 kPa. The concave shape

also had the largest longitudinal elongation, εcham , as shown in
Fig. 4(c), which means the concave-shaped chamber offered the
largest bending angle under the same loading pressure.

C. Round Edge
The third design principle, i.e., the use of a round edge be-
tween soft and rigid materials of the hybrid PneuNet (Fig. 5(a)),
was applied to increase the fingertip force. Assuming a cham-
ber with a plain shape, a round-edge between the soft chambers
increases the contact area and guides the longitudinal elonga-
tion between the chambers and rigid parts with enhanced force
transmission (Fig. 5(a)). In a simplified model, in which the
interaction between soft and rigid structures is considered as a
simple linkage (Fig. 2(c)), the bending moment can be expressed
as follows:
Mi = li sin θi Fi,x + li cos θi Fi,y
 
= li sin θi σi,x dAyz + li cos θi σi,y dAxz (6)

where σi and d A are the stress, and contact area between the
soft and rigid structure, respectively. σi,x and σi,y represent the
decomposed stresses from σi in x and y direction. Fi is decom-
posed into x and y directional forces as shown in Fig. 2(c). d A is
also decomposed into the contact area projected onto the yz and
x z planes. The decoupled force components can be expressed
as the integral forms of the decoupled stresses and contact ar-
eas. Thus, the fingertip force can be manipulated by varying the
bending moment between soft and rigid materials.
The proposed round edge design is shown in Fig. 5(a)–(c).
Assuming that the entire system volume remains the same and
that the rigid structure does not protrude from the soft chambers,
the bending moment of each linkage increases with the round
edge configuration. The additional bending moment exerted by
the round edges between the soft chambers and rigid structures
is given by the following:

Mi,top = σi,y,e dAxz

n
Mi,RE n
Mi + Mi,top
Fig. 4. Longitudinal strain comparison based on the simulation of different Ftip.RE = = (7)
chamber shapes: (a) three different types of chamber shapes, (b) strain energy i=1
l i l cos θi i=1
li cos θi
comparison for the different chambers at the mid-height, (c) longitudinal strain
comparison at the mid-height. where Mi,top is the additional bending moment of each linkage
with the round edge design. Ftip,RE and Mi,RE represent the fin-
example, since the resultant vectors are headed to the mid- gertip force and ith bending moment for each linkage of the
height of the chamber, the more strain energy is accumulated hybrid PneuNet with round edge design. σi,y,e and σi,z,e decom-
for the concave chamber, compared with the convex chamber. posed from σi , in y and z direction, represent the additional
On the other hand, the less strain energy is accumulated with stresses, exerted at round edges (Fig. 5(b) and (c)). From (7),
the mid-height of convex chamber, since the resultant vector σi,y,e creates an additional bending moment Mi,top , which in-
left out of the mid-height. Thus, the differentiated accumulated creases the fingertip force with the round edge (Ftip,RE ). Given
strain energy due to the chamber shape induces the different that the z directional stress at the edge, σi,z,e , is applied sym-
longitudinal strain and the concave chamber would have more metrically in Fig. 5(c), it does not affect the additional bending
longitudinal strain. moment term but instead guides the longitudinal expansion of
For verification, the strain energies for each chamber shape at each chamber. To verify the suggested theoretical principle,
mid-height were compared, as shown in Fig. 4(b) up to 30 kPa. the fingertip force of the hybrid PneuNet was simulated up
The concave chamber had the largest strain energy, followed by to 40 kPa, with and without the round edges (Fig. 5(d)). The
PARK et al.: HYBRID GRIPPER WITH SOFT MATERIAL AND RIGID STRUCTURES 69

Fig. 5. Round edge design between the soft chamber and rigid structures: (a) hybrid PneuNet with the round edge design, (b) the detailed configuration of
the round edges (side view), (c) the detailed configuration of the round edges (top view), (d) fingertip force comparison based on the simulation showing the
effectiveness of the round edges.

fingertip force was enhanced by the round edge design, and its
gap was increased as the applied pressure was increased.

III. EXPERIMENT VERIFICATION

A. Fabrication of the Hybrid PneuNet
The three design principles based on the simulation results
were combined simultaneously for the hybrid PneuNet with
the maximized ratio of the rigid structures γrigid = 0.61, the
concave chamber shape with the maximum radius, and round
edge design.
The detailed configuration of the hybrid PneuNet is illus-
trated in Fig. 6. Dragonskin 30 A (Dragonskin 30 A, Smooth-
On [19]) was used as the soft material, manufactured by 3D
printed molds. To accommodate the required concave chamber
shape, 3D printed molds were divided into several parts. The
rigid structures (ABS-P430, Stratasys [20]), manufactured via
3D printer, were inserted into punched holes and sealed with
the silicone. Readily available plastic film was used as an inex-
tensible layer. The top and bottom layer shown in Fig. 6, were Fig. 6. Final design of the hybrid PneuNet, satisfying design principles, for
combined monolithically with a commercial silicone adhesive. the grippers.
70 IEEE ROBOTICS AND AUTOMATION LETTERS, VOL. 4, NO. 1, JANUARY 2019

Fig. 8. Fingertip force comparison experiments: (a) experimental setups

mounted with the hybrid PneuNet, (b) fingertip force comparison for the differ-
ent SPAs.
Fig. 7. Actuation speed comparison experiment: (a) fast PneuNet (137 ×
25 × 20 mm3 ) [2], (b) fiber-reinforced SPAs (136 × 26 × 20 mm3 ) [8], (c)
measurement of the volume change during the bending motion, (d) comparison
for the normalized volume change.
SPA, and hybrid PneuNet, respectively; the normalized volume
change of each actuator, which are calculated as V
V0
, were 0.382,
The teflon tube was inserted to supply air to the chambers. For 0.471, and 0.275, respectively. The developed hybrid PneuNet
the increased grasping ability, a nail was fabricated with Drag- had the smallest normalized volume change, followed by the fast
onskin 10 A and attached to the tip of the hybrid PneuNet; a PneuNet and fiber-reinforced SPA. Thus, the actuation speed of
bumpy shape was added to the bottom layer to enhance friction. the hybrid PneuNet improved 1.3 times compared with the fast
Finally, the fabricated hybrid PneuNet was assembled into the PneuNet, irrespective of the capabilities of the attached electri-
hybrid gripper as shown in Fig. 1. cal pump.

B. Actuation Speed Comparison

Experiments were performed to evaluate the actuation speed C. Fingertip Force Comparison
and fingertip force of the single hybrid PneuNet. The pro- The fingertip force of the hybrid PneuNet was compared to
posed hybrid PneuNet demonstrated enhanced bending speed. that of the fast-PneuNet and fiber-reinforced SPA. The SPAs,
As the actuation speed is affected by the specification of the including the hybrid PneuNet, were attached to a mounting
fluidic pumps, the volume change was measured for the actu- platform. The fingertip normal force was measured using a
ation speed comparison at 180 deg of bending. The conven- force/torque (F/T) sensor (ATI Gamma; ATI [21]), as depicted
tional SPAs including the fast PneuNet (Fig. 7(a)) and fiber- in Fig. 8(a). Given that the general operating pressure range
reinforced SPA (Fig. 7(b)) were compared with the developed of SPAs is about 0–80 kPa, the three SPAs were compared up
hybrid PneuNet [2], [8]. To avoid inaccurate measurements due to 80 kPa. A commercial dispenser (Performus III; Nordson
to compressible air, incompressible water was used to inflate the EFD [22]) was used to inflate the pneumatic chambers.
chambers, as illustrated in Fig. 7(c). The hybrid PneuNet exerted about 1.5–2 times greater finger-
Fig. 7(c) shows the volume change measurements, performed tip force compared with the other SPAs, as shown in Fig. 8(b).
manually 10 times in water. The mean values of the fast PneuNet, At 80 kPa, the measured fingertip force of the fast PneuNet,
fiber-reinforced SPA, and hybrid PneuNet were 9.0, 4.0, and fiber-reinforced SPA, and hybrid PneuNet were 1.28 N, 0.68 N,
2.2 mL, respectively. The measured volume change after bend- and 2.56 N, respectively. Thus, the hybrid PneuNet showed the
ing motion was normalized with respect to the initial volume. better performance in exerting the fingertip force due to the in-
The initial volumes were calculated based on the mold param- teraction between the soft and rigid structures with increased
eters: 23.6, 8.5, and 8.3 mL for fast PneuNet, fiber-reinforced force transmission rate.
PARK et al.: HYBRID GRIPPER WITH SOFT MATERIAL AND RIGID STRUCTURES 71

Fig. 10. Test for grasping different objects with the hybrid gripper : (a) tumbler
(300 g) (b) doll (145 g) (c) tape (502 g) (d) apple (426 g) (e) heat gun (631 g)
(f) banana (144 g) (g) box (212 g) (h) screwdriver (145 g).

IV. APPLICATION
A. Grasping Objects Test
We evaluated the ability of the hybrid gripper to grasp various
objects. The various weight and size of objects were grasped
(Fig. 10) below 80 kPa pressure. Compared with previously
developed grippers, the hybrid gripper grasped heavy objects
such as a heat gun and the duct tape with increased bending
speed; notably, the operating pressure conditions were sim-
ilar to those used in previous studies [3], [24]. The hybrid
gripper was able to grasp different objects with the increased
fingertip force without the need for complicated control algo-
rithms. It was verified that different objects with large weights
were grasped with simple inflation of the hybrid grippers, com-
pared with the mechanisms used in conventional SPA grippers
[3], [7], [24].
Fig. 9. Lifting force measurement: (a) Experimental setup for measuring
the displacement and the lifting force with the motion capture system and
F/T sensor, (b) lifting force of the developed hybrid gripper depending on the B. Application for the Teleoperation System
different operating pressure.
We tested the developed hybrid gripper with a teleoperation
system; these systems require light and compact components for
D. Lifting Force Measurement
mobility. The hybrid gripper uses a small pneumatic component
The maximum lifting force was also measured for the devel- with low pressure and flow rate to exert a large lifting force.
oped hybrid gripper with the hybrid PneuNets. The lifting force Thus, the hybrid gripper offers high actuation speeds without
of the gripper indicates the maximum lifting weight for grasping the need for a bulky pump motor or complex any complicated
objects. The experimental setup for lifting force measurements control algorithms for actuation. The pneumatic system for the
is shown in Fig. 9(a). A spherical dummy object was grasped by hybrid gripper (Fig. 11(a)) consisted of two pneumatic pumps
the hybrid gripper, and the gripper was moved upward manually (DAP-370A) connected in parallel to enhance the flow rate, as
at a slow speed (about 10 mm/s) until it failed to grasp under well as a pressure sensor and solenoid valves to inflate the hybrid
different pressure (20, 40, 60, and 80 kPa). The lifting force and gripper.
displacement of the hybrid gripper were measured with the F/T As shown in Fig. 11(b)–(d), several objects were grasped
sensor and a motion capture system (Prime 17, Optitrack [23]). with the teleoperated hybrid gripper and moved into the basket.
Fig. 9(b) shows the lifting weight as a function of operat- The teleoperation system, which uses a robot arm with seven
ing pressure. The displacement when the gripper released the degrees of freedom, was operated by the motion of the operator.
object increased in proportion to the applied pressure. Due to The motion was measured with a wearable sensor system and
the increased friction between the surface of the gripper and finger motion measurement system [25]. An institutional review
the object, the displacement of the gripper showed the perturba- board (IRB) approved for human operation (UNISTIRB-17-28-
tion (Fig. 9(b)). The maximum lifting force was about 28.7 N, A approved). The baskets containing objects, which weighed in
which showed an enhanced lifting force under similar pressure 1.25 kg, were carried by the gripper. A detailed demonstration
loadings [3], [10]. is shown in a supplementary video.
72 IEEE ROBOTICS AND AUTOMATION LETTERS, VOL. 4, NO. 1, JANUARY 2019

Fig. 11. Application of the hybrid gripper for teleoperation system: (a) compact pneumatic system, grasping (b) the tape (200 g), (c) the bowl (300 g), (d) the
box (150 g), (e) the baskets containing all objects (1.3 kg).

V. CONCLUSION [7] E. Browna et al., “Universal robotic gripper based on the jamming of
granular material,” Proc. Nat. Acad. Sci. United States Amer., vol. 107,
The hybrid gripper, which was assembled with the hybrid no. 44, pp. 18809–18814, 2010.
PneuNet, was developed to achieve enhanced fingertip force [8] P. Polygerinos, Z. Wanga, K. C. Galloway, R. J. Wood, and C. J. Walsh,
“Soft robotic glove for combined assistance and at-home rehabilitation,”
and fast actuation speed based on three design principles: 1) Robot. Auton. Syst., vol. 73, pp. 135–143, 2015.
the degree of bending is proportional to the ratio of the rigid [9] J. Fras, M. Macias, Y. Noh, and K. Althoefer, “Fluidical bending actuator
structure, 2) a concave chamber design is preferred for large designed for soft octopus robot tentacle,” in Proc. IEEE Int. Conf. Soft
Robot., 2018, pp. 253–257.
longitudinal strain, and 3) a round edge shape between soft and [10] K. C. Galloway et al., “Soft robotic grippers for biological sampling on
rigid materials increases the fingertip force of the hybrid Pne- deep reefs,” Soft Robot., vol. 3, no. 1, pp. 23–33, 2016.
uNet. FEM simulations were conducted to verify these design [11] Y. Li, Y. Chen, Y. Yang, and Y. Wei, “Passive particle jamming and its
stiffening of soft robotic grippers,” IEEE Trans. Robot., vol. 33, no. 2,
principles independently. Based on our simulation results, we pp. 446–455, Apr. 2017.
constructed a hybrid gripper with the design principles via the [12] K. Mizushima, T. Oku, Y. Suzuki, T. Tsuji, and T. Watanabe, “Multi-
3D printed molds. With the hybrid design, the fingertip force in- fingered robotic hand based on hybrid mechanism of tendon-driven and
jamming transition,” in Proc. IEEE Int. Conf. Soft Robot., 2018, pp. 376–
creased 1.5–2 times and the actuation speed increased 1.3 times 381.
compared with those of conventional SPAs. The developed hy- [13] A. M. Nasab, A. Sabzehzar, M. Tatari, C. Majidi, and W. Shan, “A soft
brid gripper design included a compact pneumatic system with gripper with rigidity tunable elastomer strips as ligaments,” Soft Robot.,
vol. 4, no. 4, pp. 411–420, 2017.
the solenoid valves and the pressure sensor. As it has low oper- [14] I. Hussain et al., “Modeling and prototyping of an underactuated gripper
ating pressures (<100 kPa), a large compressor with high flow exploiting joint compliance and modularity,” IEEE Robot. Autom. Lett.,
rate was not needed to achieve a large grasping force or fast vol. 3, no. 4, pp. 2854–2861, Oct. 2018.
[15] Y. Chen, F. Wan, T. Wu, and C. Song, “Soft-rigid interaction mechanism
actuation, making it applicable to teleoperated systems without towards a lobster-inspired hybrid actuator,” J. Micromechanics Microengi-
any complicated control algorithms. neering, vol. 28, no. 1, 2017, Art. no. 014007.
[16] S. S. Rao., Mechanical Vibrations. London, U.K.: Pearson, 2011.
[17] P. Martins, R. Natal Jorge, and A. Ferreira, “A comparative study of several
REFERENCES material modelsfor prediction of hyperelastic properties:application to
silicone-rubber and soft tissues,” Strain, vol. 42, pp. 135–147, 2006.
[1] C. Lee et al., “Soft robot review,” Int. J. Control, Autom. Syst., vol. 15, [18] N. E. Dowling, Mechanical Behavior of Meterials. London, U.K.: Pearson,
no. 1, pp. 3–15, 2017. 1999.
[2] B. Mosadegh et al., “Pneumatic networks for soft robotics that actu- [19] Smooth-On, “Smooth-On Dragonskin 30,” 1895. [Online]. Available:
ate rapidly,” Adv. Functional Mater., vol. 24, no. 15, pp. 2163–2170, https://www.smooth-on.com
2014. [20] STRATASYS, “STRATASYS uPrinter SE PLUS,” 1989. [Online]. Avail-
[3] J. Zhou, S. Chen, and Z. Wang, “A soft-robotic gripper with enhanced able: http://www.stratasys.com
object adaptation and grasping reliability,” IEEE Robot. Autom. Lett., [21] ATI industrial automation, “ATI nano 17 force/torque sensor.” [Online].
vol. 2, no. 4, pp. 2287–2293, Oct. 2017. Available: http://www.ati-ia.com
[4] P. Glick, S. A. Suresh, D. Ruffatto, M. Cutkosky, M. T. Tolley, [22] NordsonEFD, “NordsonEFD Performus-V,” 1954. [Online]. Available:
and A. Parness, “A soft robotic gripper with gecko-inspired adhe- http://www.nordson.com
sive,” IEEE Robot. Autom. Lett., vol. 3, no. 2, pp. 903–910, Apr. [23] Optitrack, “Prime 13,” 2018. [Online]. Available: http://optitrack.
2018. com/products/prime-13
[5] P. Moseley, J. M. Florez, H. A. Sonar, W. C. Gunjan Agarwal, and J. [24] H. Zhang, A. S. Kumar, J. Y. H. Fuh, and M. Y. Wang, “Topology optimized
Paik, “Modeling, design, and development of soft pneumatic actuators design, fabrication and evaluation of a multimaterial soft gripper,” in Proc.
with finite element method,” Adv. Eng. Mater., vol. 18, no. 6, pp. 978–988, IEEE RoboSoft, 2018, pp. 424–430.
2015. [25] S. Park, Y. Jung, and J. Bae, “An interactive and intuitive control inter-
[6] A. D. Marchese, R. K. Katzschmann, and D. Rus, “A recipe for soft fluidic face for a tele-operated robot (AVATAR) system,” Mechatronics, vol. 55,
elastomer robots,” Soft Robot., vol. 2, no. 1, pp. 7–23, 2015. pp. 54–62, 2018.