Inexpensive monolithic additive manufacturing of silicone structures for bio-inspired soft robotic systems

In soft robotics, the fabrication of extremely soft structures capable of performing bio-inspired complex motion is a challenging task. This paper introduces an innovative 3D printing of soft silicone structures with embedded shape memory alloy (SMA) actuators, which is completed in a single printing cycle from CAD files. The proposed custom-made 3D printing setup, based on the material extrusion (MEX) method, was used in conjunction with a cartesian pick and place robot (CPPR) to completely automate the fabrication of thick silicone skins (7 mm) with embedded shape memory alloy actuators. These structures were fabricated monolithically without any assembly tasks and direct human intervention. Taking advantage of the capability to 3D print different geometries, three different patterns were fabricated over the silicone skin, resulting in remarkable dynamic motions: an out-of-plane deformation (jumping of the structure from the x-y plane to the x-z plane) was achieved for the first-time employing silicone skin, to the best of the author’s knowledge. In addition, two process parameters (printing speed and build plate temperature) and the extruded silicone curing mechanisms were investigated to enhance the printing quality. This paper aims to advance the role of additive manufacturing in the field of soft robotics by demonstrating all the benefits that a low-cost, custom-made silicone 3D printer can bring to the table in terms of manufacturing soft bio-inspired structures.


Introduction
Soft robotics is a relatively new field of study that has gained attention in recent years as a result of the capacity to create soft continuum systems with fewer rigid links or fully soft structures [1]. Bio-inspiration is one of the soft robotics' most fundamental pillars. Humans, animals, and other biological entities are capable of performing extremely complex motions by conforming to their surroundings, dispersing stress across a large volume, and modulating their stiffness [2][3][4][5].
From a manufacturing perspective, additive manufacturing (AM) technologies appear to be very attractive for the fabrication of soft robots. Despite the fact that AM is still in its infancy in the soft robotics field, multiple researchers have exploited it at various levels to reap the maximum advantages such as: (i) the ability to 3D print complex structures, (ii) the incorporation of actuators, (iii) the fabrication of intelligent soft robots, and (iv) the reduction of manufacturing steps and assembly tasks [6][7][8][9].
Recently, new advancements in the processability of extremely soft materials motivated many scientists to develop 3D printing setups capable of extruding silicone, using the well-known material extrusion (MEX) approach, to create complex soft structures. printing; in this case silicone printing, (d) Slot created by silicone 3D printing for potential SMA spring embedment, (e) After created the slot printer paused utilizing g-code instructions and SMA spring picked and placed in the slot by custom made pick and place robot (CPPR) utilizing stop n go manufacturing method, (f) Printer started to print over the SMA, (g) SMA spring completely embedded by printing more layers of silicone and traditional silicone skin complete, (h) Patterned silicone skin completed by printing pattern over the tradition silicone skin.

Literature review on mex-based silicone process parameters
Due to the fact that additive manufacturing of silicone structures is an emerging manufacturing technology, the scientific literature on process parameters is scarce. Few studies correlating process parameters to output variables (such as mechanical properties, surface finishes, and dimensional accuracy, among others) have been conducted, especially in comparison to more established AM technologies such as FFF [15,16] and SLA [17].
In this section, the key efforts made to correlate process parameters to printing outputs such as printing accuracy, printing force, intralayer adhesion, and mechanical properties are analyzed critically. Table 1 summarizes the significant process parameters, measurement units, and abbreviations.
Colpani et al [18] studied the effect of three parameters (T , b S p and D r ) on the final accuracy of silicone samples by determining the average width and variability of fabricated structures. They conducted a DoE plan and determined that (i) T b mainly affects the average line width: as T b increases, the silicone's solidification time decreases, resulting in far less material spreading during printing and a higher degree of accuracy in the extruded structure. Furthermore, they found that the highest limit of T b was about 100°C: above that point, nozzle obstruction occurred. (ii) The D r linearly influences the average width value, confirming that the material solidification is under control and (iii) The three process parameters analyzed by the authors have no impact on the variability of the width. Plott et al [19] studied a crucial aspect in extrusion-based silicone: the relationship between process parameters and voids (number, dimension, and effect on tensile strength). They observed that (i) when X is equal to 1 and 0.97 and I a is equal to +/− 45°and 90°, high level of tensile strength is obtained; even though some small voids are present when X is equal to 0.97, high tensile strength was maintained because the voids were able to orient themselves along the direction of force. (ii) The specimens with 0°infill exhibited the lowest tensile strength due to the interior tangency voids. Similar considerations have been highlighted in Miriyev et al [20] who correlated infill angle( I a ) to mechanical properties and silicone formulation. They found out that maximum strain was reached when I a is equal to 90°, while the maximum tensile force was reached by setting I a equal to 0°.
In an intriguing study, Walker et al [21] examined the relationship between Curing temperature( T c ) and intralayer adhesion using a peeling test. This work is interesting because it analyzes intralayer adhesion in this manufacturing field for the first time. Inadequate crosslinking between layers can result in anisotropy and premature failure. This is a significant issue because this manufacturing technology is frequently used to fabricate soft robots, prosthetic and rehabilitation devices which are frequently actuated by compressed air, with a long cycle life and the ability to withstand specific loads. The key to increasing the strength of silicone structures is to maximize the interfacial adhesion between two printed layers. In theory, when a printed layer cures, the number of cross-linkable groups available for bonding to the subsequent layer lowers. Their work results in a negative correlation between T c and intralayer adhesion: as T c increases, intralayer adhesion (and thus part strength) decreases. Consequently, in large-scale 3D printing with silicone, a trade-off must be considered: if the layers are over-cured, the structure's tensile strength will decrease; if they are under-cured, tall structures cannot be printed due to the structure collapsing.
The topic of large-scale extrusion-based silicone is extremely appealing since it has the potential to pave the way for molding replacement, resulting in an automated process that requires few human interventions and saves time and money. Plott et al [22] established a precedent in this field by correlating process parameters to additive manufacturing forces that occur during fabrication. The ability to fabricate tall and thin structures requires an understanding of how to minimize extrusion forces. They evaluated three process parameters: h d , , and Q in a parametric study that are linked to normal force (F n ) and tangential force (F t ). They determined that, depending on the process parameters used, four different configurations of printing forces can occur (see figure 2).
Plott et al, emphasized that the optimal configuration for minimizing printing forces is the one depicted in figure 2(a). In fact, some observations are: (i) F t increases if its major component F tn (tangential force caused by the nozzle dragging through the deposited silicone) increases as well, resulting in an order of magnitude increase in F t when present, (ii) F n increases when the nozzle comes into contact with the recently extruded filament and F nn (the normal force caused by the normal interaction between the nozzle and the extruded silicone) increases F n by an order of magnitude when present, (iii) to reduce the total printing force ( ) + F F F t n it is necessary to reduce flow rate (Q) and increase h and this conclusion is consistent with the work of Percoco et al [23], who established a relationship between the h parameter and extrusion force for FFF technology. iv) When a small nozzle (d) is used, the force-deflection ratio becomes more favorable: the force-deflection ratio also becomes less dependent on Q and t. A summary of each cited work on silicone modeling is provided in table 2.
The state-of-the-art process parameters in silicone extrusion-based additive manufacturing motivated the authors to study two aspects of a manufacturing technology that is currently underexploited.
• How do the process parameters affect the accuracy of a single 3D printed bead? Answering this question allow to fabricate more accurate structures, allowing the integration of external actuators (namely SMA) using pick and place robots.
• The relationship between the amount of heat generated by the build plate and the maximum height of 3Dprinted structures in order to fabricate tall soft robots.

Custom-made 3D printer setup
The current study demonstrated an inexpensive custom-made 3D printer based on material extrusion technology to create extremely soft silicone structures with embedded shape memory alloys (SMAs) actuators capable of complex motions. Specifically, using the proposed setup, two classes of silicone structures were fabricated: (i) the traditional one with uniform surface and capable of motions recalling several animals, and (ii) the patterned one capable of out of plane motions. Noe that silicone structures with different surface patterns are difficult to be manufactured exploiting the traditional fabrication approach or it may require several molding   [21] has been modified further with the addition of the following features: (i) A stepper motor was installed in the machine's top section which was connected to a wooden support structure.
(ii) A system consisting of two gears connected to the motor shaft and a lead screw equipped with a custom made 'pushing' part was assembled. The assembly is connected to the lead screw to convert the stepper motor motion to a linear motion capable of pushing the syringe containing silicone material and obtaining the silicone extrusion.
(iii) A syringe holder is attached to the wooden support to hold the syringe, lead screw, gears, and stepper motor.
(iv) A silicone-filled syringe reservoir is connected to a plastic tube by a plastic connector (Female Luer x 1/8' hose barb adapter), which is connected to a terminal calibrated plastic nozzle via a plastic connector (Male Luer Lock 1/8' hose barb adapter).
(v) An inexpensive additional heating system consisting of a ceramic cartridge heater, an aluminum block, a heating fan, and a temperature sensor has been incorporated beside the nozzle to heat 3D Printed structure from the top. (See figure 3(d))

Figures 3(a)-(c)
shows the above-mentioned 3D printing setup. In addition, the maximum sizes of silicone structure that can be printed in the setup are shown, which 170 mm is size.

Material preparation
For the study in this paper, Ecoflex 00-10, a dual-part silicone (part A and part B) material, was used in this research. In [10], it is proved that parts manufactured using Ecoflex 00-10 are characterized by an extremely high degree of flexibility and elongation at break equal to 1260%. Therefore, we focused on this silicone. It is worth mentioning that other silicones (Ecoflex 00-20, 00-30, Dragon Skin series etc.) based on the same curing mechanism can be employed with the proposed setup. In table 3, the most important characteristics of Ecoflex 00-10 (from the material data sheet) are listed.
After pouring equal amounts of Parts A and B into the mixing container (1A:1B volume or weight ratio), they were thoroughly stirred and mixed for at least 3 min. The silicone was poured into a 60 ml syringe and connected to the 3D printed setup following the mixing procedure. Avoiding air entrapment and bubble formation inside the syringe is critical throughout the mixing procedure. Using vacuum degassing, air bubbles were removed from the syringe; otherwise, the pressure created by the air would have caused the syringe to break.

3D Printing large structures
An important aspect of silicone 3D printing is the ability to create large and tall structures. While sections 3.2 and 4 analyze the feasibility of fabricating tall structures, which is a challenging topic in this field, this section examines the capabilities of the custom-made setup in terms of large structures. The custom-made 3D printer's build plate dimension is 200 mm × 200 mm. To demonstrate the proposed setup potential for large structures, two distinct structures consisting of only two extruded layers (layer height equal to 0.8 mm for an overall structure height of 1.6 mm) were fabricated. As illustrated in figures 3(e)-(f), a rectangular shape with two identical 175 mm sides and a circle with a 175 mm radius were 3D printed using a 0.4 mm nozzle. It was found that the maximum size of structures that can be manufactured in this 3D printing system is 175 mm × 175 mm owing to setup constraints (build plate dimension and nozzle holder) while the build player temperature was set at 40°C throughout the process. This indicates that larger structures can be fabricated if a 3D printing machine with a larger build plate had been utilized. To summarize, the size of the structures that can be printed (along the x and y axes) is fully dependent on the build plate and nozzle holder.

Process parameters study
4.1. Printing accuracy One of the most important relationships to study in silicone additive manufacturing is the correlation between process parameters and final part accuracy. This paper examines two process parameters: build plate temperature, abbreviated as T b (°C), and printing speed, abbreviated as S p (mm s) −1 . There were three levels of variation for each process parameter: low (50°C and 10 mm s) −1 , medium (75°C and 20 mm s −1 ), and high (100°C and 30 mm s −1 ). A −1 factorial plan with 3 repetitions as shown in table 4 was computed (see Supplementary S1 for details) for printing a single layer, single line silicone bead (see figures 4(a), (b)) and two outputs were experimentally measured: bead width and bead height. For the measurements, an optical benchtop microscope (PSM1000, Motic), equipped with a vertically positioned camera (Moticam 3+, Moticom) for photographing the magnified sample, was employed (see Supplementary S1). The most important results in terms of bead width and bead height are shown respectively in figures 4(c)-(d). With regards to the bead width investigation, the following conclusions can be drawn: , and 100°C, respectively. This conclusion is particularly intriguing when compared to its FDM counterpart. For FDM technology, it is well known from literature [23] that there is a negative correlation between printing speed and accuracy; in this case(DIW), the correlation is positive. To understand why this behavior occurs, it is imperative to analyze the setup and take into consideration that the flow value was constant for every experiment. The silicone inside the syringe was pushed by a mechanism activated by a stepper motor and let it flow through a PETF tube and finally into the plastic nozzle (see figures 3(a)-(b)), both of which are vertically adjacent to the reservoir silicone syringe. Gravity plays a significant part in this process, as silicone is less viscous than melted plastic filaments. Even though the volumetric flow is constant for each printing speed, when a low printing speed is being used, a greater amount of silicone flows out of the nozzle due to the gravity effect. The printing speed mitigates the unwanted extra silicone generated by gravity in the following way: when the build plate temperature is set to 50°C, 75°C, and 100°C, switching from the low level (10 mm s) −1 to the medium level (20 mm s −1 ), the sample width is reduced by 45%, 31%, and 42%, respectively. The sample width is reduced by less than 5%, 13%, and 7.6%, respectively, when the build plate temperature is 50°C, 75°C, and 100°C, while the speed is increased from medium (20 mm s) −1 to high (30 mm s −1 ).
• The build plate temperature influences the sample line width. The T b affects the width of the sample in the following way-when the best printing speed (namely 30 mm s) −1 in terms of accuracy is set: switching from 50°C to 75°C and from 75°C to 100°C, the sample width is reduced from 10.5% and 2.48% respectively. It can be observed that increasing the build plate temperature results in the abrupt decrease in the curing time of the silicone. This means that the expansion of the recently extruded silicone is reduced as well because as soon as the silicone flows out of the nozzle, it gets cured. This behavior is more noticeable when the build plate temperature is increased from 50°C to 75°C than when it increased from 75°C to 100°C. With regards to the sample height, the following conclusion can be pointed out (see figures 4(e), (f)).
• When the build plate temperature is set to 50°C, the printing speed does not affect the sample height, which is nearly 50 μm (150 μm less than the desired height). The low values of layer height obtained when the build plate temperature is set to 50°C (and the insensitivity to print speed) could be due to the material less heat to get instantly cured. In fact, the line width is significantly larger when the build plate temperature is set to 50°C, resulting in a decrease in the layer height. When the build plate temperature is 75°C and 100°C, the printing speed produces an interesting effect: as the printing speed increases, the layer height decreases. Because the amount of silicone extruded at a printing speed of 10 mm s −1 is greater than at 20 mm s −1 , which is greater than at 30 mm s −1 due to the gravity effect, and since the build plate temperature is high enough to instantly cure the extruded silicone, the measured layer height is greater when a low printing speed value is set.
In summary, the cross-section study yielded the following significant findings: Gravity has a significant effect on the precision of silicone extrusion-based 3D printing (Ecoflex 00-10). If the printing speed is too slow (10 mm s −1 ), the amount of silicone that flows out of the nozzle becomes excessive, resulting in a 'collapse' (a bigger line width than the one set and a smaller height than the one set). The 20 and 30 mm s −1 provide a better outcome in terms of width accuracy, while only the 20 mm s −1 provides a good result in terms of height accuracy. In terms of build plate temperature, both the medium (75°C) and high (100°C) values produce nearly identical results in terms of width accuracy and layer height. In conclusion, among the parameters tested, the ones that should be selected to achieve a suitable balance between width and height accuracy are: (i) printing speed set at 20 mm s −1 and, (ii) build plate temperature set at 75°C.

Heating effect for tall structure
Extrusion-based additive manufacturing techniques utilize a heated build plate to cure the extruded silicone, thereby enabling the fabrication of multilayer structures. This curing method imposes a major limitation on the maximum height of the structure: the heat supplied by the build plate is inadequate to cure the taller silicone structure layer after layer. Therefore, the heated build plate method cannot be employed to fabricate the tall 3Dprinted structure.
The current section examined the drop in heating after each layer using an infrared thermal camera (see supplementary S2 for more details) at different build plate temperatures. The printed sample is depicted in figure  S4(a), and it has a rectangular shape with a square length of 25 mm and 8 mm height. It was determined arbitrarily to use a 0.4 mm nozzle and a layer height of 1 mm for the slicing; in this manner, every manufactured layer corresponds to 1 mm. The set process parameters and the plots with infrared images are shown in Supplementary S2. The experiment was performed at two different build plate temperatures (55°C and 70°C) and the linear regression equation (equation (1), and 2) determined from experimental data are: for the 70°C build plate and this slight drop in temperature can be attributed to room conditions. Indeed, the custom-built setup is based on an open-chamber design, making it sensitive to its surrounding environment. In the experiment with the 55°C build plate temperature, the structure began collapsing at the 7th layer (after reaching 7 mm of height). When the temperature read from the thermal camera reached 44°C, it collapsed completely, and the temperature at the next layer (8th layer) was 43.2°C. Furthermore, when 70°C was set as the build plate temperature, it was not possible to create an 8 mm tall structure in this situation (not due to the structure collapsing, as was the case with the 55°C build plate temperature), since the silicone was cured inside the tube-nozzle due to the build plate's high temperature. Around the 7th layer, the silicone began to cure; indeed, as illustrated in figure S4(d), the final printed layer (8th) is affected by the under-extrusion problem.
To address this issue, the tube-nozzle assembly was thermally insulated using a commercially available thermal insulator spray (Loctite Insulating Spray Foam), as illustrated in Supplementary S2 (figure S4(c)). Using the new insulated pipe, we fabricated taller structures: setting the build plate temperature at 70°C, 18 mm tall structure was fabricated, finding a temperature drop of 1.51  .
mm C In this case, at the 18th layer when the measured temperature over the recently extruded layer was around 44°C the structure collapsed, as expected. The advantage of the proposed approach consists of the possibility to increase the height of 3D printed silicone (Ecoflex 00-10) structures from 7 mm to 17 mm. It is important to point out that higher build plate temperatures than 70°C were found to be not suitable for the following application. When the higher temperature (from 75°C to 100°C) is set, the first layer gets cured too fast and adhesion between the first one and second one is not good enough, involving accuracy problems. For this reason, we claim that using only the build plate temperature as a heating source to cure the extruded silicone (Ecoflex 00-10), the maximum height of structure that can be reached without any collapse is around 17 mm (if the nozzle-tube system is thermally insulated).
To overcome curing issue, an additional heating source was added near the calibrated nozzle ( figure 3(d)), to locally cure the recently extruded silicone. In this way, tall structures of almost 30 mm were fabricated, showing that with the proposed method, the only constraint is related to the curing process inside the syringe. After 2 h and 20 min the silicone starts getting cured. The whole experiment is discussed in Supplementary S3.

Application in soft robotics: silicone skin with embedded sma spring
There has been a lot of interest in the fabrication of bio-inspired soft structures with embedded actuators [24] [25] due to the ability to achieve several types of movements for various applications. In the state of the art, silicone is poured into a mold (typically a 3D printed mold) having certain slots for the manual integration of actuators such as shape memory alloy (SMA) in the form of wires or springs [26][27][28][29][30][31][32][33].
For the first time, the whole process of fabricating a silicone skin has been automated in this research article, using the proposed 3D printer setup and a custom-made cartesian pick and place robot (CPPR), as shown in figure 5(a) (see Supplementary S4). The stop and go method [34] was employed in this fabrication. First, the silicone print was paused through g-code instructions. Then the SMA spring was placed into the manufactured channel utilizing the CPPR, and after that, the print was resumed (see Video S1, https://youtu.be/MkV-E_ n4kzk).
The usage of the process parameters studied in section 4 (i.e., printing speed and build plate temperature) led to the fabrication of accurate structures. In this way, SMA spring actuators were successfully embedded into the channel when the printing was stopped.
The dimensions of the proposed structures and the most relevant process parameters are listed in table 5 (more information at supplementary S4. SMA springs (Dynalloy, USA) were employed, and the properties are listed in table S5. It is important to point out that the austenitic start (A s ) temperature of the spring is equal to 90°C, while the constant heat provided locally by the additional heating element was 50°C, preventing the SMA activation during the manufacturing process. Using the following approach, it was possible to automate the integration of SMA springs into 3D printed soft structures during the fabrication process. This goal has never been achieved in the literature because: (i) the melting temperature of the filaments used in material extrusion processes is higher than A , s and (ii) silicone additive manufacturing is still in its early years. Two types of silicone skins have been manufactured: traditional and patterned, as shown in figures 5(b), (c). Three different skins were fabricated for the traditional group: 'Middle' (an SMA positioned in the middle), 'Diagonal' (an SMA positioned diagonally), and 'Dual' (two SMA positioned linearly). For the patterned group, a single SMA was placed at the center of each structure and three different patterns (0.4 mm tall) were fabricated for each silicone skin: 'Chess', 'Crown', and 'Three Lines'. Both the versions were characterized using the same measurement setup ( figure 5(d)).
The aim of the proposed artificial skin manufacturing is multipurpose: • Improve the traditional manufacturing process for the artificial silicone skin by increasing the degree of automation.
• Demonstrate that the proposed 3D printing custom-made setup is capable of fabricating patterned structures that are difficult to fabricate employing traditional fabrication methods, without mold and requiring long fabrication time.
• Demonstrate capability of manufacturing structures performing complex motions and mimicking animals with the proposed inexpensive setup, reducing the investment cost.
Three different current inputs (3A, 3.5A, and 4A) were used to examine the motions in the x-y plane for each silicone skin. The identical testing protocol was utilized for traditional and patterned silicone skins. Each current input was applied for 2 s followed by a 20 s off period; the cycle was repeated ten times in total and standard deviation also calculated. Although there exist several more sophisticated approaches to achieve a more precise activation of SMA, for the scope of this work, a simple activation with direct current has been preferred to guarantee complete activation and verify the deformed shape of the silicone structure. The silicone skin characterization demonstrates that extremely high repeatability is achievable, with a standard deviation of less than 0.1 mm per point (each marker) and all graphs representing the average of ten measurements. Regarding the three silicone skins belonging to the traditional group (see Video S2, https://youtu.be/FqSsx6s19Gk), increasing the applied current results in an improvement in the silicone skin motions, which mimic the animal domain. When a 4A was applied, the Middle, Diagonal, and Dual (applied current to both the SMA actuators) structures recalled the elephant trunk, cuttlefish, and caterpillar behavior, respectively (see figures 6(a)-(d)). In Supplementary S4 further data is provided. Figure 6(a) show the trajectory of the points in the skin samples at different input currents to SMA, showing the final morphed shape in the X and Y planes.
Regarding the patterned structures (see Video S3, https://youtu.be/s__VNIZ07zo), the main outcomes for each structure are summarized here: • When 3 A is applied to the 'chess' structure, a slight deformation occurs (mostly on the left side of the structure). Increasing the applied current to 3.5 A, results in a completely different motion. The silicone skin keeps contact with the base only on the left and right sides and is elevated by nearly 15 mm from the base.  • For the '3 lines' structure, depending on the applied current, radically different movements are obtained. By applying 3A, a deformation mimicking the inchworm is achieved (see figures 7(a)-(c)). When 3.5 A and 4 A currents are applied, respectively, an out-of-plane deformation occurs while the skin base is in contact with the flat base jumps from the x-y to the x-z plane. To the author's knowledge, this outcome has never been shown in silicone skin and is only achievable because of the additional pattern fabricated by employing the 3D printer. Figure 7(e) illustrates the leap dynamic (t = 0 s, t = 1 s, t = 1.5 s, and t = 2 s). This motion is also relatively rapid.
• For the 'Crown' structure, no motion was observed at 3A. When the current was increased to 3.5 A, a bending deformation was achieved (see figures 7(a)-(d)), while at 4 A, an out-of-plane motion (from x-y plane to x-z plane) was obtained as shown in figure 7(d).
The achieved results demonstrate how soft robotics can benefit from silicone-based material extrusion combined with a pick and place robot capable of embedding active actuators during the manufacturing process autonomously without direct human intervention. Future work will focus on mathematical modeling of patterned structures to predict their motions during the fabrication process. Numerous intriguing applications are feasible with the proposed 3D printed artificial skins, including soft grippers ('Middle' at 4A), walking robots ('Chess' at 4A), a soft robot capable of carrying objects ('Chess' at 3.5 A), and jumping robots ('3 lines' at 3.5 A and 4 A). Figure 8 depicts cyclic actuation results of the SMA spring-based silicone skin, obtained by tracking the vertical displacement over time. For calculating vertical displacement with time changes, one of the traditional silicone skins (middle) and one of the patterned silicone skins (chess) were actuated. Actuation experiment was conducting using programmable DC power supply shown earlier, a current input of 4A was applied for 2 s heating period followed by a 20 s cooling period for silicone skin (f = 0.046 Hz) and the cycle was repeated five times. The result demonstrates high repeatability, and the rise time (Tr) was so fast, and the cooling time (fall time, Tf) is relatively longer because of the 20 s cooling period. Referring to the width of the vertical displacement, the patterned sample can potentially actuate at a period of 4.8 sec (0.25 Hz) frequency. The manufacturing method discussed in this work represents the first attempt to create functionalized soft robotic structures with embedded actuators, taking full advantage of the silicone 3D printing technique. As a proof of concept, plain and patterned structures have been 3D printed and characterized to determine their behaviors. Future work will focus on the fabrication of more complex structures from a design standpoint, to replicate animals such as octopus, jellyfish, and starfish. The usage of some artifacts for critical 3D printing issues (i.e., supports and hollow structures) will enable the fabrication of non-conventional geometries by employing the proposed manufacturing approach.

Conclusions
In this paper, a custom-made inexpensive 3D printer capable of extruding silicone Ecoflex 00-10 through a calibrated nozzle is presented and characterized; specifically, circular, and square objects with maximum dimensions of 175 mm are fabricated. The two process parameters (printing speed and build plate temperature) were studied and correlated with the dimensional accuracy of single-layer printed beads, discovering an interesting relationship: increasing the printing speed improves the accuracy, which is the opposite of the behavior observed with FFF technology. The authors attribute this behavior to the gravity effect that occurs during printing. Due to gravity, more material flows out of the nozzle at lower printing speeds, resulting in a decrease in accuracy. Another critical component is the silicone curing mechanism. The conventional technique of curing 3D printed silicone relied on the heat generated by the build plate, which permitted the construction of structures up to 17 mm in height (under certain conditions, such as the use of thermal insulation for the system nozzle-tube). Another curing technique employs installing an additional local heating source from the top to provide heat directly to the recently extruded silicone, which results in structures with a maximum height of 30 mm. Finally, the proposed custom-made 3D printer was employed to manufacture multiple types of 'silicone skin' with embedded SMA actuators in a single manufacturing step autonomously without direct human intervention. The whole manufacturing process was automated using a cartesian pick and place robot that embedded the SMA using the stop and go approach. The proposed structures were tested at different current inputs and demonstrated complex motions that recalled the behavior of several animals, including an elephant trunk, cuttlefish, and caterpillar; additionally, the patterned structures demonstrated unique behavior and outof-plane deformations. The present work paves the way for the enormous exploitation of silicone-based printing in soft robotics, especially for the fabrication of bio-inspired structures which can potentially be fabricated monolithically, resulting in cost, fabrication time, and assembly step reductions.

Data availability statement
The data generated and/or analysed during the current study are not publicly available for legal/ethical reasons but are available from the corresponding author on reasonable request.