Articulo de Actuadores
Articulo de Actuadores
Articulo de Actuadores
XIANGYU GONG
ALL RIGHTS RESERVED
ii
ACKNOWLEDGEMENTS
iii
Table of Contents
ABSTRACT
ii
ACKNOWLEDGEMENTS
iii
1.
Introduction
2.
Background
3.
Experimental Design
3.1.
3.2.
3.3.
3.4.
10
3.5.
15
4.
16
4.1.
16
4.2.
19
5.
Conclusions
24
Reference
26
Supporting Information
29
29
32
35
39
41
45
iv
47
50
52
1. Introduction
Rotation plays a significant role in machines. Conventional robots often depend
heavily on joints and torque-providing motors, but recent advances in pneumatic,
elastomeric actuators are not capable of delivering torque through pure rotation.
Pneumatic channels/networks based on elastomeric materials of differing geometries
have gained wide applications. The pneumatic, artificial McKibben muscles, for example,
developed in 1950s and 1960s, contract when their volume increases due to
pressurization, which usually provides a linear motion[1].
In the 1990s, Suzumori et al. used flexible microactuators (FMAs) to create multifingered robotic manipulators[2]. They also developed soft-bodied swimming robots,
inspired by the manta ray[3]. Both of these designs relied on cooperative motions of their
pneumatically-driven components, in order to complete complicated tasks, such as
manipulating an object, screwing a bolt, or swimming underwater. Pneumatic networks
(PneuNets) designed by the Whitesides Group consisted of a series of inflatable chambers
embedded in elastomers[4]. With these inflatable structures, they demonstrated complex
motion at varying speeds. Relying on the simple motion of bending, soft grippers
manipulated fragile objects with pressurized fingers. Multigait soft robots crawled and
undulated by changing their gaits[5]. Further demonstrations have also included
camouflage[6], resistance to puncture and impact[7], elastomeric origami-like devices[8],
soft robots powered by explosions[9], fast PneuNets (fPNs)[10], and an untethered soft
robot resilient to adverse environments[11]. However, these soft robots introduced above
have generally relied on two fundamental modes of motion: bending and extension.
2
Researchers have also made rolling, belt-like robots[12,13] and a spherical rolling
robot[14], which did not employ soft wheel-and-axle structures. Their structures (circular
belts and spheres) enabled rotation and might be capable of higher rotational speeds in
the future. Nonetheless, vehicles that rotate completely upon themselves might have
difficulty in transporting payloads and handling pneumatic tubing.
We explore a unique configuration of inflatable stators paired with rotors. The design
presented in this paper provides another mode: rotary motion, which is independent of
bending or extension. This paper introduces two types of designs for the actuators
integrated with multiple methods of fabrication and different materials. To demonstrate
the application for these actuators, we built a winch equipped with rotary actuators,
which is capable of applying torque to lift weights. For this type of actuator (Type 1), the
stator with air-filled bladders surrounds an internal rotor. Another type of actuator (Type
2) has the stator inside its rotor, and the assembly functions as a wheel. Using the latter
design with the stator inside the rotor, this work highlights the capability of fabricating
physically compliant vehicles with soft wheels, which are capable of withstanding
mechanical impact and maneuvering on land and in water.
2. Background
Current pneumatically-driven soft robots, such as starfish-like grippers[4,15], octopus
arms[1618], multigait locomotors[5], robotic tentacles[19], and manta rays[3] can mimic the
motion and behavior of many kinds of soft animals. Nevertheless, in nature, many
creatures consist of rigid structures with bones coupled to soft tissues, such as muscle,
cartilage, and skin. In the case of limbs, these composite structures are capable of moving
relative to each other, in part, because of their interconnecting joints. The ability to create
joints that can rotate with multiple degrees of freedom remains an open challenge for soft
robots. From the perspective of biologically inspired (bio-inspired) design, the rotary
actuators described in this work are a step toward rotating, joint-like structures that might
provide torque in future limbed robots with soft, synthetic components.
Furthermore, while many soft robots have their basis in bio-inspired design or
evolution[18,20,21], wheels on axles distinct from rolling mechanisms employed by
animals, such as, armadillos[22], pangolins, wheel spiders and pebble toads for soft
machines are a notable exception[33,34], as these fully rotating components do not exist in
nature at the mesoscale. Future limbed or wheeled robots with elastomeric components
might find uses in search and rescue missions in extreme environments or over various
terrains. With the low sensitivity of elastomers to electromagnetic radiation, wheeled
robots or rovers made of elastomers might also facilitate travel in space or manipulation
of objects during magnetic resonance imaging (MRI). Lacking sharp or metal
components, soft rotary actuators may also serve as safe motors and be compatible with
future human-friendly robots or vehicles.
4
Since the work of Karl Sims in the 1990s, there have been continual efforts to use
3. Experimental Design
3.1. Peristaltic Motion as a Mechanism for Providing Torque
Peristaltic motion is common in nature. Peristalsis generated by the circular muscle of
human esophagus helps push a bolus of food toward the stomach[26]. People have applied
peristaltic actuation into polymer-gel-based actuators that transport objects via oscillating
reactions[27]. Researchers have also fabricated micropumps for drug delivery using
peristaltic mechanics[28]. A variety of limbless crawlers, such as earthworms, snails and
snakes, use peristaltic mechanics and have inspired multiple worm-like robots[29,30].
The inspiration of our design comes from the combination of peristaltic motion with
the standard layout of a rotary motor consisting of a rotor and a stator. The stator is a
circular, elastomeric structure with embedded, hollow bladders. Each bladder has an inlet
that provides pressurized air. We divide or distribute the flow of pressurized air into
multiple subgroups of bladders with each subgroup being inflated and deflated
sequentially. The sequential inflations and deflations generate a peristaltic wave
surrounding the stator. The peristaltic wave, offset some radial distance from the
rotational center of the rotor, exerts a set of circumferential forces to provide torque on
the rotor.
This work demonstrates two configurations of rotary actuators that enable the
conversion of peristaltic motion to torque. In the first configuration, which we name Type
1, the stator surrounds the rotor. In the second configuration named Type 2, the stator sits
inside the rotor. Figure 1A shows one complete cycle of sequential inflation of four
subgroups of bladders on stators without rotors for both types of configurations. These
Thus, in each step (from one subgroup to the next subgroup), the rotor turns by 22.5
(360/16).
3.2. Meso-scale Soft Lithography to Fabricate the Actuators
Using computer-based, geometric models created in SolidWorks (Dassault), a threedimensional (3-D) printer (FlashForge Creator) produced Type 1 rotors and molds for
elastomeric components. The Supporting Information (Figure S1) discusses more details
concerning fabrication. In this paper, we made stators of a silicone-based rubber Ecoflex
00-50 (Smooth-On Inc.) by bonding the cured top layer to the partially cured bottom
layer without applying an intermediate layer of adhesive as shown in Figure S1A. With
these two layers, a stator contains closed rectangular bladders (length: 0.5 cm, width 0.15
cm, depth: 0.5 cm) inside of it.
We also present methods for fabricating the two types of rotors (Type 1: 3.65-cm
diameter, Type 2: 9.2-cm diameter). We fabricated the hard internal rotor (Type I) from
thermoplastic acrylonitrile butadiene styrene (ABS) directly through the 3-D printer using
computer-aid design (Figure S1B). On the other hand, we bonded three structured layers
of molded elastomers to make external rotors (Type II) (Figure S1C). A 3-D printer
produced the molds for each layer.
There is friction between the stator and the rotor. In order to reduce the friction and
smoothe the rotation, we applied lubricant between the sliding interfaces of the rotor and
stator. After qualitative experiments with different lubricants, including dish soap, water,
silicone oil, and Super Lube grease (Synco Chemical Corp.), we concluded that Super
Lube grease provided the smoothest rotation of these options. Super Lube grease is a
nontoxic synthetic lubricant with PTFE, and it does not drip or dry out easily. Moreover,
it does not react with or significantly alter the texture of the elastomers.
3.3.Analysis of Strain with Pressurization of the Stators
A sufficient inflation of the bladders without damage is a prerequisite for enabling a
continual rotation of the rotors. To understand how the pressurized air acted on the
deformation of bladders, we estimated the strains on a stator during inflation with finiteelement methods (FEMs) (Figure S2A, C). With uniaxial tensile test data for Ecoflex 0050, we fitted the stress-strain curve to a 3rd-ordered Yeoh model of hyperelasticity
(Figure S3) in a similar fashion to the work completed by Mosadegh, et al.[10] Given a
specified static internal pressure, the simulations showed the volumes of a subgroup of
bladders and their corresponding maximum principal strains when pressurized. The
maximum principal strain (~200%) did not exceed maximum strain at break of the
Ecoflex 00-50 (980%) provided by the manufacturer, which means the bladders might
not yield at the applied pressure (varying from 0 to 62.1 kPa).
We then validated the FEM simulations qualitatively by comparing the deformation
of stators caused by internal pressure with experimentally acquired images (Figure S2 B,
D). Because a stator consists of two separate layers that were bonded together, these
experiments also proved the applied internal pressure that sufficiently enabled the
rotation would not cause the delamination between the bonded parts. More details
regarding the results of the simulations and the experimental images are in the Supporting
Information.
10
11
when a solenoid then cut off the airflow before the actual internal pressure in the bladders
reached the pressure set on the regulator.
In addition to controlling the rate of inflation, we needed to allow a period of time to
exhaust the air from a group of bladders before inflating an adjacent group. Thus, the
rotational speed of the actuators relied on both times of inflation and deflation for each
group of bladders. Figure 2A depicts the staggered inflation and deflation of four
subgroups of bladders embedded in a stator. The blue solid lines represent the command
signals for the opening and closing of the valves. Each valve associated with a group of
bladders opened and subsequently closed to control the air flowing into these
corresponding bladders. Here, the period between opening and closing of the valve was
the inflation time tinflation , which was also the time needed to inflate a group of bladders to
the minimum volume. The red dotted lines indicate the change in the internal
pressure/volume within the bladders. As the internal pressure approached the pressure
corresponding to the minimum volume that could enable a consistent rotation of the rotor,
we halted the inflation by switching the position on the valves to cut the pressurized
airflow and exhaust the air in the bladders to the atmosphere. Unlike inflation times,
which are dependent on the pressure set on the regulator, the deflation times remain
consistent for a given volume of the inflated bladders. To get the time of deflation tdeflation ,
we used a high-speed camera (120 frames/second) to record the deflation of each group
(Figure S5, Movie S1). In this way, we found the average deflation times of Type 1 and
Type 2 actuators were approximately 130 msec and 190 msec, respectively.
12
13
Figure 2. Controllable setups used for characterization. (A) The sequential inflation and
deflation of groups of bladders. Four solenoid valves opened and closed to deliver
pressure to four groups of bladders. (B) Experimental setup for characterizing set
pressure rotational speed for Type 1 actuators. (C) A Type 2 actuator attached with a
light-weighted string for external load. We used this setup to characterize this actuator.
(D) Experimental setup for easily profiling the relationships between rotational speed of
Type 2 actuator and the external torque acting on it. The scale bar is 2 cm.
14
With an approach similar to that of controlling a DC motor by changing the applied
voltage, we manipulated the speed of the rotary actuator by varying the pressure set on
the regulator. With a specific inflation time corresponding to a specific set pressure, we
calculated the rotational speed with the following formula:
(1)
15
Discussion includes the characteristic curves of set pressure inflation time and
rotational speed set pressure.
3.5. Characterization of Torque and Rotational Speed
People have characterized the relationship between external torque and rotational
speed for motors. Similarly, in addition to our ability to control the rotational speed by
manipulating the air pressure, we were also interested in how the rotational speed
changed with an external load acting on the actuator at a fixed pressure on the regulator,
in terms of applications of these rotary actuators in the real world. To examine the
relationships between torque and the rotational speeds of both types of actuators, we kept
using the basic setup for the Type 2 actuator shown in Figure 2C, by attaching external
loads on the string. On the other hand, we built a symmetric experimental setup for the
Type 1 actuator (Figure 2D), because we found it difficult to directly attach external
loads to a single Type 1 rotor. The new setup consisted of two 3-D printed brackets, each
fixing a stator inside. A cylindrical shaft bridged two Type 1 rotors on each of its ends,
and a string was attached to the shaft. Detailed designs of the brackets and the shaft are in
the Supporting Information (Figure S6). We then inserted the shaft with rotor-shaped
ends into the stators. This winch-like setup allowed us to add up the weight easily and
keep the rotors balanced and stable inside the stators when profiling the torque
rotational speed curve. By using this setup to perform the experiments, we had to divide
the weight by 2 to get the external torque acting on one actuator.
For both actuators, we set a pressure of 68.9 kPa (around 10 psi) on the regulator and
started pressurizing to make these actuators turn without any external loads. While
adding up the external loads on the actuators, we expected more inflation time to
16
overcome the increased pressure force and friction between the rotor and stator caused by
the increased load. Thus, increasing external torque acting on the actuators would lower
the overall rotational speed due to longer inflation time needed.
In addition to the capability of transporting payloads as already shown during the
characterization, we also demonstrated how the soft rotary actuators could be integrated
with others work (i.e., a soft gripper imbedded with PneuNet)[4]. A winch equipped with
a soft gripper going up and down was able to manipulate/transport an object. (Figure S7,
Movie S2)
17
Figure 3. Characteristics of soft rotary actuators. (A) Measured relationships between set
pressure on the regulator and inflation time for both types of actuators without any
external load. (B) Calculated rotational speed as a function of set pressures for both types
of actuators based on the measurements of (A). (C) Torque-speed characteristics of
actuator Type 1 and the estimated power based on its linearized characteristics. (D)
Torque-speed characteristic of actuator Type 2 and the estimated power based on its
linearized characteristics.
18
rotors at a distance closer to the axis of rotation than in the Type 2 actuators. The
bladders in both types inflate by approximately the same amount at the same rate, and as
this action moves closer to the axis of rotation, the speed of rotation increase. There was
also a remarkable change in the speed for the Type 1 actuator at around 55.2 kPa (8 psi)
caused by the snap-through behavior of the inflatable elastomers[31,32].
Figure 3C-D characterized the rotational speeds of the two types of actuators,
respectively, as a function of the torque at a specified pressure set on the regulator. Given
the diameters of rotating units of the setups (the shaft of the winch: 4.05 cm; rotor of
Type 2 actuator: 9.2 cm), we calculated the torque based on the external loads we added
on them. At a fixed pressure of 68.95 kPa (10 psi), starting with no load, the winch for
Type 1 rotated at the speed of 17 RPM approximately, and the Type 2 had the rotational
speed of around 10.2 RPM. Here, we noted that the speed of Type 2 actuator matched the
characteristics of set pressure rotational speed (Figure 3B) well at a set pressure of
70 kPa (10.2 psi). However, the speed of the winch was higher than that of a single Type
1 shown in Figure 3B, which might be due to increased power of doubled actuators.
Generally, the rotational speeds of the actuators dropped when the external loads
went up, until the torque reached 0.0065 Nm for Type 1, and 0.0375 Nm for Type 2. At
that point, the actuators were no longer able to bear the external torque or turn continually,
with the rotors sliding back inside the stators. Using linear regression, we approximated
the relationships between rotational speed and external torque to a linear model, based on
the experimental data. In this way, we could also estimate the power of the actuators as a
function of the speed of rotation. According to the modeled power rotational speed
curves (orange dotted-lines) in Figure 3C-D, the maximum powers that the actuators
19
could achieve were 0.0084 W and 0.026 W, when the rotational speeds were at a
minimum. The overall power of the Type 1 actuator was lower than that of Type 2, but
the Type 1 actuator could achieve a higher rotational speed.
4.2. A Squishy, Wheeled Vehicle
To demonstrate the application of the rotary actuators powered by the pneumatic
system, we built prototypes of squishy vehicles equipped with the actuators working as
wheels with a diameter of 9.2 cm. Combining multiple controllable motions, a twowheeled vehicle demonstrated the capability of navigating around an obstacle at a high
speed (4.9 cm/s) when the pressure (set on the regulator) was 86.2 kPa (12.5 psi). (Figure
S8, Movie S3)
We also extended the two-wheeled device to a rover-like vehicle with four wheels
(Figure 4A-B). A structure made of Ecoflex 00-50 served as a chassis to connect two
two-wheeled modules. In this paper, we did not address the issue of making the fourwheeled vehicle turn, which means it only moved in a straight line.
At the same set pressure of 86.2 kPa (12.5 psi) used to drive the two-wheeled vehicle,
we needed longer inflation time (300 msec) to compensate for the extra loading
associated with an increase in the total weight (1.23 kg) and friction between stators and
rotors. The rotational speed of each wheel was 7.7 RPM, and therefore, the speed of the
vehicle was 3.7 cm/sec (Figure 4C-E, Movie S4).
20
21
This four-wheeled vehicle demonstrates several potential advantages when it comes
22
23
Figure 6. A four-wheeled vehicle negotiating a rocky terrain with a puddle on its way. (A)
The vehicle starts at higher level of rocks. (B) The vehicle went down into the puddle
with its body bending. (C) The vehicle went up to the other side of rocks. (Movie S6)
24
5. Conclusions
The soft rotary actuators generate motion through biologically-inspired peristalsis that
uses alternating inflation and deflation of pneumatic bladders (Figure 1). We describe
two types of rotary actuators with their complementary configurations of stators and
rotors. Type 1 has an internal plastic (ABS) rotor fabricated by a 3-D printer, while Type
2 has an external soft rotor surrounding a stator. Lubricant helps reduce the friction
between stators and rotor to achieve a smoother rotation. FEM simulations estimated the
local strain of the inflated bladders to ensure the material (Ecoflex 00-50) would not yield
or fail. Also the experiments conducted in the same fashion verified the FEM results and
implied that there was no delamination on the stators at the pressure ranging from 0 to
55.5kPa (8 psi) (Figure S2).
For a given design with specified dimensions (i.e., dimensions of air-filled bladders,
wall thickness of the bladders, and the distributions of these bladders on a stator), the
speed and torque of the soft actuators depended on at least three factors: pressure/flow
rate of the compressed air delivered to the bladders, timing of the solenoid valves, and the
load on the actuators. Based on the external load applied to the actuators and a desired
rotational speed, we could set the pressure on the regulator and the timing to open/close
the valves through a programmable pneumatic system. These rotary actuators have
torque-speed characteristics similar to those of DC motors (Figure 3). The quantitative
characterization also serves as an initial baseline to follow for future soft, rotary
applications.
A winch equipped with Type 1 actuators combined with a soft gripper in the work
demonstrated how different soft robotic systems cooperate and might be capable of more
25
complex tasks (Figure S7). Type 2 actuators enabled the fabrication of a squishy,
wheeled vehicle with the ability to withstand mechanical impact (Figure 5), negotiate
rocky and wet terrains (Figure 6), and work under water as an amphibious vehicle
(Figure S9),
Although the configuration of stator and rotor has solved the issue of the tangled
tubing caused by rotational motion, the tubing that supplies pressurized air is still limiting
the workspace of the actuators. An onboard, autonomous system for a soft robot is
desirable, and still remains a somewhat open challenge[11]. To better understand the soft
rotary actuators, we still need to conduct fatigue tests for them and see how the
alternating inflation and deflation might change the properties of the material. We can
also optimize the design to reduce or avoid stress concentration, and explore alternative
elastomers with higher strengths, so that the actuators can possibly have better
performance (i.e., longer life and higher rotational speed). In terms of characterization
and control, dynamic simulations for the interactions between a stator and a rotor at
varied pressures are desirable, to reveal the intrinsic relationship of the inflation and
rotation of a rotary actuator.
There is a limitation of the squishy, four-wheeled vehicle: unlike a traditional car, the
vehicle has a body with four fixed axles, which requires a different mechanism to make
the vehicle turn. Thus, in the demonstrations, we were only focused on the motion in a
straight line. To extend the degrees of freedom, we can develop more sophisticated
control patterns for each wheel, based on rotational speed and overall dimension of the
vehicle. An alternative design, such as a pneumatic, morphing vehicle body, will also
potentially inspire more complex motions.
Reference
26
[1]
C.-P.
Chou,
B.
Hannaford,
IEEE
Trans.
Robot.
Autom.
1996,
12,
90.
[2]
K.
Suzumori,
S.
Iikura,
H.
Tanaka,
in
1991
IEEE
Int.
Conf.
Robot.
Autom.
1991
Proc.,
1991,
pp.
16221627
vol.2.
[3]
K.
Suzumori,
S.
Endo,
T.
Kanda,
N.
Kato,
H.
Suzuki,
in
2007
IEEE
Int.
Conf.
Robot.
Autom.,
2007,
pp.
49754980.
[4]
F.
Ilievski,
A.
D.
Mazzeo,
R.
F.
Shepherd,
X.
Chen,
G.
M.
Whitesides,
Angew.
Chem.
Int.
Ed.
2011,
50,
1890.
[5]
R.
F.
Shepherd,
F.
Ilievski,
W.
Choi,
S.
A.
Morin,
A.
A.
Stokes,
A.
D.
Mazzeo,
X.
Chen,
M.
Wang,
G.
M.
Whitesides,
Proc.
Natl.
Acad.
Sci.
U.
S.
A.
2011,
108,
20400.
[6]
S.
A.
Morin,
R.
F.
Shepherd,
S.
W.
Kwok,
A.
A.
Stokes,
A.
Nemiroski,
G.
M.
Whitesides,
Science
2012,
337,
828.
[7]
R.
V.
Martinez,
A.
C.
Glavan,
C.
Keplinger,
A.
I.
Oyetibo,
G.
M.
Whitesides,
Adv.
Funct.
Mater.
2014,
24,
3003.
[8]
R.
V.
Martinez,
C.
R.
Fish,
X.
Chen,
G.
M.
Whitesides,
Adv.
Funct.
Mater.
2012,
22,
1376.
[9]
R.
F.
Shepherd,
A.
A.
Stokes,
J.
Freake,
J.
Barber,
P.
W.
Snyder,
A.
D.
Mazzeo,
L.
Cademartiri,
S.
A.
Morin,
G.
M.
Whitesides,
Angew.
Chem.
2013,
125,
2964.
[10]
B.
Mosadegh,
P.
Polygerinos,
C.
Keplinger,
S.
Wennstedt,
R.
F.
Shepherd,
U.
Gupta,
J.
Shim,
K.
Bertoldi,
C.
J.
Walsh,
G.
M.
Whitesides,
Adv.
Funct.
Mater.
2014,
24,
2163.
[11]
M.
T.
Tolley,
R.
F.
Shepherd,
B.
Mosadegh,
K.
C.
Galloway,
M.
Wehner,
M.
Karpelson,
R.
J.
Wood,
G.
M.
Whitesides,
Soft
Robot.
2014,
1,
213.
[12]
N.
Correll,
.
D.
nal,
H.
Liang,
E.
Schoenfeld,
D.
Rus,
in
Exp.
Robot.
(Eds.:
O.
Khatib,
V.
Kumar,
G.
Sukhatme),
Springer
Berlin
Heidelberg,
2014,
pp.
227240.
[13]
A.
D.
Marchese,
C.
D.
Onal,
D.
Rus,
in
2011
IEEERSJ
Int.
Conf.
Intell.
Robots
Syst.
IROS,
2011,
pp.
756761.
[14]
K.
W.
Wait,
P.
J.
Jackson,
L.
S.
Smoot,
in
2010
IEEE
Int.
Conf.
Robot.
Autom.
ICRA,
2010,
pp.
37573762.
[15]
E.
Brown,
N.
Rodenberg,
J.
Amend,
A.
Mozeika,
E.
Steltz,
M.
R.
Zakin,
H.
Lipson,
H.
M.
Jaeger,
Proc.
Natl.
Acad.
Sci.
U.
S.
A.
2010,
107,
18809.
[16]
L.
Margheri,
C.
Laschi,
B.
Mazzolai,
Bioinspir.
Biomim.
2012,
7,
025004.
[17]
B.
Mazzolai,
L.
Margheri,
M.
Cianchetti,
P.
Dario,
C.
Laschi,
Bioinspir.
Biomim.
2012,
7,
025005.
[18]
S.
Kim,
C.
Laschi,
B.
Trimmer,
Trends
Biotechnol.
2013,
31,
287.
[19]
R.
V.
Martinez,
J.
L.
Branch,
C.
R.
Fish,
L.
Jin,
R.
F.
Shepherd,
R.
M.
D.
Nunes,
Z.
Suo,
G.
M.
Whitesides,
Adv.
Mater.
2013,
25,
205.
[20]
D.
Trivedi,
C.
D.
Rahn,
W.
M.
Kier,
I.
D.
Walker,
Appl.
Bionics
Biomech.
2008,
5,
99.
[21]
R.
Pfeifer,
M.
Lungarella,
F.
Iida,
Commun
ACM
2012,
55,
76.
[22]
F.
P.
Melo,
J.
A.
Siqueira,
B.
A.
Santos,
O.
lvares-da-Silva,
G.
Ceballos,
E.
Bernard,
Biotropica
2014,
46,
257.
[23]
K.
Sims,
in
Proc.
21st
Annu.
Conf.
Comput.
Graph.
Interact.
Tech.,
ACM,
New
York,
NY,
USA,
1994,
pp.
1522.
[24]
K.
Sims,
Artif.
Life
1994,
1,
353.
[25]
N.
Cheney,
R.
MacCurdy,
J.
Clune,
H.
Lipson,
in
Proc.
15th
Annu.
Conf.
Genet.
Evol.
Comput.,
ACM,
New
York,
NY,
USA,
2013,
pp.
167174.
[26]
W.
G.
Paterson,
GI
Motil.
Online
2006,
DOI
10.1038/gimo13.
[27]
S.
Maeda,
Y.
Hara,
R.
Yoshida,
S.
Hashimoto,
Angew.
Chem.
2008,
120,
6792.
[28]
M.
M.
Teymoori,
E.
Abbaspour-Sani,
Sens.
Actuators
Phys.
2005,
117,
222.
[29]
S.
Kim,
E.
Hawkes,
K.
Cho,
M.
Joldaz,
J.
Foleyz,
R.
Wood,
in
IEEERSJ
Int.
Conf.
Intell.
Robots
Syst.
2009
IROS
2009,
2009,
pp.
22282234.
[30]
S.
Seok,
C.
D.
Onal,
K.-J.
Cho,
R.
J.
Wood,
D.
Rus,
S.
Kim,
IEEEASME
Trans.
Mechatron.
2013,
18,
1485.
[31]
A.
N.
Gent,
Int.
J.
Non-Linear
Mech.
2005,
40,
165.
[32]
G.
Mao,
T.
Li,
Z.
Zou,
S.
Qu,
M.
Shi,
Int.
J.
Solids
Struct.
2014,
51,
2109.
[33]
S.
J.
Gould,
Kingdoms
without
wheels,
Natural
History,
vol.
90,
no.
3,
p.
42,1981.
[34]
R.
Dawkins,
Why
dont
animals
have
wheels?,
The
Sunday
Times,
vol.
24,
1996.
27
28
Supporting Information
29
Supporting Information
Rotary Actuators Based on Pneumatically-Driven Elastomeric Structures
Fabrication of Soft Rotary Actuators
Fabrication of the stators (Figure S1A) required the molding and assembly of two
separate components: one for the structure with groups of inflatable bladders (top layer)
and one for a layer to seal these chambers (bottom layer). We took off the top layer of the
stator from the mold at first and then put it on the half-cured bottom layer in the mold
after cleaning the surface of the top layer with isopropyl alcohol. Usually it took 10
minutes for the bottom layers to be partially cured at room temperature. After the bottom
layer was fully cured, the complete stator was ready to be taken off. The designated dots
on the bottom layer indicate the positions of the inlets for compressed air. After punching
16 holes on the bottom layer with needles, we inserted silicone rubber tubing (OD:1/16,
ID:1/32) into each bladders through the holes. As the final step, we sealed the gap
between the tubing and the holes with silicone rubber adhesive Sil-Poxy (Smooth-On
Inc.). In general, the fabrication for the two types of the stators is the same and both of
them are made of silicone-based elastomer. However, we have made Type 1 rotors
(Figure S1B) of a hard, 3-D printed thermoplastic (i.e., ABS), and Type 2 rotor of
thermosetting elastomer Mold Star 30 (Smooth-On Inc.).
For Type 2 rotor, 3-D printer produced the molds for the different layers of the external
rotor and then we molded these layers (Figure S1C). We bonded the layers with Sil-Poxy.
There are three layers: one layer with the main structure designed to interact with the
stator inside of it, one PDMS layer to constrain the slide on the axial direction between
30
the rotor and the stator, and one layer that constrains and fixes the stator on the other side.
The transparency of the PDMS layer also shows the internal structures and helps us
monitor how the rotor and the stator interact when controlling an actuator. The material
for the other two layers (blue) is Mold Star 30 (Smooth-On Inc.). Mold Star 30 is a type
of platinum silicone rubber, which is also an elastomer, but it is stiffer than the Ecoflex
series.
31
Figure S1. Meso-scale soft lithography of the rotary actuators. (A) A 3-D printed mold
patterns cast Ecoflex into the thick portion of a stator. After curing the thick portion of
the stator, it bonds to a thin cover of Ecoflex that is partially cured and adhesive. The
tubing then goes into the molded bladders of the stator with Sil-Poxy providing an
adhesive sealing. (B) A rotor of Type 1, which was fabricated with ABS by a 3-D printer.
(C) A rotor of Type II consists of three layers, which surrounded an internal stator (not
pictured).
32
33
snap-through instability) did not show any delamination between the two layers of the
stator, which proved the feasibility of the design, and validated the methods of fabrication.
34
Figure S2. (A, C) Experimental tests on the behavior of two types of stator with static
internal pressure up to 55 kPa (8 psi). (B, D) Simulated behavior of the stators showing
the deformation and maximum principal strain as a function of internal pressure up to 53
kPa (7.7 psi). We hid the tubing that provided pressurized air behind the meshed
background.
35
U = Ci0 I1 3 +
i=1
i=1
2i
1
J 1
Di
(2)
T$
!
where I1 = tr #dev ( FF) & , J = det F , and F is the deformation gradient and Ci 0 and Di
"
%
( )
are the materials parameters. When we set N = 3 (3rd-ordered Yeoh model), the curve
fitting module in ANSYS calculated the parameters. For the mean data of the five
samples, we have C10 = 1.90 102 , C20 = 9.0 104 , C30 = 4.75106 , D1 = D2 = D3 = 0 .
For the data of the sample, the parameters are C10 = 1.75102 , C20 = 6.70 104 ,
C30 = 2.65106 , D1 = D2 = D3 = 0 .
36
In this paper, we chose the sample data to perform the static simulations. By varying the
applied pressure inside of embedded bladders, we got the deformation of the bladders and
the estimated principal strain shown in Figure S1B, D. According to results of the
simulations, the maximum principal strain of the inflated bladders was 2, which means
the region of most interest on the curve is the strain of 0-200% indicated in Figure S3
(Bottom).
37
38
Figure S3. Stress-strain curve and Yeoh model of Ecoflex 00-50. Top: Uniaxial tensile
test (blue solid line) of the elastomer (Ecoflex 00-50) used for simulations of the stators
inflated by varied static pressures. We fitted the stress-stain curve with Yeoh model of
hyperelasticity (red dotted line). Bottom: The region of most interest (strain: 0 to 250%)
since the principal strains were less than 2, according to the simulations.
39
40
Figure S4. Pneumatic System. (A) Schematic of the programmable pneumatic system
used for soft rotary actuators. (B) A photo of the setup carried on a cart. (The pressure
regulator and the camera are not shown in the photo.)
41
42
deflation time tdeflation . We indicated the calculated tdeflation for each experiment in the
figure. The average tdeflation for Type 1 is 125 msec (130 msec if using nine experiments),
and the average tdeflation for Type 2 is 190 msec.
43
44
Figure S5. Estimation of deflation times using high-speed videos. (A) Four groups of
deflation of a Type 1 stator suggested an average deflation time tdeflation = 125 msec. (B)
Four groups of deflation of a Type 2 stator gave an average deflation time tdeflation = 190
msec. The red boxes indicate the LED light that would go out in the next frame.
45
46
Figure S6. A prototype of the winch designed for Type 1 actuators. (A) A 3-D printed
cylindrical shaft coupled with two Type 1 rotors. (B) Two 3-D printed brackets fixed on a
wooden stand.
47
48
PneuNets through a syringe. Then, the winch started lifting the gripped with the object.
The external torque on the winch increased to 9.2 103 Nm due to the mass of the object.
In order for the winch to carry heavier load at the same speed, we slightly increased the
pressure on the regulator. In Figure S7 C-D (Height: from 0 to 14 cm), the winch raised
the gripper and the object at a rate of 1.6 cm/sec, turning in around 17 steps.
49
Figure S7. A winch equipped with two Type 1 stators acting in parallel and a soft gripper
(16.4 grams) that could grasp and lift an object weighted 30.7 grams. (A) The gripper
started at the top. (B) The winch lowered the gripper at a speed of 1.6 cm/sec. (11 steps
of the rotor). (C) The gripper inflated and grasped the object. (D) The winch raised the
gripper and grasped object at a rate of 1.6 cm/sec (17 steps of the rotor).
50
51
Figure S8. (A) Design of the squishy two-wheeled vehicle, (B) The highlighted pieces
are two paperclips used to balance the vehicle. (C-H) To navigate around an obstacle on
its way, the two-wheeled soft vehicle moved forward, stopped, moved backward, made a
left turn, went forward, turned back to straight and then went forward. The arrows
indicate the moving directions of the vehicle.
52
53
Figure S9. A squishy, four-wheeled vehicle carried rocks and moved forward under
water at a speed of 3.6 cm/sec, driven by the pneumatic system. A floating plastic ball
indicated the level of the water in tank.