REVIEW

published: 31 July 2018

doi: 10.3389/frobt.2018.00084

Soft Robots Manufacturing: A Review

François Schmitt, Olivier Piccin, Laurent Barbé and Bernard Bayle*

ICube laboratory, University of Strasbourg/INSA Strasbourg/CNRS, Strasbourg, France

The growing interest in soft robots comes from the new possibilities offered by these
systems to cope with problems that cannot be addressed by robots built from rigid
bodies. Many innovative solutions have been developed in recent years to design soft
components and systems. They all demonstrate how soft robotics development is closely
dependent on advanced manufacturing processes. This review aims at giving an insight
on the current state of the art in soft robotics manufacturing. It first puts in light the
elementary components that can be used to develop soft actuators, whether they use
fluids, shape memory alloys, electro-active polymers or stimuli-responsive materials.
Other types of elementary components, such as soft smart structures or soft-rigid
hybrid systems, are then presented. The second part of this review deals with the
manufacturing methods used to build complete soft structures. It includes molding, with
possibly reinforcements and inclusions, additive manufacturing, thin-film manufacturing,
shape deposition manufacturing, and bonding. The paper conclusions sums up the pros
and cons of the presented techniques, and open to developing topics such as design
Edited by: methods for soft robotics and sensing technologies.
Matteo Cianchetti,
Scuola Sant’Anna di Studi Avanzati, Keywords: soft robotics, manufacturing process, soft components, prototyping, design
Italy

Reviewed by:
Andrew T. Conn,
1. INTRODUCTION
University of Bristol, United Kingdom
Ali Sadeghi, The interest in soft robots has significantly increased in recent years. This evolution is more than
Fondazione Istituto Italiano di just a trend. The scientific community is seeking to carry out a real technological breakthrough,
Technologia, Italy justified by the need to evolve toward human friendly robotics. Industrial robots are fast and
*Correspondence: precise systems, based on rigid-body mechanisms, which ensure high throughput in the production
Bernard Bayle of manufactured goods. The further development of robotic manufacturing now relies on the
[email protected] integration of workers in the manufacturing systems, allowing to perform tasks that require
cognitive capacities still beyond the reach of artificial systems. In this context, collaborative
Specialty section: manipulation has been a noticeable evolution in recent years. Industrial robots of a new type
This article was submitted to have appeared, with design and control strategies focused on the ability to perform safe physical
Soft Robotics,
human-robot interactions. In parallel, the development of flexible systems in robotics, e.g., serial
a section of the journal
elastic actuators, has also contributed to the emergence of new mechanisms, with similar safety
Frontiers in Robotics and AI
and interaction objectives. If this new generations of robots has sometimes been referred to as soft
Received: 22 December 2017
robotics (Albu-Schaffer et al., 2008), such systems are still rigid link robots, which embed sensing
Accepted: 25 June 2018
and control capabilities allowing to operate more safely in a human environments. Soft robots are
Published: 31 July 2018
a step further in the attempt to benefit from mechanical compliance in order to offer safety and,
Citation:
simultaneously, approach the incredible capabilities of evolved living systems in complex tasks.
Schmitt F, Piccin O, Barbé L and
Bayle B (2018) Soft Robots
Bio-inspiration, which has long been controversial in the robotics community, is certainly one
Manufacturing: A Review. of the precursors of soft robotics. Bio-inspired systems, which mimic animal or human capabilities,
Front. Robot. AI 5:84. have however long been designed using mostly rigid-body architectures, associated to soft parts.
doi: 10.3389/frobt.2018.00084 Pioneer works in soft robotics are to be found in research efforts that proposed to simultaneously

Frontiers in Robotics and AI | www.frontiersin.org 1 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

inspire themselves from natural properties, and to imagine and technological opportunities raised by the development of soft
innovative ways to produce them (see Figure 1). Cho et al. (2009), robotic systems has been such that a new scientific community
in a review dating back from the early period of soft robotics, well has gathered in a very short time, leading to the apparition of
emphasized the very close relationships between manufacturing SoRo in 2014, the first scientific journal dedicated to soft robotics
evolutions and the design and fabrication of systems of a totally (SORO, 2014). Since then, the recent open-access Frontiers
novel generation. Soft robots are systems built from materials in Robotics and AI has also opened a section dedicated to
with mechanical properties similar to those of living tissues, soft robotics (Frontiers, 2016), that started publishing articles
designed and manufactured in a very innovative way rather in late 2016. During this same period, soft robotics has even
than artificially assembled by serial or parallel arrangements found a greater visibility in the prestigious Nature publication,
of elementary blocks, as it was the case for rigid-body a multidisciplinary scientific journal in which robotic science
robots. and technologies are very rarely published. In this very well
The enthusiasm generated by soft robotics comes from the documented review, Rus and Tolley show the extent of research
convergence of different scientific communities for the design of fields concerned by soft robotics developments (Rus and Tolley,
these new machines. Born at the crossroads between chemistry, 2015). The recent scientific development of soft robotics has
plastics engineering, and mechatronics (Trivedi et al., 2008; taken place in parallel with successful popular initiatives, like the
Ilievski et al., 2011), soft robots have now spread in a great one carried out by the Soft Robotics Toolkit project (Holland
number of directions, leading computer scientists to work on et al., 2017). This project, initiated at Harvard University in order
design processes adapted to their non conventional structural to inspire and develop skills in undergraduate design students,
analysis, physicists and material engineers to innovate in sensing, has expanded both geographically and in terms of audience,
power supply and information processing. The endless scientific including contributions ranging from high school projects to
research competitions.
Though soft robotics is a recent scientific and technological
topic, a significant number of review articles have already been
published (Trivedi et al., 2008; Ilievski et al., 2011; Laschi and
Cianchetti, 2014; Majidi, 2014; Rus and Tolley, 2015; Laschi
et al., 2016), justified by the exponentially growing interest in
the scientific community, and by the number of excellent recent
contributions. Some observations can be made, based on these
literature reviews. Up to now, most of the works that have been
published in soft robotics research focus on the development
of elementary functionalities, underlying the development of
soft robotics. Very few research has already been concerned
with the development of fully functional systems, such as those
mentioned in futuristic forecasts (Majidi, 2014), highlighting the
fact that soft robotics is still in its infancy. In our view, it has
to be particularly emphasized that soft robotics development
will be very closely related to the development of advanced
manufacturing processes within the robotics community, and
to the overall development of new materials and manufacturing
technologies. If new solutions will come from innovations in
manufacturing, conversely, soft robotics development is an
opportunity to boost innovation in manufacturing and design.
Very few review papers focus on soft robotics manufacturing
methods. Cho et al. (2009) published a review article on the
subject very early in the development of soft robotics. In the
present article, we propose an update on the technologies
that have been used or developed during the last decade of
expansion of soft robotics. It will complement other recent review
papers (Rus and Tolley, 2015; Laschi et al., 2016) that cover a
broader spectrum. The paper focus is deliberately limited to the
manufacturing of components and systems that constitute the
structures of soft robots. In spite of the numerous developing
technologies for soft sensors, embedded soft electronics or soft
FIGURE 1 | Example of a bioinspired soft octopus tentacle (Laschi et al., energy sources, we chose not to include them in the article body,
2012), based on a braided polymeric network that can be constricted using as their presentation would deserve much more than a section in
SMA actuators. Image courtesy of C. Laschi and M. Cianchetti, reproduced
with permission.
a review paper dedicated to the manufacturing of soft robots in
general.

Frontiers in Robotics and AI | www.frontiersin.org 2 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

The present State of the Art is organized as follows. In

section 2, we review the elementary components that can be
used to develop soft robots. They all include original properties
that differentiate them totally from the building blocks of
conventional robots. They offer solutions to both the design and
fabrication of hybrid rigid/soft systems, or to the development
of fully soft systems. Section 3 deals with the different
manufacturing methods used to obtain complete soft structures,
including molding, with possibly reinforcements and inclusions,
additive manufacturing, shape deposition manufacturing, and
bonding. The conclusion will sum up the lessons learnt from
this literature review and will open to various concerns that are
closely related to the problem of manufacturing soft robots. In
particular it puts in light the recent research in this field, i.e., the
interconnection between soft robots manufacturing and design. FIGURE 2 | Soft worm robot as proposed in Calderón et al. (2016). The radial
It also provides the reader with somes references to start the expansion of the axial actuator is limited by the use of stiffer o-rings. The radial
exploration of the field of soft sensors. actuators have not been reinforced, allowing them to expand in all the
directions. Image courtesy of A. A. Calderón and N. O. Pérez-Arancibia,
reproduced with permission.

2. SOFT COMPONENTS FOR ROBOT

BUILDING
2.1. Soft Fluidic Actuation when inflated with pressurized fluids, have been proposed. Since
The easiest way to create a homogeneous load on a deformable then, multiple systems and variants inspired by the McKibben
piece of material is to raise the applied pressure by the use of some PAM have been proposed, including the Pleated PAM (Daerden,
fluidic medium. This concept is at the core of a large number 1999) or tendon driving devices (Tsujiuchi et al., 2006). Some
of soft robotic actuators and their various implementations. other approaches are based on the use of vacuum in order to
Simplest soft fluidic actuator designs involve bladders that may collapse a buckling structure (Yang et al., 2016a) or to contract a
quickly be inflated with compressed air, thus generating the compressible skeleton using the mechanical tension of a flexible
impulse needed for jumping robots (Ni et al., 2015), or creating sheet on a more rigid structural element (Li et al., 2017a) (see
fluid motion from mechanical action (Giorgio-Serchi et al., 2016) Figure 4).
or from internal combustion (Loepfe et al., 2014; Schumacher Another widely studied actuator category is the family of soft
et al., 2014). By giving the bladder a specific spatial architecture fluidic bending actuators which feature a general beam-shaped
such as bellows (Tolley et al., 2014b; Digumarti et al., 2017), geometry with an inflatable chamber along their longitudinal
or by thickening its extremities (Qi et al., 2015), one expansion axis. Their bending motion results from a change of extensibility
direction may be favored with respect to the others. Alternatively, of two opposing sides of the beam. This property can also be
thin anisotropic films can also be used to build bladders (Niiyama modulated by design. The obtained actuators may exhibit a large
et al., 2014). Their high transverse flexibility with respect to range of motion in one or more directions depending on the
their in plane stiffness can be used to favor deformation in number of internal chambers, their topology and the actuation
specific directions. Additionally, pleats can be added to the method. In single-material structures, simple variations in the
design (Nishioka et al., 2017; Sareen et al., 2017), allowing wall thickness leads to varying degrees of deformations under
the motion to follow prescribed trajectories. This “mechanical the same pressure (Gorissen et al., 2013), creating the desired
programming” can also be performed using materials with bending motion. Another possibility is to reinforce one side of the
different stiffness values. Indeed, by limiting the radial expansion actuator using a stiffer material (Polygerinos et al., 2015b). These
of tubular bladder using stiffer materials (Calderón et al., 2016; actuators may also present bellowed geometries (Polygerinos
Robertson et al., 2016) (see Figure 2), it is possible to build et al., 2013; Mosadegh et al., 2014) (see Figure 5) that allow
actuators able to generate large linear extension. Some work has faster actuation due to a lower volume change of the internal
also been done to use the elastic instabilities between pre-stressed chamber and lower strain level, which in turn increases the
elastic tubes and compressed air to generate large displacements actuator lifetime. In order to improve the deformation along the
using small fluid volume variations (Overvelde et al., 2015) (see bending direction, it is also possible to coil inextensible cables
Figure 3). around the tubular section, thus limiting the radial deformation
Using pressurized fluid, it is also possible to obtain contraction in favor of the axial deformation. Inflating a single chamber
instead of expansion. Those works have been mainly motivated bending actuator usually results in a displacement in only
by the idea of mimicking the behavior of biological muscles one direction. However actuating two chambers in opposition
that contract when stimulated. In the 1950s, pneumatic may generate a bidirectional planar bending (Yap et al., 2016),
artificial muscles (PAM) such as McKibben muscles (Chou and or a multidirectional spatial bending when more chambers
Hannaford, 1996; Tsagarakis and Caldwell, 2000), which contract are available (Martinez et al., 2013). Alternatively, the planar

Frontiers in Robotics and AI | www.frontiersin.org 3 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

FIGURE 3 | Demonstration of the fast acting system based on elastic instabilities, as presented in Overvelde et al. (2015). The soft actuator comprises two
interconnected fluidic segments, with different tube and braid lengths. The actuator is first inflated with an initial volume of 16 mL, then decoupled from the syringe
pump and connected to a small reservoir containing only 1 mL of water. When the system is inflated with water, it takes more than 1 s for the changes in length,
pressure, and internal volume to fully take place. By replacing water with air, the time is reduced from 1.4 s to 300 ms. Moreover, by adding an additional reservoir of
air to increase the energy stored in the system, the actuation time can be further decreased to 100 ms. Image courtesy of K. Bertoldi, reproduced with permission.

FIGURE 4 | Illustrations of the origami-inspired artificial muscle proposed in Li et al. (2017a). The leftmost figure shows several actuator scales with respect to a
quarter. The figures on the right presents the muscle used to pull on a finger before and after actuation using air vacuum. Image courtesy of S. Li, reproduced with
permission.

bidirectional motion is also available when the actuator is fed chambers that can deform along the bending profile (Ilievski
either with positive or negative relative pressure (Ogura et al., et al., 2011; Shepherd et al., 2011; Tolley et al., 2014a). Like
2009). It is also possible to “mechanically program” the bending tubular bending actuators, pneunets also rely on an inextensible
motion of the actuator by varying the section of the internal layer to ease differential deformation. This technique has also
chamber (Deimel and Brock, 2013, 2016), by changing the been adapted to create actuators exhibiting specific motions
configuration of the reinforcements (Polygerinos et al., 2015a) or such as undulatory snake-like or fish-like motion (Onal and
even by adding an external inextensible sheath (Galloway et al., Rus, 2012; Onal et al., 2013; Marchese et al., 2014), or even
2013). cyclic motion (Correll et al., 2014). The bending motion could
Another possible way to generate a bending motion is to also be obtained using internal explosion, creating impulsion as
create a pneumatic network (or pneunet) of small distributed required by jumping robots (Shepherd et al., 2013). Additionally,

Frontiers in Robotics and AI | www.frontiersin.org 4 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

hinges on foldable systems (Firouzeh et al., 2013), or stiffness

tuning layers for soft fluidic actuators (Firouzeh et al., 2015).
SMA can also act as tendon-like actuators, embedded in a rigid
structure (Meisel et al., 2015) or acting as constrictive muscles
for biomimetic soft tentacles (Mazzolai et al., 2012; Cianchetti
et al., 2015). Similar material behavior can also be found in Shape
Memory Polymers (SMP) (Yang et al., 2016b; Paulino et al., 2017).
They exhibit both lower density and stiffness, limiting the energy
they can release during their transition. Similarly, by using pre-
stressed coiled fibrous materials, such as low cost nylon fishing
line, it is possible to create thermo-activated synthetic muscle
fascicles (Haines et al., 2014). Alternatively, it is also possible to
use thermodynamic effects, such as phase transition, to power
thermo-active actuators. From this concept, a solution using an
ethanol and silicon emulsion has been proposed (Miriyev et al.,
2017). This emulsion can then be shaped and used similarly
to classical soft fluidic actuators, without the need of a fluid
FIGURE 5 | Examples of rehabilitation gloves, as proposed in Polygerinos
et al. (2013). Each of the four actuated fingers (A) is equipped with a bellowed
compression/distribution system, as the variation of the internal
pneunet, which can be inflated separately using the flexible tubes (B). The volume is obtained by the vaporization under heat of the ethanol
shape of the pneunet allows the external walls of the bellows to push on each micro-bubbles trapped in the silicon matrix.
other, as visible in (C), allowing a faster actuation of the bending actuator. The heat needed for these thermo-active actuator to work
Image courtesy of C. J. Walsh, reproduced with permission.
is generally brought by electro-resistive elements such as the
material itself, for instance in the case of SMA, or the
addition of resistive wiring in the structure of the actuator
when its constitutive material electric conductivity is low. In
taking inspiration from continuum and hyper-redundant robots,
order to reduce the integration complexity of soft systems,
vertebra vacuum-actuated bending robots can achieve spatial
Electro-Active Polymers (EAP) have been proposed (Carrico
bending motion (Agarwal et al., 2017). Finally, the integration of
et al., 2017a). Several families of EAP exist, such as ionic
radial unstretchable layers instead of longitudinal ones makes it
metal polymer composite (Carrico et al., 2017b), dielectric
possible to create pure rotation instead of bending motion (Song
polymers/elastomers (Suo, 2010) or ionic hydrogels (Ionov,
et al., 2013; Sun et al., 2013).
2014). Their working principle, generally based on the migration
Though less widespread, some methods have been proposed
of ionic elements under an electrical field, limits their application
to create motions using fluids. Taking inspiration from stepper
to thin film or small/micro scale systems as the generated forces
motors, some systems implement peristaltic motion generated by
are usually low and the response delay increases dramatically
the sequential actuation of chained pneumatic chambers (Chen
at larger scales. They allow to implement systems even at
et al., 2014; Gong et al., 2016) or soft material integrating soft
micro scales, ranging from micro-manipulators (Jager et al.,
actuator in order to generate bio-inspired pumping motion
2000) to aquatic micro-walker/swimmers (Kwon Gu Han et al.,
similar to that of the heart (Roche et al., 2014).
2008). Although their implementation on larger system is
While soft fluidic actuators have major advantages such as a
more limited, notable applications of these technologies are for
high specific power and a wide range of available motions, their
example bending actuators for miniature vertebra (Choi et al.,
major drawback lies in the need of a fluidic power source and
2005), electro-activated muscles (Kovacs et al., 2007) or tunable
a complex electro-mechanical system, which may result quite
lenses (Maffli et al., 2015).
bulky. If some research focused on mobile applications has led
Robots are generally built by associating a mechanical
to interesting solutions to overcome these constraints (Marchese
structure, actuators, and their control system. Smart materials
et al., 2014; Tolley et al., 2014a), this remains an issue that
such as Stimuli-Responsive Materials (SRM) allow to combine
deserves further research efforts.
both actuation and control functions, as they passively respond
to external stimuli, such as temperature, light, or chemical
2.2. Alternative Soft Actuation Methods compounds for instance. Similarly to EAP, a wide variety of
Material science community has developed several solutions photo-responsive shape-memory and shape-changing polymers
that rely on alternative energy sources and/or activation stimuli, are readily available (Iqbal and Samiullah, 2013). They can also
giving more options to soft robots designers. be obtained by adding light absorbing particles to classical heat-
Some of these smart materials rely on thermal energy in order sensitive materials (Breuer et al., 2017). Other examples include
to change their state. This is the case with materials such as Shape smart materials that react to the presence of specific fluids, such
Memory Alloys (SMA), for which transitions in the material as graphene monolayer paper (Mu et al., 2015) that react to the
crystalline structure under temperature changes allow the release present of water, or inverse opal polyionic microstructures (Wu
of stored elastic energy. Their high specific power allow them to et al., 2016) that react to various solvents. Even if the use of
be used in applications where compactness is critical, such as live these SRM is limited to specific application, they may effectively

Frontiers in Robotics and AI | www.frontiersin.org 5 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

allow to fully integrate autonomous elements when external of both soft and rigid components, combining safe interactions
energy sources are scarce or on micro-scale integrated systems. with high precision, or low weight with high forces, for instance.
Application of SRM on larger systems however is limited due to In this case, some compromises can be made, resulting in hybrid
their low specific power and the difficulty of stimulus diffusion to systems that are neither soft nor rigid.
the core of thick elements. While not strictly oriented toward soft robot design, some
particularly interesting works have been done on stiffer materials
2.3. Soft Smart Structures in order to create compliant mechanisms (Wood et al., 2008;
Applications of soft materials are not limited to the scope of soft Delimont et al., 2015). Some authors also take inspiration from
actuators. Indeed, they can be used to build new structures that origami and kirigami techniques in order to design structures
can have a smart behavior when exposed to specific conditions. that exhibit both flexible motion about given axis, and a high
These smart properties can either be permanent, or tunable in off-axis stiffness due to mechanical overconstraints (Onal and
order to adapt the structure to an evolving environment. Rus, 2013). Using specific structural patterns it is actually possible
The simplest of these structures are inflatable structures, to create soft actuators exhibiting both a soft behavior and
which differ from fluidic actuators based on chambers by the high forces (Martinez et al., 2012). Further boosted by recent
fact that they are not actively controlled. Pressurized fluid is developments in multimaterial additive manufacturing, these
injected into the structure in order to transform it from a deflated emerging concepts provide new solutions to design complex
to an inflated state, thus changing its mechanical properties. structures that can be produced with minimal amount of
Several examples of such systems are found in the literature operations (Bruyas et al., 2015; Wang and Lee, 2017).
such as inflatable links (Stilli et al., 2017) or robots (Best
et al., 2015; Gillespie et al., 2016) for safe human interactions, 3. SHAPING SOFT STRUCTURES
high payload-to-weight ratio manipulation arms for operation
in inaccessible areas (Voisembert et al., 2013) and inflatable Alongside the development of the technological elements
furniture or architecture (Sareen et al., 2017) that can deploy presented in the previous section, in-depth work has also
starting from a compact initial footprint. been done in order to propose solutions to manufacture these
Other works have been carried out to make soft structures elements. As soft elements are mainly made of polymers, it
compatible with high interaction forces occurring when is natural that most proposed methods have been derived
manipulating or grabbing objects. From this viewpoint, the from solution that have been used for plastics engineering.
ability to switch from a soft to a stiffer structure becomes a Consequently, proposed building methods will mainly focus on
desirable feature. In order to obtain this behavior, some systems molding and adding reinforcements and inclusions of interest.
are based on the use of low temperature fusible alloys. This Some additive manufacturing solutions will also be presented
allows the creation of metamorphic mobile robots that can be as they may offer convenient solutions in order to craft soft
converted into grippers (Nakai et al., 2002), flexible deployable systems. In parallel, a consequent part of the research in soft
structures (Wang et al., 2016) (see Figure 6), tunable stiffness robotics is oriented toward the design of small scale systems. As
materials (Shan et al., 2013) or actuators (Shintake et al., 2015). such, manufacturing methods initially developed for thin film
Another possibility is to use particles or flat sheet layers that electronics manufacturing that were adapted in order to produce
can be jammed together using vacuum, effectively stiffening a soft mechanical systems will also be presented.
structure. This has been used for bending actuators (Cianchetti
et al., 2014; Ranzani et al., 2016; Li et al., 2017b), continuum 3.1. Molding
robots (Kim et al., 2012) or adaptive graspers (Brown et al., A large majority of soft structures are built using catalyzed
2010). polymer such as silicone rubbers that are obtained by mixing
Finally, auxetic structures or metamaterials are an example of two component before molding operations. The homogenization
purely passive structure that can be interesting from a robotic steps that are needed add however air bubbles to the mix. Because
standpoint. They exhibit a negative Poisson’s ratio, meaning that they may add weaknesses to the final structure, those bubbles
they expand orthogonally, instead of contracting, when uniaxially need to be removed, generally by vacuum degassing the mix.
stretched. They have found some use on soft robots as passive Alternatively, spinning the mold and using the centrifugal forces
clutch for the movement of a worm robot (Mark et al., 2016) or can increase the pressure gradient and degas more effectively
in the industry as shock-absorbing materials. than using only gravity (Mazzeo and Hardt, 2013). The degassing
could also be forced by applying vaccuum at strategic locations
2.4. Soft-Rigid Hybrid Systems of the mold during the injection of the polymer. This technique,
Until now, the majority of the presented systems were mostly called vacuum casting, allows to replicate details even in the
based on soft materials. Their major advantages is their ability sub-millimetric range (Zhao et al., 2012).
to deform when actuated and their high specific power while In the case of thermoplastic material molding, such as paraffin
allowing soft interactions with their environment. However this wax, high shrinkage due to thermal retraction can be observed,
comes also with some notable drawbacks such as low positioning leading to dimensional inaccuracies or even shape warping. In
accuracy, the fact they cannot generate or sustain high forces, order to limit these adverse effects, the material should be molded
and also the complexity of the instrumentation and control under packing pressure during the cooling phase, in order to
implementation. Some tasks at hand may however need the best provide additional material and compensate for the shrinkage.

Frontiers in Robotics and AI | www.frontiersin.org 6 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

FIGURE 6 | Metamorphic deployable structure (Wang et al., 2016). The top-left figure illustrates the structure assembly, before curing the polymer. The top-right figure
represents the structure after curing and removing the mold. The deployment of the assembled structure is represented in the bottom figure. Image reproduced from
Springer Nature with the permission of W. Wang.

Another issue appears for multi-step molding where fusible or soluble material, and be destroyed after molding the
functional interfaces, reinforcement (Polygerinos et al., part (Schumacher et al., 2014; Marchese, 2015; Lawrence et al.,
2013), actuators (Roche et al., 2014) or even electronic 2016). Another solution is to use the flexible nature of the molded
components (Correll et al., 2014) may be overmolded. These part. Galloway et al. (2016) use a soft core that can be removed out
added components need to be correctly located and fixed to of the molded part using for example vacuum to ease the process.
the mold, and have to adhere to the inclusion material. This Alternatively, instead of a fully empty volume inside the part,
operation can be facilitated by using specific primer and glues Argiolas et al. (2016) propose to use soft foam material crafted
that may enclose the overmolded component and also allow using soluble filler (salt) mixed with silicone. After curing, the
adhesion with the inclusion material. salt is then dissolved, leaving porous internal volume. Agarwal
Molding is very difficult when it comes to manufacturing et al. use an overmolded foam core to simplify shaping operations
internal volumes and undercuts. Molds are well suited to form (Agarwal et al., 2017).
external shapes in semi-round or full-round, that can then be Internal volumes can also be molded from only one side. If
extracted by simple pulling motion. In the case of internal the precision of the internal geometry is not a priority, rotational
volumes and undercuts, however, pulling is generally not possible molding can be used (Zhao et al., 2015). This method involves
as the mold or the internal core would collide with the molded filling and closing mold, then spinning it about two axes in
material. Several solutions have been developed to tackle this order to homogeneously cover all of its surfaces with a layer of
issue. The easiest solution is to mold the part in several subparts polymer. The obtained parts offer empty internal spaces without
that can then be sealed together by gluing (Tolley et al., 2014b) the need of any core for the molding. The main drawback of
or dipping (Onal and Rus, 2012) the parts in uncured material. this method is that the wall thickness is not constant and tends
The internal volume can also be closed in a later operation by to produce meniscus in area with large depth variations. If, on
adding a layer of uncured material (Ilievski et al., 2011; Sun et al., the contrary, the internal walls details are important, dip coating
2013). The seam obtained through this method may however be can be used. This methods, notably used in the industry to mold
structurally weak because of material heterogeneity. bellows, inflatable balloons and plastic gloves, works by dipping
When more complex internal structures are needed, these a molding core or an insert part in a bath of liquid polymer.
method may become insufficient. In this case, an internal core In order to reduce risks of streaks and excessive thickness, the
is needed during molding operations, an thus the challenge polymer should exhibit both proper viscosity and adhesion to its
becomes to extract this core after the molding operation. For support, and should also be set shortly after dipping.
tubular shapes, the solution is to simply mold the part with Alternatively, when one side of the part is more important
a core passing throughout the mold, and to plug the holes at than the other, infusion molding can be used (Brouwer et al.,
the extremities with additional material afterwards (Martinez 2003). This method, developed in order to mold fiber-reinforced
et al., 2013; Polygerinos et al., 2015b). Other classical shapes composites, can also be adapted to soft composites with fibrous
may take some inspiration from industrial molding: the core matrix. The mold is covered on its opposite side with a plastic bag
could be formed by multiple parts that can be disassembled and and the polymer is then pulled using vacuum through the fibrous
extracted by an access hole. Similarly, the core could be made of reinforcement until completely infused. Although this method

Frontiers in Robotics and AI | www.frontiersin.org 7 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

has initially been designed for molding large parts such as boat with a shore hardness of 00-30 and an elongation at break
hulls, it can also be used to mold smaller parts. of about 900%, as the body of an actuator, in combination
with a slightly harder reinforcement, made of a PDMS rubber
3.2. Reinforcements with a shore A hardness of 50 and an elongation at break
Even if the final shape of a soft robot is dependent on the intrinsic of about 200% (see Figure 8). The contrast between stiffnesses
compliance of its material, and therefore not always predictable, allowed the creation of a pneumatically actuated bending grasper.
the designer has some leverage on the behavior of the structure by Another approach is to embed in the structure a layer of highly
modulating its stiffness along specific directions. This is generally anisotropic material such as paper (Mosadegh et al., 2014)
done by adding reinforcements to the structure. or fabric (Sun et al., 2013) sheets. These layers exhibit very
The simplest way to reinforce a system is by modifying its high stretching stiffness but can bend very easily, making them
geometries, modulating the overall material thickness (Gorissen suitable reinforcement for bending actuators. When in opposite,
et al., 2013) or adding ribs. This allows to increase stiffness in the linear extension of the structure is desired, unstretchable
areas where deformation is not desired for the task. This is threads can be coiled around the structure, allowing it to extend
particularly important for soft fluidic actuator, as an increase with very low diametrical expansion (Robertson et al., 2016).
in the internal pressure may lead to the extrusion of the A combination of both a strain limiting sheet and a coiled
actuator walls, and, ultimately, to the rupture of the chamber. thread can also lead to interesting designs combining twisting
Another possibility to avoid this pitfall is to implement bellowed with bending, or even modulating the bending radius of the
geometries (Tolley et al., 2014b; Yap et al., 2016) or exploit the actuator (Polygerinos et al., 2015a).
collapse of the material due to buckling (Yang et al., 2016a) Reinforcements can be added at any time during the molding
(see Figure 7) to limit the final deformation of the shape. When process. However, in order to avoid friction or hysteresis, they
modulating the geometry is not possible, the modulation of the should be correctly bounded to the soft structure. Strain limiting
material anisotropy may also allow to limit the deformations in layers could be pre-dipped in uncured polymer, and then sticked
specific directions. This is the case for extruded plastic films that to the rest of the structure (Sun et al., 2013). The reinforcements
exhibit high bending flexibility with respect to their high stiffness can also be molded in the initial molding operations (Mosadegh
in traction (Niiyama et al., 2014). et al., 2014) or even overmolded (Polygerinos et al., 2015b)
Modulating the flexibility of soft body parts can also be or glued (Memarian et al., 2015b) later in the process. The
obtained by the combining materials with different mechanical silicon can also be reinforced using fillers such as short fiber or
properties. The combination of two (or more) heterogeneous beads (Tolley et al., 2014a), modifying both the hardness and the
material provides the anisotropy needed to obtain the soft mechanical resistance of the material. Complex behavior can also
mobility required while limiting transverse deformation. It is be obtained using an internal skeleton (Martinez et al., 2012) (see
easily possible to combine materials with similar chemical Figure 9), or additional external reinforcements, such as a fabric
composition but contrasting mechanical properties. For example, sheath (Galloway et al., 2013) or a plastic exoskeleton (Kim and
Shepherd et al. (2011) propose to use of a soft Siloxane rubber, Gillespie, 2015; Tao et al., 2015; Paez et al., 2016). These hybrid
designs benefit both from the low stiffness of the soft core for soft
interactions, and from their more rigid sections for higher forces
generation. The addition of those more rigid components may
also be used to enforce a motion compatible with more classical
kinematics, allowing an easier transposition of dynamic models
from stiff to soft robots.

FIGURE 7 | Example of a soft actuator using buckling rectangular cells to

produce displacements when actuated using vacuum. Such actuators can be
used as synthetic muscles on an exoskeleton as they contract similarly to FIGURE 8 | Walking robot composed of five separately inflatable pneunets
biological muscles when actuated. Image reproduced from Yang et al. (2016a) molded in silicone (Shepherd et al., 2011). The strain limiting layer is
with the permission of G. M. Whitesides and Wiley (Copyright Wiley-VCH composed of PDMS. Image reproduced from PNAS with the permission of
Verlag GmbH & Co. KGaA). G. M. Whitesides.

Frontiers in Robotics and AI | www.frontiersin.org 8 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

this case, the 3D shape is formed by sweeping each layer using

a thermoplastic extruding nozzle that creates a thin polymer
string. Although parts obtained through this process usually
exhibit worse spatial resolutions than those obtained through
photo-activated processes, the low price of FDM machines as
well as their high modularity allowed a wide adoption in the
recent years by both research laboratories and competent do-
it-yourselfers. Another readily available method is the inkjet
printing of polymers. In this process, the printing machine is
equipped with a printhead that can travel in 2D. Each material
layer is printed in a fashion similar to paper desktop printers.
After each layer, the printing tray is displaced as to allow the next
layer to be printed. This process is a good compromise between
printing speed and resolution, while keeping the production
FIGURE 9 | Soft linear actuator based on a silicon cylinder overmolded on a costs relatively low. Finally, halfway between FDM and inkjets,
sheet of paper folded with an origami pattern. The obtained system can be
direct ink writing (Lewis, 2006) is a method that uses solidifying
actuated using pressurized air, exhibiting both large displacements and high
force level. Panels A–C show consecutive states of the actuator during viscous inks to create complex shapes. Although technically more
inflation. Image reproduced from Martinez et al. (2012) with the permission of complex than the previous methods, it makes it possible to
G. M. Whitesides and Wiley (Copyright Wiley-VCH Verlag GmbH & Co. KGaA). produce very small patterns, with a high control over the material
orientation at each point. Unlike more classical methods, the
printing direction is not restricted to one plane at a time, which
permits to build shapes with anisotropic materials, such as short-
3.3. Additive Manufacturing fiber-filled polymers.
Unlike classical machining methods, also called subtractive Comparatively to molding or machining methods, AM offers
manufacturing, where a tool removes scraps from a workpiece more freedom to design complex geometries. For instance, it is
to produce the required shapes, additive manufacturing (AM) possible to build entanglements of convoluted shapes without
is based on the local deposition of small volumes of material any complementary part or internal core that would need to be
in order to directly form the shape. While early AM methods machined using a 5-axis milling machine, if even machinable.
were limited to the production of rigid thermoset polymers, These processes have however some limitations. Indeed, AM
the available technologies have been improving constantly since often relies on scaffolding in order to build precisely overhang
the early 1990’s (Kruth, 1991; Kruth et al., 1998, 2005). It geometries. These scaffolds can be constructed either in model
is now possible to create parts from polymers and possibly material, later removed by machining, or in support material
elastomers, metals, ceramics, and even in some cases to combine removed by mechanical (brittle support) or chemical (soluble
or mix two or more construction materials in a same monolithic support) action. Some manufacturing methods could be used to
part. This allows for instance to include electro-conductive build overhangs without support, but in this case it is possible by
materials in a mechanical structure in order to produce parts either limiting the draft angle (Anver et al., 2017) or the distance
with potentially complex shapes that include 2D or 3D electrical between both ends of the overhang (Yap et al., 2016). Should
circuits (Macdonald et al., 2014; Goh et al., 2017). Depending on these constraints not be respected, the obtained geometries may
the technology, it is possible to build parts with sizes ranging from show excessive sagging, holes and/or mechanical weaknesses.
tens of micrometers to several meters long. Additionally, the scaffolding could be ignored or removed during
Methods most commonly used for additive manufacturing the manufacturing process in order to include elements of
are based on the selective solidification of a liquid or powdered different nature (Meisel et al., 2015), and then be printed over.
material in order to create the desired shape. Stereolithography Additive manufacturing can also give the possibility to
and selective laser sintering technologies use a laser scanning produce components including multiple materials. Indeed, FDM
the shape of each additional layer of the desired 3D shape. and inkjet processes may allow this multimaterial additive
In the case of stereolithography, the laser is focused in order manufacturing by including multiple extrusion/printing heads,
to polymerize a small volume of pre-polymer available as a each loaded with different materials. Using this principle, it is
liquid bath. Selective laser sintering uses the laser to locally fuse possible to create mechanical device combining soft deformable
powdered material. When each layer is finished, the construction parts with more rigid elements, or even to pattern both materials
tray is lowered and a new layer can be constructed on top of in order to obtain intermediary compounds (Bartlett et al., 2015).
it. In the case of selective laser sintering, an additional layer of These methods may open new pathways for hybrid soft-rigid
powder is added using a roller. Newer methods such as digital robotic systems, benefiting both from soft compliant elements
projection lithography (Sun et al., 2005) and continuous liquid and structural reinforcements, similarly to what can be found
interface production (Tumbleston et al., 2015) allow to print in the nature. The contrast in material properties is also not
each layer directly, without the need of 2D-scanning, effectively limited to mechanical properties. Finally, new manufacturing
speeding up the building process. Another method commonly methods based on AM methods but using smart materials
used for AM is the fused deposition modeling (FDM) process. In such as shape memory polymers also emerge. These so-called

Frontiers in Robotics and AI | www.frontiersin.org 9 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

4D-printed or shape-morphing systems are mechanical systems parts that are then bound together. While a number of solutions
obtained through classical multimaterial additive manufacturing are available in order to bound rigid parts together, this choice is
methods. When exposed to specific interactions, for example much more limited for soft parts. Indeed, the interface between
with water (Tibbits, 2014; Gladman et al., 2016) or UV (Liu et al., both parts of a bonded assembly should not exhibit excessive
2012), they morph into a programmed shape. stiffness while still guaranteeing enough adhesion.
A first approach to take advantage of the manufacturing
3.4. Thin-Film Manufacturing process of the parts themselves. Most soft robotic systems
When trying to build soft systems at smaller scales, some of the are obtained through the curing of polymeric materials. It is
aforementioned manufacturing methods become limited. In this possible to slow the material curing at some point in order
case, techniques adapted from thin-film manufacturing process to obtain parts in an intermediate state, where they exhibit
can be preferred. These techniques, historically developed by a solid gel-like behavior, and keep their molded shape. This
the microelectronics industry, allow to create films of precisely allows preparing the bonding interface. Then, two separate
controlled thickness that can be cut and stacked in order parts can be merged simply by first adding a thin layer of
to produce the required functions. It is critical to create a uncured material similar to a glue, and then finishing the curing
homogeneous thin layer of material to cast thin-films properly. process (Ilievski et al., 2011; Marchese et al., 2014). However,
This is made possible by techniques such as blade coating, this method is limited to chemically compatible materials.
where a blade is moved along the manufacturing area, removing Another possibility offered by thermoplastics is mechanical or
excess material to ensure that the thickness of the layer remains thermal fusion. These methods are used to create pockets or
constant. Another technique, called spin coating (Hall et al., pleats: several sheets are thermally sealed together to obtain
2004), consists in spinning a flat bed wetted with the required complex closed geometries (Niiyama et al., 2014; Nishioka
casting material. After spinning at a sufficient speed, a thin layer et al., 2017; Sareen et al., 2017). However, because of the low
of material of constant thickness remains on the bed while the thermal conductivity of polymers, it is difficult to use this same
excess material is ejected by centrifugal forces. method on thicker parts. Finally, the last common, possibly
Another concept widely used in thin film manufacturing is most obvious bonding method, is to use glues. The choice of
layer transfer. These techniques allow to build elements on a an adapted adhesive for a gluing operation may however be
temporary support and then transfer them on the main film. Roll quite challenging. Several criteria affect this choice, including but
transfer process (Sharma et al., 2012) allow to attach or detach not limited to chemical compatibility (and adhesion) with both
elements from a layer using the pressure of a rolling cylinder. Pad glued material, adhesive strength, flexibility, resistance to solvent
printing (Krebs, 2009) is a technique also used in the industry for and to temperature, setting time, viscosity or contact surface
decorating purpose. It uses a transfer pad that collects a pattern roughness.
formed in a stencil and then transfers it on the destination part or Although the aforementioned methods are of an irreversible
assembly by applying pressure on it. nature, there are a few reversible bonding processes. This is of
Using these manufacturing methods with a combination of very high interest from a robotics standpoint as this reversible
etching, cutting and lamination steps, it is possible to create bond may be controlled and thus used for some specific
planar multi-layered systems that can exhibit out of plane applications. This has already been illustrated in the design of
motion (Wood et al., 2008; Jacobsen et al., 2009; Russo et al., climbing (Wang et al., 2013) or reconfigurable (Wang et al., 2014)
2017). robots, but could also be adapted to other soft systems.

3.5. Shape Deposition Manufacturing 3.7. Architectural Considerations

Shape deposition manufacturing is not actually a manufacturing When manufacturing a soft robotic system, material and
method, but a concept to create complex and possibly manufacturing methods are not the only elements that need
multimaterial structures. It is based on the use of additive and to be considered. Though most classical “rigid” actuators have
subtractive operations in a sequence to obtain the desired shape. the ability to naturally move in both directions about their
It is not easy to achieve a fully automated process, as multiple motion axis, soft actuators are often single action actuators.
manufacturing tools must be available at the same spot. Though, If the application requires motions in the two directions, the
this method allows to make a compromise between the pros architecture of the system needs to be adapted. If motions in
and cons of additive and subtractive manufacturing processes. both directions require similar performances, two actuators can
Because there is not a single shape deposition manufacturing be mounted in an antagonistic fashion, as in bio-mechanical
method, there is almost no limit in the combination of systems (Verrelst et al., 2005). An advantage of this configuration
manufacturing processes, allowing a designer to create structures is that when the combined actuators exhibit non-linear stiffness,
combining heterogeneous materials (Dollar and Howe, 2006). both the position of the antagonistic mechanism and its global
stiffness can be controlled separately (Vanderborght et al., 2013).
3.6. Bonding When considering a system with multiple degrees of freedom,
As exemplified previously, it is difficult to produce complex more than two actuators can be mounted antagonistically in
shapes, with multiple functional geometries and potentially order to create complex motion (Bishop-Moser et al., 2012;
several materials. One of the most straightforward approach to Martinez et al., 2013). Alternatively, some actuators may use their
tackle this challenge is to decompose the final system into several internal stiffness or external elastic elements to return back to

Frontiers in Robotics and AI | www.frontiersin.org 10 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

their initial configuration. While this solution is used quite often by single degree of freedom actuators such as classical motors or
for soft fluidic actuators (e.g., Mosadegh et al., 2014; Calderón synthetic muscles.
et al., 2016) or for shape memory material-actuated systems (e.g.,
Hines et al., 2012; Jin et al., 2016), the control of the return
motion is more limited. 4. CONCLUSIONS AND PERSPECTIVES
Another major issue for some soft actuator designs is buckling.
Although this phenomenon also exists in stiff structures, the low 4.1. Various Novel Tools With Pros and
stiffness of some soft actuators render them prone to this elastic Cons
instability. For this reason, it is preferable to use actuators with Soft robotics is a relatively recent topic. However, a wide range
high length to bending modulus ratio in pulling motion than in of systems have already been developed, with a great number
pushing motion, as buckling is generated by compression forces. of applications (Laschi et al., 2016). This technological push
Finally, some actuators rely on internal stress, as for instance has been closely tied to the development of new manufacturing
coiled artificial muscles (Haines et al., 2014) or some SMA-based techniques, allowing to produce increasingly performant devices,
systems (Meisel et al., 2015). From a design standpoint, such that can even include additional functions in top of their purely
systems have to be prestressed at manufacturing time, and later soft-mechanical behavior. The introduction of new innovative
be able to conserve this prestress in order to avoid the destruction materials, such as self-healing polymers (Terryn et al., 2015,
of either the actuator or the rest of the structure. 2018), or biocompatible elastomers will also lead to further
Finally, some systems use the flexibility offered by tendon- expansions for soft robotics the field of medical applications. The
like elements (Manti et al., 2015; Rateni et al., 2015; Mutlu et al., underlying manufacturing technologies already allow to tackle
2016) in order to transmit forces between the actuator and the some of the challenges emerging from the use of soft materials
structure. This architecture allows the mechanical elements to in robotics, and they will go on developing with the advances of
accommodate unstructured environments while being actuated several scientific communities converging on soft robotics.

TABLE 1 | Summary of the principal manufacturing methods presented in this article, the typical scale range of the parts produced and the main applications and
features of each method.

Manufacturing method Part scale [mm] Applications Advantages Drawbacks

Gravity molding 1–103 Structural elements Low implementation cost Degassing required
Fluidic actuators Wide adaptability Long molding time
Centrifugal molding 10−3 –101 Microfluidics, thin systems High precision details More adapted to thin parts
Degassing without vacuum Internal volumes difficult
Vacuum molding 10−3 –101 Microfluidics High precision details Complex positioning
Systems with small details of bubble traps
for some details
Spin casting 1–102 Void elements with thin walls Void chamber easy to cast Internal walls with no details
Simple molds Wall thickness control
Dip coating 1–102 Void elements with thin walls Void chamber easy to cast External walls with no details
Balloons, gloves, bellows Simple molds Streaks/excessive thickness
Vacuum infusion 101 –104 Fiber reinforced elements Low implementation cost Details only on one side
Very large parts Relies on reinforcement

Stereolithography 10−2 –103 Prototyping Low surface roughness Mono-material

Small series manufacturing Large range of materials
Fused deposition 10−1 –102 Prototyping, do-it-yourself Low cost Poor mechanical properties
modeling Low-strength parts Large range of materials High surface roughness
manufacturing Long manufacturing time
Inkjet printing 10−2 –102 Prototyping, 4D printing, Multi-material patterning Highly relying on scaffoldings
Small series manufacturing Low production time High material cost
Direct ink writing 10−3 –1 Micro-machining, 4D printing, High precision Long production time
Precision manufacturing Scaffoldings optional
Continuous liquid 10−2 –101 Prototyping, Low production time Mono-material
interface production Small series manufacturing Low surface roughness

Spin coating 10−4 –10−1 Thin film Fast and simple Thickness control only
Flat film
Tape coating 10−3 –1 Thin film Liquid or powder Thickness control only
Flat or cylindrical film
Pad printing 10−3 –10−1 Thin film, Decoration 2D patterns Stencil required
Works on irregular surfaces

Frontiers in Robotics and AI | www.frontiersin.org 11 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

FIGURE 10 | Example of a soft tubular actuators with external reinforcements, as presented in Agarwal et al. (2016). The left column shows a bending variant of the
actuator while the right column shows a linear extension one. On the top, FEM simulation results are presented: (a4) and (b4) represent the von Mises stress,
respectively for the full actuator and for the soft core; (d1) and (d2) show the von Mises stress for the linear actuator, respectively in free and blocked extension. The
pictures at the bottom show the corresponding experimental results. Image reproduced from Springer Nature with permission from J. Paik.

Frontiers in Robotics and AI | www.frontiersin.org 12 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

A wide range of manufacturing methods have been proposed Hopkins et al. (2015) presents a synthesis and analysis methods
throughout the literature, ranging from methods adapted from for generating soft parallel robots using pneumatic trichamber
their industrial couterparts to new emerging solutions that could actuators. Other methods relying on topological optimization
help shaping new designs in the future. Table 1 sums up the can also be useful in the design of soft robots. First adapted
main manufacturing methods presented in this paper with their from structural cases (Bendsøe and Kikuchi, 1988) to compliant
principal features and applications. Because of the wide diversity mechanisms synthesis (Yin and Ananthasuresh, 2003), they can
of the produced parts in terms of size, geometry and material be used to synthesize soft and multimaterial mechanisms (Hiller
properties, manufacturing methods cannot be readlily compared. and Lipson, 2012; Meisel et al., 2013). Finally, bioinspiration is a
In the first block of the table are gathered the principal vast source of inspiration for soft robots (Kim et al., 2013) as it
molding techniques described in section 3.1. These methods has fueled the field since its early beginnings.
give the designer the best control over material properties but
requires the constuction of molds. The second block presents
AM methods detailed in section 3.3. They offer broad ranges
4.3. Instrumentation and Control
of possibilities to create freeform geometries that could not be Perspectives
obtained otherwise. However the choice in material types is Designers of soft robots also face many other challenges that
generally more restricted than with molding techniques. The are yet to be properly addressed (Rus and Tolley, 2015;
last block shows more specific techniques for manufacturing Laschi et al., 2016). Some ambitious challenges lie in the
multilayer thin-film systems as discussed in section 3.4. instrumentation and control of such systems, and also in the
design of embedded components adapted to the specificities
4.2. Modeling and Design Perspectives of soft robotics. Multimaterial manufacturing makes it today
Modeling is of central importance for characterizing existing possible to embed various elements in the soft matrix of the
soft devices, but also for the design and the synthesis of novel robot, in order to improve the integration of the functionalities.
soft robots. Soft robots can be best modeled as continuous Modularity increases Onal and Rus (2012), with soft robots
deformable media, but analytic models are difficult (if possible) including their own pressure sources (Onal et al., 2011) and
to derive for soft robots exhibiting geometric and behavorial control valves (Marchese et al., 2011). The inclusion of channels
non-linearities. Advances in the field of compliant mechanisms for liquid metal injection (Park et al., 2012; Farrow and
have provided possible approaches for modeling the load- Correll, 2015) or the embedding of conductive hydrogels and
displacement behavior of devices comprising flexure members. electroactive fillers in flexible materials (Larson et al., 2016),
The pseudo-rigid body model approach (Howell, 2001) consists open perspectives in soft electronics (Correll et al., 2014), a step
in setting a correspondence between deformable members and an further in the development of sensing systems for soft robots. To
equivalent rigid-body mechanism where flexures are modeled as date, reliable sensors able to measure the state of a soft system
massless spring elements. Other models can be chosen depending only provide partial information on the state of the system,
on the deflection level, such as Sen and Awtar (2013) for though more and more solutions have been proposed, using for
small to medium deflections, and Holst et al. (2011) for larger example liquid metals (Park et al., 2012; White et al., 2017) or
displacements, but are however limited to 1D beam shape microstructured metal on polymeric substrate (Araromi et al.,
geometries. Most research toward analytic modeling of soft 2016; Atalay et al., 2017). The control of soft robots is also a very
robots is oriented toward fiber-reinforced structures, such as open challenge, and it has to be acknowledged that modeling soft
bending actuators (Memarian et al., 2015a; Wang et al., 2017) device for realtime control is currently at a very early stage in spite
or fiber-reinforced elastofluidic enclosures (Krishnan et al., 2015; of noticeable contributions (Duriez, 2013; Coevoet et al., 2017).
Bruder et al., 2017). All these challenges may currently limit the adoption of
In most case, however, an analytical model is not available. soft robots technologies in the industry, at least in the
Then, iterative finite element methods (FEM) can be used short term. However, they are currently an extraordinary
to solve the equation of continuum mechanics. Although source of investigations for research groups, suggesting many
computationally expensive, they only offer little parametric improvements in the years to come.
design insight (Sen and Awtar, 2013) and are more adapted
at the design stage to check the validity of a closed-form AUTHOR CONTRIBUTIONS
model, or to provide comparative results with experiments (as
shown in Figure 10). Recently, however, real-time FEM for FS has made the in-depth bibliographic research and has written
soft robot simulation and control has been proposed (Duriez, the article body. OP and LB have reviewed the introduction and
2013). The computational speed of such approach comes the sections 2, 3, and 4. BB has written the introduction, and
however with some associated tradeoffs such as simplistic reviewed the whole article.
material constitutive laws and low dynamics. Other approaches
adapted from physics simulation and computer graphics such FUNDING
as those proposed by Hiller and Lipson (2014) allow the
modeling of voxelized structures built with heterogeneous This work was supported by French state funds managed by the
material properties. ANR within the Investissements d’Avenir programme (Robotex
The synthesis of new soft robots can also be adapted from ANR-10-EQPX-44, Labex CAMI - ANR-11-LABX-0004) and by
classical methods already available in the field of mechanism. the Région Grand Est.

Frontiers in Robotics and AI | www.frontiersin.org 13 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

REFERENCES Conference on Robotics and Automation (ICRA) (Singapore), 4313–4320.

doi: 10.1109/ICRA.2017.7989496
Agarwal, G., Besuchet, N., Audergon, B., and Paik, J. (2016). Stretchable materials Carrico, J. D., Tyler, T., and Leang, K. K. (2017b). A comprehensive review of
for robust soft actuators towards assistive wearable devices. Sci. Rep. 6:34224. select smart polymeric and gel actuators for soft mechatronics and robotics
doi: 10.1038/srep34224 applications: fundamentals, freeform fabrication, and motion control. Int. J.
Agarwal, G., Robertson, M. A., Sonar, H., and Paik, J. (2017). Design Smart Nano Mater. 8, 144–213. doi: 10.1080/19475411.2018.1438534
and computational modeling of a modular, compliant robotic assembly Chen, F.-J., Dirven, S., Xu, W., Li, X.-N., and Bronlund, J. (2014). “Design and
for human lumbar unit and spinal cord assistance. Sci. Rep. 7:34224. fabrication of a soft actuator for a swallowing robot,” in Robot Intelligence
doi: 10.1038/s41598-017-14220-3 Technology and Applications 2, Vol. 274, eds J.-H. Kim, E. T. Matson, H. Myung,
Albu-Schaffer, A., Eiberger, O., Grebenstein, M., Haddadin, S., Ott, C., Wimbock, P. Xu, and F. Karray (Cham: Springer International Publishing), 483–493.
T., et al. (2008). Soft robotics: from torque feedback controlled lightweight Cho, K.-J., Koh, J.-S., Kim, S., Chu, W.-S., Hong, Y., and Ahn, S.-H. (2009).
robots to intrinsically compliant systems. IEEE Robot. Automat. Mag. 15, 20–30. Review of manufacturing processes for soft biomimetic robots. Int. J. Precis.
doi: 10.1109/MRA.2008.927979 Eng. Manufactur. 10:171. doi: 10.1007/s12541-009-0064-6
Anver, H. M., Mutlu, R., and Alici, G. (2017). “3D printing of a thin-wall soft and Choi, H. R., Jung, K., Ryew, S., Nam, J.-D., Jeon, J., Koo, J. C., et al.
monolithic gripper using fused filament fabrication,” in 2017 IEEE International (2005). Biomimetic soft actuator: design, modeling, control, and applications.
Conference on Advanced Intelligent Mechatronics (AIM) (Munich), 442–447. IEEE/ASME Trans. Mechatr. 10, 581–593. doi: 10.1109/TMECH.2005.856108
doi: 10.1109/AIM.2017.8014057 Chou, C.-P., and Hannaford, B. (1996). Measurement and modeling of McKibben
Araromi, O. A., Walsh, C. J., and Wood, R. J. (2016). “Fabrication of pneumatic artificial muscles. IEEE Trans. Robot. Automat. 12, 90–102.
stretchable composites with anisotropic electrical conductivity for doi: 10.1109/70.481753
compliant pressure transducers,” in 2016 IEEE SENSORS (Orlando, FL), Cianchetti, M., Calisti, M., Margheri, L., Kuba, M., and Laschi, C. (2015).
1–3. doi: 10.1109/ICSENS.2016.7808424 Bioinspired locomotion and grasping in water: the soft eight-arm OCTOPUS
Argiolas, A., Mac Murray, B. C., Van Meerbeek, I., Whitehead, J., Sinibaldi, E., robot. Bioinspir. Biomimet. 10:035003. doi: 10.1088/1748-3190/10/3/035003
Mazzolai, B., et al. (2016). Sculpting soft machines. Soft Robot. 3, 101–108. Cianchetti, M., Ranzani, T., Gerboni, G., Nanayakkara, T., Althoefer, K., Dasgupta,
doi: 10.1089/soro.2016.0004 P., et al. (2014). “Soft robotics technologies to address shortcomings in today’s
Atalay, A., Sanchez, V., Atalay, O., Vogt, D. M., Haufe, F., Wood, R. J., et al. minimally invasive surgery: the STIFF-FLOP approach. Soft Robot. 1, 122–131.
(2017). Batch fabrication of customizable silicone-textile composite capacitive doi: 10.1089/soro.2014.0001
strain sensors for human motion tracking. Adv. Mater. Technol. 2:1700136. Coevoet, E., Escande, A., and Duriez, C. (2017). Optimization-based inverse model
doi: 10.1002/admt.201700136 of soft robots with contact handling. IEEE Robot. Automat. Lett. 2, 1413–1419.
Bartlett, N. W., Tolley, M. T., Overvelde, J. T., Weaver, J. C., Mosadegh, B., Bertoldi, doi: 10.1109/LRA.2017.2669367
K., et al. (2015). A 3D-printed, functionally graded soft robot powered by Correll, N., Önal, Ç. D., Liang, H., Schoenfeld, E., and Rus, D. (2014).
combustion. Science 349, 161–165. doi: 10.1126/science.aab0129 “Soft autonomous materials–using active elasticity and embedded distributed
Bendsøe, M. P., and Kikuchi, N. (1988). Generating optimal topologies in computation,” in Experimental Robotics (Berlin; Heidelberg: Springer),
structural design using a homogenization method. Comput. Methods Appl. 227–240.
Mech. Eng. 71, 197–224. doi: 10.1016/0045-7825(88)90086-2 Daerden, F. (1999). Conception and Realization of Pleated Pneumatic Artificial
Best, C. M., Wilson, J. P., and Killpack, M. D. (2015). “Control of a pneumatically Muscles and Their Use as Compliant Actuation Elements. PhD thesis, Vrije
actuated, fully inflatable, fabric-based, humanoid robot,” in 2015 IEEE-RAS Universiteit Brussel.
15th International Conference on Humanoid Robots (Humanoids) (Seoul), Deimel, R., and Brock, O. (2013). “A compliant hand based on a novel pneumatic
1133–1140. doi: 10.1109/HUMANOIDS.2015.7363495 actuator,” in 2013 IEEE International Conference on Robotics and Automation
Bishop-Moser, J., Krishnan, G., Kim, C., and Kota, S. (2012). “Design of soft (Karlsruhe), 2047–2053. doi: 10.1109/ICRA.2013.6630851
robotic actuators using fluid-filled fiber-reinforced elastomeric enclosures in Deimel, R., and Brock, O. (2016). A novel type of compliant and underactuated
parallel combinations,” in 2012 IEEE/RSJ International Conference on Intelligent robotic hand for dexterous grasping. Int. J. Robot. Res. 35, 161–185.
Robots and Systems (Vilamoura), 4264–4269. doi: 10.1109/IROS.2012.638 doi: 10.1177/0278364915592961
5966 Delimont, I. L., Magleby, S. P., and Howell, L. L. (2015). A family of dual-segment
Breuer, L., Guthmann, E., Schöning, M. J., Thoelen, R., and Wagner, T. (2017). compliant joints suitable for use as surrogate folds. J. Mech. Design 137:092302.
Light-stimulated hydrogels with incorporated graphene oxide as actuator doi: 10.1115/1.4030875
material for flow control in microfluidic applications. Proc. Eurosens. Conf. Digumarti, K. M., Conn, A. T., and Rossiter, J. (2017). Euglenoid-inspired giant
1:524. shape change for highly deformable soft robots. IEEE Robot. Automat. Lett. 2,
Brouwer, W. D., van Herpt, E. C. F. C., and Labordus, M. (2003). Vacuum injection 2302–2307. doi: 10.1109/LRA.2017.2726113
moulding for large structural applications. Compos. A Appl. Sci. Manufactur. 34, Dollar, A. M., and Howe, R. D. (2006). A robust compliant grasper via
551–558. doi: 10.1016/S1359-835X(03)00060-5 shape deposition manufacturing. IEEE/ASME Trans. Mechatr. 11, 154–161.
Brown, E., Rodenberg, N., Amend, J., Mozeika, A., Steltz, E., Zakin, M. R., et al. doi: 10.1109/TMECH.2006.871090
(2010). Universal robotic gripper based on the jamming of granular material. Duriez, C. (2013). “Control of elastic soft robots based on real-time finite element
Proc. Natl. Acad. Sci. U.S.A. 107, 18809–18814. doi: 10.1073/pnas.1003250107 method,” in 2013 IEEE International Conference on Robotics and Automation
Bruder, D., Sedal, A., Bishop-Moser, J., Kota, S., and Vasudevan, R. (2017). (ICRA) (Karlsruhe: IEEE), 3982–3987. doi: 10.1109/ICRA.2013.6631138
“Model based control of fiber reinforced elastofluidic enclosures,” in 2017 IEEE Farrow, N., and Correll, N. (2015). “A soft pneumatic actuator that can sense grasp
International Conference on Robotics and Automation (ICRA) (Singapore), and touch,” in 2015 IEEE/RSJ International Conference on Intelligent Robots and
5539–5544. doi: 10.1109/ICRA.2017.7989653 Systems (IROS) (Hamburg), 2317–2323. doi: 10.1109/IROS.2015.7353689
Bruyas, A., Geiskopf, F., and Renaud, P. (2015). “Toward unibody robotic Firouzeh, A., Salerno, M., and Paik, J. (2015). “Soft pneumatic actuator
structures with integrated functions using multimaterial additive with adjustable stiffness layers for Multi-DoF Actuation,” in 2015 IEEE/RSJ
manufacturing: case study of an MRI-compatible interventional device,” International Conference on Intelligent Robots and Systems (IROS) (Hamburg),
in 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems 1117–1124. doi: 10.1109/IROS.2015.7353510
(IROS) (Hamburg), 1744–1750. doi: 10.1109/IROS.2015.7353603 Firouzeh, A., Sun, Y., Lee, H., and Paik, J. (2013). “Sensor and actuator
Calderón, A. A., Ugalde, J. C., Zagal, J. C., and Pérez-Arancibia, N. O. integrated low-profile robotic origami,” in 2013 IEEE/RSJ International
(2016). “Design, fabrication and control of a multi-material-multi-actuator Conference on Intelligent Robots and Systems (Tokyo), 4937–4944.
soft robot inspired by burrowing worms,” in 2016 IEEE International doi: 10.1109/IROS.2013.6697069
Conference on Robotics and Biomimetics (ROBIO) (Qindao), 31–38. Frontiers (2016). Frontiers in Robotics and AI, Soft Robotics Section.
doi: 10.1109/ROBIO.2016.7866293 Galloway, K. C., Becker, K. P., Phillips, B., Kirby, J., Licht, S., Tchernov, D., et al.
Carrico, J. D., Kim, K. J., and Leang, K. K. (2017a). “3D-printed ionic (2016). Soft robotic grippers for biological sampling on deep reefs. Soft Robot.
polymer-metal composite soft crawling robot,” in 2017 IEEE International 3, 23–33. doi: 10.1089/soro.2015.0019

Frontiers in Robotics and AI | www.frontiersin.org 14 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

Galloway, K. C., Polygerinos, P., Walsh, C. J., and Wood, R. J. (2013). Kim, S., Laschi, C., and Trimmer, B. (2013). Soft robotics: a bioinspired evolution
“Mechanically programmable bend radius for fiber-reinforced soft actuators,” in robotics. Trends Biotechnol. 31, 287–294. doi: 10.1016/j.tibtech.2013.03.002
in 2013 16th International Conference on Advanced Robotics (ICAR) Kim, Y.-J., Cheng, S., Kim, S., and Iagnemma, K. (2012). “Design of a tubular
(Montevideo), 1–6. doi: 10.1109/ICAR.2013.6766586 snake-like manipulator with stiffening capability by layer jamming,” in 2012
Gillespie, M. T., Best, C. M., and Killpack, M. D. (2016). “Simultaneous position IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
and stiffness control for an inflatable soft robot,” in 2016 IEEE International (Vilamoura), 4251–4256. doi: 10.1109/IROS.2012.6385574
Conference on Robotics and Automation (ICRA) (Stockholm), 1095–1101. Kovacs, G., Lochmatter, P., and Wissler, M. (2007). An arm wrestling robot
doi: 10.1109/ICRA.2016.7487240 driven by dielectric elastomer actuators. Smart Mater. Struct. 16:S306.
Giorgio-Serchi, F., Arienti, A., and Laschi, C. (2016). Underwater soft-bodied doi: 10.1088/0964-1726/16/2/S16
pulsed-jet thrusters: actuator modeling and performance profiling. Int. J. Robot. Krebs, F. C. (2009). Pad printing as a film forming technique for polymer solar cells.
Res. 35, 1308–1329. doi: 10.1177/0278364915622569 Solar Energy Mater. Solar Cells 93, 484–490. doi: 10.1016/j.solmat.2008.09.003
Gladman, A. S., Matsumoto, E. A., Nuzzo R. G., Mahadevan, L., and Lewis, J. A., Krishnan, G., Bishop-Moser, J., Kim, C., and Kota, S. (2015). Kinematics of a
(2016). Biomimetic 4D printing. Nat. Mater. 15:413. doi: 10.1038/nmat4544 generalized class of pneumatic artificial muscles. J. Mech. Robot. 7:041014.
Goh, G. L., Agarwala, S., Goh, G. D., Tan, H. K. J., Zhao, L., Chuah, T. K., doi: 10.1115/1.4029705
et al. (2017). Additively manufactured multi-material free-form structure Kruth, J.-P. (1991). Material incress manufacturing by rapid prototyping
with printed electronics. Int. J. Adv. Manufactur. Technol. 94, 1309–1316. techniques. CIRP Ann. 40, 603–614. doi: 10.1016/S0007-8506(07)61136-6
doi: 10.1007/s00170-017-0972-z Kruth, J.-P., Leu, M. C., and Nakagawa, T. (1998). Progress in additive
Gong, X., Yang, K., Xie, J., Wang, Y., Kulkarni, P., Hobbs, A. S., et al. (2016). Rotary manufacturing and rapid prototyping. CIRP Ann. 47, 525–540.
actuators based on pneumatically driven elastomeric structures. Adv. Mater. 28, doi: 10.1016/S0007-8506(07)63240-5
7533–7538. doi: 10.1002/adma.201600660 Kruth, J.-P., Van Den Broucke, B., Van Vaerenbergh, J., and Mercelis, P. (2005).
Gorissen, B., Vincentie, W., Al-Bender, F., Reynaerts, D., and Volder, “Benchmarking of different SLS/SLM processes as Rapid Manufacturing
M. D. (2013). Modeling and bonding-free fabrication of flexible fluidic techniques,” in International Conference Polymers and Moulds Innovations
microactuators with a bending motion. J. Micromech. Microeng. 23:045012. (PMI), 20-24 April 2005, Gent Belgium, Paper 525 (Ghent).
doi: 10.1088/0960-1317/23/4/045012 Kwon, G. H., Park, J. Y., Kim, J. Y., Frisk, M. L., Beebe D. J., and Lee, S. H.
Haines, C. S., Lima, M. D., Li, N., Spinks, G. M., Foroughi, J., Madden, J. D., (2008). Biomimetic soft multifunctional miniature aquabots. Small 4, 2148–
et al. (2014). Artificial muscles from fishing line and sewing thread. Science 343, 2153. doi: 10.1002/smll.200800315
868–872. doi: 10.1126/science.1246906 Larson, C., Peele, B., Li, S., Robinson, S., Totaro, M., Beccai, L., et al. (2016). Highly
Hall, D. B., Underhill, P., and Torkelson, J. M. (2004). Spin coating of thin and stretchable electroluminescent skin for optical signaling and tactile sensing.
ultrathin polymer films. Polym. Eng. Sci. 38, 2039–2045. doi: 10.1002/pen.10373 Science 351, 1071–1074. doi: 10.1126/science.aac5082
Hiller, J., and Lipson, H. (2012). Automatic design and manufacture of soft robots. Laschi, C., and Cianchetti, M. (2014). Soft robotics: new perspectives
IEEE Trans. Robot. 28, 457–466. doi: 10.1109/TRO.2011.2172702 for robot bodyware and control. Front. Bioeng. Biotechnol. 2:035003.
Hiller, J., and Lipson, H. (2014). Dynamic simulation of soft multimaterial 3D- doi: 10.3389/fbioe.2014.00003
printed objects. Soft Robot. 1, 88–101. doi: 10.1089/soro.2013.0010 Laschi, C., Cianchetti, M., Mazzolai, B., Margheri, L., Follador, M., and Dario,
Hines, L., Arabagi, V., and Sitti, M. (2012). Shape memory polymer-based flexure P. (2012). Soft robot arm inspired by the octopus. Adv. Robot. 26, 709–727.
stiffness control in a miniature flapping-wing robot. IEEE Trans. Robot. 28, doi: 10.1163/156855312X626343
987–990. doi: 10.1109/TRO.2012.2197313 Laschi, C., Mazzolai, B., and Cianchetti, M. (2016). Soft robotics: technologies
Holland, D. P., Abah, C., Velasco-Enriquez, M., Herman, M., Bennett, G. J., Vela, and systems pushing the boundaries of robot abilities. Sci. Robot. 1:eaah3690.
E. A., et al. (2017). The soft robotics toolkit: strategies for overcoming obstacles doi: 10.1126/scirobotics.aah3690
to the wide dissemination of soft-robotic hardware. IEEE Robot. Automat. Mag. Lawrence, A. B., Alspach, A. N., and Bentivegna, D. C. (2016). “Mechanical
24, 57–64. doi: 10.1109/MRA.2016.2639067 implementation of a variable-stiffness actuator for a softly strummed ukulele,”
Holst, G. L., Teichert, G. H., and Jensen, B. D. (2011). Modeling and experiments in 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems
of buckling modes and deflection of fixed-guided beams in compliant (IROS) (Daejeon), 376–382. doi: 10.1109/IROS.2016.7759082
mechanisms. J. Mech. Design 133:051002. doi: 10.1115/1.4003922 Lewis, J. A. (2006). Direct ink writing of 3D functional materials. Adv. Funct.
Hopkins, J. B., Rivera, J., Kim, C., and Krishnan, G. (2015). Synthesis and analysis Mater. 16, 2193–2204. doi: 10.1002/adfm.200600434
of soft parallel robots comprised of active constraints. J. Mech. Robot. 7:011002. Li, S., Vogt, D. M., Rus, D., and Wood, R. J. (2017a). Fluid-driven origami-
doi: 10.1115/1.4029324 inspired artificial muscles. Proc. Natl. Acad. Sci. U.S.A. 114, 13132–13137.
Howell, L. L. (2001). Compliant Mechanisms. New York, NY: A Wiley-Interscience doi: 10.1073/pnas.1713450114
publication, Wiley. Li, Y., Chen, Y., Yang, Y., and Wei, Y. (2017b). Passive particle jamming
Ilievski, F., Mazzeo, A. D., Shepherd, R. F., Chen, X., and Whitesides, G. M. and its stiffening of soft robotic grippers. IEEE Trans. Robot. 33, 446–455.
(2011). Soft robotics for chemists. Angew. Chem. Int. Edn. 50, 1890–1895. doi: 10.1109/TRO.2016.2636899
doi: 10.1002/anie.201006464 Liu, Y., Boyles, J. K., Genzer, J., and Dickey, M. D. (2012). Self-folding
Ionov, L. (2014). Hydrogel-based actuators: possibilities and limitations. Mater. of polymer sheets using local light absorption. Soft Matter 8, 1764–1769.
Today 17, 494–503. doi: 10.1016/j.mattod.2014.07.002 doi: 10.1039/C1SM06564E
Iqbal, D., and Samiullah, M. H. (2013). Photo-responsive shape-memory Loepfe, M., Schumacher, C. M., and Stark, W. J. (2014). Design, performance and
and shape-changing liquid-crystal polymer networks. Materials 6, 116–142. reinforcement of bearing-free soft silicone combustion-driven pumps. Indust.
doi: 10.3390/ma6010116 Eng. Chem. Res. 53, 12519–12526. doi: 10.1021/ie501991d
Jacobsen, J. O., Winder, B. G., Howell, L. L., and Magleby, S. P. (2009). Lamina Macdonald, E., Salas, R., Espalin, D., Perez, M., Aguilera, E., Muse, D., et al. (2014).
emergent mechanisms and their basic elements. J. Mech. Robot. 2, 011003– 3D printing for the rapid prototyping of structural electronics. IEEE Access 2,
011003–9. doi: 10.1115/1.4000523 234–242. doi: 10.1109/ACCESS.2014.2311810
Jager, E. W., Inganäs, O., and Lundström, I. (2000). Microrobots for micrometer- Maffli, L., Rosset, S., Ghilardi, M., Carpi, F., and Shea, H. (2015). Ultrafast all-
size objects in aqueous media: potential tools for single-cell manipulation. polymer electrically tunable silicone lenses. Adv. Funct. Mater. 25, 1656–1665.
Science 288, 2335–2338. doi: 10.1126/science.288.5475.2335 doi: 10.1002/adfm.201403942
Jin, H., Dong, E., Alici, G., Mao, S., Min, X., Liu, C., et al. (2016). A starfish robot Majidi, C. (2014). Soft robotics: a perspective—current trends and prospects for
based on soft and smart modular structure (SMS) actuated by SMA wires. the future. Soft Robot. 1, 5–11. doi: 10.1089/soro.2013.0001
Bioinspir. Biomimet. 11:056012. doi: 10.1088/1748-3190/11/5/056012 Manti, M., Hassan, T., Passetti, G., D’Elia, N., Laschi, C., and Cianchetti, M. (2015).
Kim, D., and Gillespie, R. B. (2015). “Origami structured compliant actuator A bioinspired soft robotic gripper for adaptable and effective grasping. Soft
(OSCA),” in 2015 IEEE International Conference on Rehabilitation Robot. 2, 107–116. doi: 10.1089/soro.2015.0009
Robotics (ICORR) (Singapore), 259–264. doi: 10.1109/ICORR.2015.72 Marchese, A. D. (2015). Design, Fabrication, and Control of Soft Robots with Fluidic
81209 Elastomer Actuators. Ph.D. thesis, Massachusetts Institute of Technology.

Frontiers in Robotics and AI | www.frontiersin.org 15 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

Marchese, A. D., Komorowski, K., Onal, C. D., and Rus, D. (2014). “Design and Onal, C. D., and Rus, D. (2012). “A modular approach to soft
control of a soft and continuously deformable 2d robotic manipulation system,” robots,” in 2012 4th IEEE RAS EMBS International Conference on
in Robotics and Automation (ICRA), 2014 IEEE International Conference On Biomedical Robotics and Biomechatronics (BioRob) (Roma), 1038–1045.
(IEEE) (Hong Kong), 2189–2196. doi: 10.1109/ICRA.2014.6907161 doi: 10.1109/BioRob.2012.6290290
Marchese, A. D., Onal, C. D., and Rus, D. (2011). “Soft robot actuators Onal, C. D., and Rus, D. (2013). Autonomous undulatory serpentine locomotion
using energy-efficient valves controlled by electropermanent magnets,” in utilizing body dynamics of a fluidic soft robot. Bioinspir. Biomimet. 8:026003.
2011 IEEE/RSJ International Conference on Intelligent Robots and Systems doi: 10.1088/1748-3182/8/2/026003
(San Francisco, CA), 756–761. doi: 10.1109/IROS.2011.6095064 Onal, C. D., Wood, R. J., and Rus, D. (2011). “Towards printable robotics: origami-
Mark, A. G., Palagi, S., Qiu, T., and Fischer, P. (2016). “Auxetic inspired planar fabrication of three-dimensional mechanisms,” in 2011 IEEE
metamaterial simplifies soft robot design,” in 2016 IEEE International International Conference on Robotics and Automation (Shanghai), 4608–4613.
Conference on Robotics and Automation (ICRA) (Stockholm), 4951–4956. doi: 10.1109/ICRA.2011.5980139
doi: 10.1109/ICRA.2016.7487701 Onal, C. D., Wood, R. J., and Rus, D. (2013). An origami-inspired
Martinez, R. V., Branch, J. L., Fish, C. R., Jin, L., Shepherd, R. F., Nunes, R. M., et al. approach to worm robots. IEEE/ASME Trans. Mechatr. 18, 430–438.
(2013). Robotic tentacles with three-dimensional mobility based on flexible doi: 10.1109/TMECH.2012.2210239
elastomers. Adv. Mater. 25, 205–212. doi: 10.1002/adma.201203002 Overvelde, J. T., Kloek, T., D’haen, J. J., and Bertoldi, K. (2015). Amplifying the
Martinez, R. V., Fish, C. R., Chen, X., and Whitesides, G. M. (2012). Elastomeric response of soft actuators by harnessing snap-through instabilities. Proc. Natl.
origami: programmable paper-elastomer composites as pneumatic actuators. Acad. Sci. U.S.A. 112, 10863–10868. doi: 10.1073/pnas.1504947112
Adv. Funct. Mater. 22, 1376–1384. doi: 10.1002/adfm.201102978 Paez, L., Agarwal, G., and Paik, J. (2016). Design and analysis of a soft
Mazzeo, A. D., and Hardt, D. E. (2013). Centrifugal casting of microfluidic pneumatic actuator with origami shell reinforcement. Soft Robot. 3, 109–119.
components with PDMS. J. Micro Nano Manufactur. 1:021001. doi: 10.1089/soro.2016.0023
doi: 10.1115/1.4023754 Park, Y. L., Chen, B. R., and Wood, R. J. (2012). Design and fabrication of soft
Mazzolai, B., Margheri, L., Cianchetti, M., Dario, P., and Laschi, C. (2012). artificial skin using embedded microchannels and liquid conductors. IEEE Sens.
Soft-robotic arm inspired by the octopus: II. From artificial requirements J. 12, 2711–2718. doi: 10.1109/JSEN.2012.2200790
to innovative technological solutions. Bioinspir. Biomimet. 7:025005. Liu, K., Wu, J., Paulino, G. H., and Qi, H. J. (2017). Programmable deployment
doi: 10.1088/1748-3182/7/2/025005 of tensegrity structures by stimulus-responsive polymers. Sci. Rep. 7:1.
Meisel, N., Gaynor, A., Williams, C., and Guest, J. (2013). “Multiple-material doi: 10.1038/s41598-017-03412-6
topology optimization of compliant mechanisms created via polyjet 3d Polygerinos, P., Lyne, S., Wang, Z., Nicolini, L. F., Mosadegh, B., Whitesides, G. M.,
printing,” in 24th Annual International Solid Freeform Fabrication Symposium et al. (2013). “Towards a soft pneumatic glove for hand rehabilitation,” in 2013
an Additive Manufacturing Conference (Austin, TX), 28. IEEE/RSJ International Conference on Intelligent Robots and Systems (Tokyo),
Meisel, N. A., Elliott, A. M., and Williams, C. B. (2015). A procedure for creating 1512–1517. doi: 10.1109/IROS.2013.6696549
actuated joints via embedding shape memory alloys in PolyJet 3D printing. J. Polygerinos, P., Wang, Z., Galloway, K. C., Wood, R. J., and Walsh, C. J. (2015a).
Intell. Mater. Syst. Struct. 26, 1498–1512. doi: 10.1177/1045389X14544144 Soft robotic glove for combined assistance and at-home rehabilitation. Robot.
Memarian, M., Gorbet, R., and Kulić, D. (2015a). “Control of soft pneumatic Auton. Syst. 73, 135–143. doi: 10.1016/j.robot.2014.08.014
finger-like actuators for affective motion generation,” in 2015 IEEE/RSJ Polygerinos, P., Wang, Z., Overvelde, J. T. B., Galloway, K. C., Wood, R. J., Bertoldi,
International Conference on Intelligent Robots and Systems (IROS) (Hamburg), K., et al. (2015b). Modeling of soft fiber-reinforced bending actuators. IEEE
1691–1697. doi: 10.1109/IROS.2015.7353595 Trans. Robot. 31, 778–789. doi: 10.1109/TRO.2015.2428504
Memarian, M., Gorbet, R., and Kulić, D. (2015b). “Modelling and experimental Qi, Q., Teng, Y., and Li, X. (2015). “Design and characteristic study of a
analysis of a novel design for soft pneumatic artificial muscles,” in 2015 pneumatically actuated earthworm-like soft robot,” in 2015 International
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) Conference on Fluid Power and Mechatronics (FPM) (Harbin), 435–439.
(Hamburg), 1718–1724. doi: 10.1109/IROS.2015.7353599 doi: 10.1109/FPM.2015.7337156
Miriyev, A., Stack, K., and Lipson, H. (2017). Soft material for soft actuators. Nat. Ranzani, T., Cianchetti, M., Gerboni, G., Falco, I. D., and Menciassi, A.
Commun. 8:596. doi: 10.1038/s41467-017-00685-3 (2016). A soft modular manipulator for minimally invasive surgery: design
Mosadegh, B., Polygerinos, P., Keplinger, C., Wennstedt, S., Shepherd, R. F., Gupta, and characterization of a single module. IEEE Trans. Robot. 32, 187–200.
U., et al. (2014). Pneumatic networks for soft robotics that actuate rapidly. Adv. doi: 10.1109/TRO.2015.2507160
Funct. Mater. 24, 2163–2170. doi: 10.1002/adfm.201303288 Rateni, G., Cianchetti, M., Ciuti, G., Menciassi, A., and Laschi, C. (2015).
Mu, J., Hou, C., Zhu, B., Wang, H., Li, Y., and Zhang, Q. (2015). A multi-responsive Design and development of a soft robotic gripper for manipulation in
water-driven actuator With instant and powerful performance for versatile minimally invasive surgery: a proof of concept. Meccanica 50, 2855–2863.
applications. Sci. Rep. 5:9503. doi: 10.1038/srep09503 doi: 10.1007/s11012-015-0261-6
Mutlu, R., Alici, G., Panhuis, M., and Spinks, G. M. (2016). 3D printed Robertson, M. A., Sadeghi, H., Florez, J. M., and Paik, J. (2016). Soft pneumatic
flexure hinges for soft monolithic prosthetic fingers. Soft Robot. 3, 120–133. actuator fascicles for high force and reliability. Soft Robot. 4, 23–32.
doi: 10.1089/soro.2016.0026 doi: 10.1089/soro.2016.0029
Nakai, H., Kuniyoshi, Y., Inaba, M., and Inoue, H. (2002). “Metamorphic Roche, E. T., Wohlfarth, R., Overvelde, J. T., Vasilyev, N. V., Pigula, F. A., Mooney,
robot made of low melting point alloy,” in IEEE/RSJ International D. J., et al. (2014). A bioinspired soft actuated material. Adv. Mater. 26,
Conference on Intelligent Robots and Systems, Vol. 2 (Lausanne), 2025–2030. 1200–1206. doi: 10.1002/adma.201304018
doi: 10.1109/IRDS.2002.1044053 Rus, D., and Tolley, M. T. (2015). Design, fabrication and control of soft robots.
Ni, F., Rojas, D., Tang, K., Cai, L., and Asfour, T. (2015). “A jumping Nature 521, 467–475. doi: 10.1038/nature14543
robot using soft pneumatic actuator,” in 2015 IEEE International Russo, S., Ranzani, T., Walsh, C. J., and Wood, R. J. (2017). An additive
Conference on Robotics and Automation (ICRA) (Seattle, WA), 3154–3159. millimeter-scale fabrication method for soft biocompatible actuators
doi: 10.1109/ICRA.2015.7139633 and sensors. Adv. Mater. Technol. 2:1700135. doi: 10.1002/admt.2017
Niiyama, R., Rus, D., and Kim, S. (2014). “Pouch motors: printable/inflatable 00135
soft actuators for robotics,” in 2014 IEEE International Conference Sareen, H., Umapathi, U., Shin, P., Kakehi, Y., Ou, J., Ishii, H., et al. (2017).
on Robotics and Automation (ICRA) (Hong Kong), 6332–6337. “Printflatables: printing human-scale, functional and dynamic inflatable
doi: 10.1109/ICRA.2014.6907793 objects,” in Proceedings of the 2017 CHI Conference on Human Factors
Nishioka, Y., Uesu, M., Tsuboi, H., Kawamura, S., Masuda, W., Yasuda, T., et al. in Computing Systems, CHI ’17 (New York, NY: ACM), 3669–3680.
(2017). Development of a pneumatic soft actuator with pleated inflatable doi: 10.1145/3025453.3025898
structures. Adv. Robot. 31, 753–762. doi: 10.1080/01691864.2017.1345323 Schumacher, C. M., Loepfe, M., Fuhrer, R., Grass, R. N., and
Ogura, K., Wakimoto, S., Suzumori, K., and Nishioka, Y. (2009). “Micro Stark, W. J. (2014). 3D printed lost-wax casted soft silicone
pneumatic curling actuator - Nematode actuator,” in 2008 IEEE monoblocks enable heart-inspired pumping by internal
International Conference on Robotics and Biomimetics (Guilin), 462–467. combustion. RSC Adv. 4, 16039–16042. doi: 10.1039/C4RA0
doi: 10.1109/ROBIO.2009.4913047 1497A

Frontiers in Robotics and AI | www.frontiersin.org 16 July 2018 | Volume 5 | Article 84

Schmitt et al. Soft Robots Manufacturing: A Review

Sen, S., and Awtar, S. (2013). A closed-form nonlinear model for the constraint Tumbleston, J. R., Shirvanyants, D., Ermoshkin, N., Janusziewicz, R., Johnson, A.
characteristics of symmetric spatial beams. J. Mech. Design 135, 031003– R., Kelly, D., et al. (2015). Continuous liquid interface production of 3D objects.
031003–11. doi: 10.1115/1.4023157 Science 347, 1349–1352. doi: 10.1126/science.aaa2397
Shan, W., Lu, T., and Majidi, C. (2013). Soft-matter composites with Vanderborght, B., Albu-Schaeffer, A., Bicchi, A., Burdet, E., Caldwell, D. G.,
electrically tunable elastic rigidity. Smart Mater. Struct. 22:085005. Carloni, R., et al. (2013). Variable impedance actuators: a review. Robot. Auton.
doi: 10.1088/0964-1726/22/8/085005 Syst. 61, 1601–1614. doi: 10.1016/j.robot.2013.06.009
Sharma, B. K., Bongkyun, J., Lee, J. E., Bae, S.-H., Kim, T. W., Lee, H.-J., Verrelst, B., Ham, R. V., Vanderborght, B., Daerden, F., Lefeber, D.,
et al. (2012). Load-controlled roll transfer of oxide transistors for stretchable and Vermeulen, J. (2005). The pneumatic biped “Lucy” actuated
electronics. Adv. Funct. Mater. 23, 2024–2032. doi: 10.1002/adfm.201202519 with pleated pneumatic artificial muscles. Auton. Robots 18, 201–213.
Shepherd, R. F., Ilievski, F., Choi, W., Morin, S. A., Stokes, A. A., Mazzeo, doi: 10.1007/s10514-005-0726-x
A. D., et al. (2011). Multigait soft robot. Proc. Natl. Acad. Sci. U.S.A. 108, Voisembert, S., Mechbal, N., Riwan, A., and Aoussat, A. (2013). Design of
20400–20403. doi: 10.1073/pnas.1116564108 a novel long-range inflatable robotic arm: manufacturing and numerical
Shepherd, R. F., Stokes, A. A., Freake, J., Barber, J., Snyder, P. W., Mazzeo, A. D., evaluation of the joints and actuation. J. Mech. Robot. 5, 045001–045001–9.
et al. (2013). Using explosions to power a soft robot. Angew. Chem. 125, doi: 10.1115/1.4025025
2964–2968. doi: 10.1002/ange.201209540 Wang, L., Brodbeck, L., and Iida, F. (2014). Mechanics and energetics in tool
Shintake, J., Schubert, B., Rosset, S., Shea, H., and Floreano, D. (2015). “Variable manufacture and use: a synthetic approach. J. R. Soc. Interf. 11:20140827.
stiffness actuator for soft robotics using dielectric elastomer and low-melting- doi: 10.1098/rsif.2014.0827
point alloy,” in 2015 IEEE/RSJ International Conference on Intelligent Robots Wang, L., Graber, L., and Iida, F. (2013). Large-payload climbing in complex
and Systems (IROS) (Hamburg), 1097–1102. doi: 10.1109/IROS.2015.7353507 vertical environments using thermoplastic adhesive bonds. IEEE Trans. Robot.
Song, Y. S., Sun, Y., Van Den Brand, R., Von Zitzewitz, J., Micera, S., Courtine, 29, 863–874. doi: 10.1109/TRO.2013.2256312
G., et al. (2013). “Soft robot for gait rehabilitation of spinalized rodents,” in Wang, W., Rodrigue, H., and Ahn, S. H. (2016). Deployable soft composite
Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International Conference structures. Sci. Rep. 6:20869. doi: 10.1038/srep20869
On (IEEE) (Tokyo), 971–976. doi: 10.1109/IROS.2013.6696468 Wang, Y., and Lee, K. (2017). “3D-printed semi-soft mechanisms
SORO (2014). Soft Robotics (SoRo). Mary Ann Liebert, Inc. Publishers. inspired by origami twisted tower,” in 2017 NASA/ESA Conference
Stilli, A., Wurdemann, H. A., and Althoefer, K. (2017). A novel concept on Adaptive Hardware and Systems (AHS) (Edinburgh, UK), 161–166.
for safe, stiffness-controllable robot links. Soft Robot. 4, 16–22. doi: 10.1109/AHS.2017.8046373
doi: 10.1089/soro.2016.0015 Wang, Z., Polygerinos, P., Overvelde, J. T. B., Galloway, K. C., Bertoldi, K., and
Sun, C., Fang, N., Wu, D. M., and Zhang, X. (2005). Projection micro- Walsh, C. J. (2017). Interaction forces of soft fiber reinforced bending actuators.
stereolithography using digital micro-mirror dynamic mask. Sens. Actuat. A IEEE/ASME Trans. Mechatr. 22, 717–727. doi: 10.1109/TMECH.2016.2638468
Phys. 121, 113–120. doi: 10.1016/j.sna.2004.12.011 White, E. L., Case, J. C., and Kramer, R. K. (2017). Multi-mode strain and curvature
Sun, Y., Song, Y. S., and Paik, J. (2013). “Characterization of silicone sensors for soft robotic applications. Sens. Actuat. A Phys. 253, 188–197.
rubber based soft pneumatic actuators,” in 2013 IEEE/RSJ International doi: 10.1016/j.sna.2016.11.031
Conference on Intelligent Robots and Systems (Tokyo), 4446–4453. Wood, R. J., Avadhanula, S., Sahai, R., Steltz, E., and Fearing, R. S. (2008).
doi: 10.1109/IROS.2013.6696995 Microrobot design using fiber reinforced composites. J. Mech. Design
Suo, Z. (2010). Theory of dielectric elastomers. Acta Mech. Solida Sin. 23, 549–578. 130:052304. doi: 10.1115/1.2885509
doi: 10.1016/S0894-9166(11)60004-9 Wu, H., Kuang, M., Cui, L., Tian, D., Wang, M., Luan, G., et al. (2016).
Tao, W., Skorina, E. H., Chen, F., McInnis, J., Luo, M., and Onal, C. D. (2015). Single-material solvent-sensitive actuator from poly(ionic liquid) inverse
“Bioinspired design and fabrication principles of reliable fluidic soft actuation opals based on gradient dewetting. Chem. Commun. 52, 5924–5927.
modules,” in 2015 IEEE International Conference on Robotics and Biomimetics doi: 10.1039/C6CC01442A
(ROBIO) (Zhuhai), 2169–2174. doi: 10.1109/ROBIO.2015.7419095 Yang, D., Verma, M. S., So, J.-H., Mosadegh, B., Keplinger, C., Lee, B., et al. (2016a).
Terryn, S., Brancart, J., Lefeber, D., Van Assche, G., and Vanderborght, B. (2018). Buckling pneumatic linear actuators inspired by muscle. Adv. Mater. Technol.
A pneumatic artificial muscle manufactured out of self-healing polymers 1:1600055. doi: 10.1002/admt.201600055
that can repair macroscopic damages. IEEE Robot. Automat. Lett. 3, 16–21. Yang, Y., Chen, Y., Wei, Y., and Li, Y. (2016b). Novel design and three-
doi: 10.1109/LRA.2017.2724140 dimensional printing of variable stiffness robotic grippers. J. Mech. Robot.
Terryn, S., Mathijssen, G., Brancart, J., Lefeber, D., Assche, G. V., 8:061010. doi: 10.1115/1.4033728
and Vanderborght, B. (2015). Development of a self-healing soft Yap, H. K., Ng, H. Y., and Yeow, C.-H. (2016). High-force soft printable
pneumatic actuator: a first concept. Bioinspir. Biomimet. 10:046007. pneumatics for soft robotic applications. Soft Robot. 3, 144–158.
doi: 10.1088/1748-3190/10/4/046007 doi: 10.1089/soro.2016.0030
Tibbits, S. (2014). 4D printing: multi-material shape change. Architect. Design 84, Yin, L., and Ananthasuresh, G. K. (2003). Design of distributed
116–121. doi: 10.1002/ad.1710 compliant mechanisms. Mech. Based Design Struct. Mach. 31, 151–179.
Tolley, M. T., Felton, S. M., Miyashita, S., Aukes, D., Rus, D., and Wood, doi: 10.1081/SME-120020289
R. J. (2014a). Self-folding origami: shape memory composites activated by Zhao, D. Y., Huang, Z.-P., Wang, M.-J., Wang, T., and Jin, Y. (2012).
uniform heating. Smart Mater. Struct. 23:094006. doi: 10.1088/0964-1726/23/9/ Vacuum casting replication of micro-riblets on shark skin for drag-
094006 reducing applications. J. Mater. Process. Technol. 212, 198–202.
Tolley, M. T., Shepherd, R. F., Mosadegh, B., Galloway, K. C., Wehner, M., doi: 10.1016/j.jmatprotec.2011.09.002
Karpelson, M., et al. (2014b). A resilient, untethered soft robot. Soft Robot. 1, Zhao, H., Li, Y., Elsamadisi, A., and Shepherd, R. (2015). Scalable manufacturing
213–223. doi: 10.1089/soro.2014.0008 of high force wearable soft actuators. Extreme Mech. Lett. 3(Suppl. C):89–104.
Trivedi, D., Rahn, C. D., Kier, W. M., and Walker, I. D. (2008). Soft robotics: doi: 10.1016/j.eml.2015.02.006
biological inspiration, state of the art, and future research. Appl. Bionics
Biomech. 5, 99–117. doi: 10.1155/2008/520417 Conflict of Interest Statement: The authors declare that the research was
Tsagarakis, N., and Caldwell, D. G. (2000). “Improved modelling and assessment conducted in the absence of any commercial or financial relationships that could
of pneumatic muscle actuators,” in Proceedings 2000 ICRA. Millennium be construed as a potential conflict of interest.
Conference. IEEE International Conference on Robotics and Automation.
Symposia Proceedings (Cat. No.00CH37065), Vol. 4 (San Francisco, CA), Copyright © 2018 Schmitt, Piccin, Barbé and Bayle. This is an open-access article
3641–3646. doi: 10.1109/ROBOT.2000.845299 distributed under the terms of the Creative Commons Attribution License (CC BY).
Tsujiuchi, N., Koizumi, T., Shirai, S., Kudawara, T., and Ichikawa, Y. (2006). The use, distribution or reproduction in other forums is permitted, provided the
“Development of a low pressure driven pneumatic actuator and its original author(s) and the copyright owner(s) are credited and that the original
application to a robot hand,” in IECON 2006 - 32nd Annual Conference on publication in this journal is cited, in accordance with accepted academic practice.
IEEE Industrial Electronics (Paris), 3040–3045. doi: 10.1109/IECON.2006.3 No use, distribution or reproduction is permitted which does not comply with these
47883 terms.

Frontiers in Robotics and AI | www.frontiersin.org 17 July 2018 | Volume 5 | Article 84