Edible Soft Actuators Based on Konjac Glucomannan for Underwater Operation

Soft robots are being increasingly developed and deployed for underwater applications such as exploration, monitoring, and rescue owing to their innate compliance, mechanical and chemical stability, and waterproof properties. However, they are still predominantly manufactured using silicones and acrylic elastomers, which pose environmental risks due to their limited biodegradability. To tackle this issue, a method for manufacturing water‐resistant, liquid‐driven soft actuators made from a novel type of konjac glucomannan (KGM) is presented, which is plant‐derived, insoluble, and edible. The material offers impressive stretchability, reaching up to 67.2% of its initial length, surpassing previous related work, and withstands over a 1000 cycles of tensile stress. Using this material, we develop soft actuators that can bend up to 47.1°, and the maximum blocking force of 0.1 N is achieved with a pressure input of 20 kPa. A gripper that is capable of grasping various objects underwater is also evaluated. In essence, this article introduces a sustainable and environmentally friendly alternative to conventional soft robotic materials, paving the way for innovative and eco‐conscious underwater applications. The utilization of edible, water‐resistant materials like konjac provides a promising solution to reduce pollution and mitigate negative environmental impacts associated with traditional underwater soft robots.


Introduction
Soft robots offer promise for diverse applications in aquatic environments, such as exploration, monitoring, and rescue. [1,2]esearchers have developed underwater soft robots in variety of forms, including grippers, [3,4] fish, [5][6][7][8] octopus, [9] jellyfish, [10] ray, [11] and even a school of small autonomous robotic fish. [12]uch soft robots aim to interact effectively with the surrounding environment, by displaying animal-like mobility, efficiency, and adaptability.[15] Ironically, these properties become problematic near the end-oflife stage of these robots, preventing degradation and potentially causing environmental pollution.Tiny particles of silicone and acrylic polymer are already found inside the bodies of aquatic animals, [16,17] with harm caused by these particles actively being studied.As such, there is a growing interest in the development of soft robots using more sustainable and environmentally friendly materials.For example, soft robotic devices have been made of gelatin or candy, [18][19][20][21][22][23][24][25] but these materials are highly soluble in water and therefore unsuited for operation in aquatic environments.Quite recently, a water insoluble, millimeter-scaled, 3 d printed edible actuator made of a marine-sourced hydrogel was reported. [26]Building on these advancements, there remain certain limitations in the fabrication methods of previous studies.For instance, the fabrication process of the reported marinesourced hydrogel actuator involved the use of hydrochloric acid, potentially introducing safety concerns.In stark contrast, our research prioritizes safety by exclusively utilizing materials deemed safe for consumption.A comparison of the quantitative characteristics (e.g., size, force output, etc.) between ref. [26] and our approach is provided in Table S1, Supporting Information.
Here we present a plant origin, water-resistant soft actuator made of edible materials that can be reliably operated underwater, with the potential to replace conventional non-degradable materials typically used in underwater soft robotics.Our actuator is made of cross-linked konjac glucomannan (c-KGM).Konjac glucomannan (KGM) is a dietary fiber extracted from the tubers of Amorphophallus konjac. [27]KGM has been widely consumed in China, Japan, and Southeast Asia as a food source and a component in traditional medicine. [28]Its high dietary fiber content promotes satiety and supports digestive health, making KGM a valuable food source for both land-based and aquatic animals. [29]oreover, the low toxicity of KGM reduces potential harm to humans and aquatic ecosystems, making it suitable for DOI: 10.1002/aisy.202300473Soft robots are being increasingly developed and deployed for underwater applications such as exploration, monitoring, and rescue owing to their innate compliance, mechanical and chemical stability, and waterproof properties.However, they are still predominantly manufactured using silicones and acrylic elastomers, which pose environmental risks due to their limited biodegradability.To tackle this issue, a method for manufacturing water-resistant, liquid-driven soft actuators made from a novel type of konjac glucomannan (KGM) is presented, which is plant-derived, insoluble, and edible.The material offers impressive stretchability, reaching up to 67.2% of its initial length, surpassing previous related work, and withstands over a 1000 cycles of tensile stress.Using this material, we develop soft actuators that can bend up to 47.1°, and the maximum blocking force of 0.1 N is achieved with a pressure input of 20 kPa.A gripper that is capable of grasping various objects underwater is also evaluated.In essence, this article introduces a sustainable and environmentally friendly alternative to conventional soft robotic materials, paving the way for innovative and eco-conscious underwater applications.The utilization of edible, water-resistant materials like konjac provides a promising solution to reduce pollution and mitigate negative environmental impacts associated with traditional underwater soft robots.
eco-friendly underwater soft robots.Its biodegradability enables decomposition within a couple of weeks in seawater (see Figure S1, Supporting Information), offering a substantial ecological advantage over conventional non-degradable materials.At the same time, our actuator samples retained their shape well in distilled water for more than 4 months (Figure S2, Supporting Information).
c-KGM exhibits larger elongation at break (53%) [30] than other water-resistant edible materials (refer Table S2, Supporting Information), such as oleogel (15.3%), [31] sodium alginate (14%), [32] and ethyl cellulose (0.8%). [33]Owing to this property, c-KGM has been used in several stretchable devices, such as skin protectants, [34] sponges, [35] humidity sensors, [36] and stretchable gels. [37]However, the fabrication processes of the aforementioned examples involved toxic chemicals, resulting in reduced edibility and biocompatibility.In addition, the Young's moduli (E) of previous c-KGM-based materials was on the order of MPa.[40] Here we describe a new fabrication method to develop softer c-KGM materials (i.e., E = 60-80 kPa) without the use of any toxic chemicals.The resulting material properties are thoroughly characterized by tensile and water absorption tests.We use the proposed method to develop a soft gripper and characterize its performance.The results describe a soft edible gripper that can be reliably operated underwater, and the proposed method could be used to manufacture a variety of other underwater edible robots that can be safely ingested by animals or degrade in the environment.

Results and Discussion
The structure of our c-KGM actuator (Figure 1a: see also Supplementary Video S1, Supporting Information) is based on PneuNets, a widely used actuator design in soft robotics. [41]gure 1.Edible soft actuator that can be operated underwater, where P stands for gauge pressure.a) The actuator remains straight in an unpressurized state (P = 0).In a pressurized state (P > 0), the actuator bends due to the strain difference between its bottom layer and the chamber.b) The actuator is made of c-KGM (main component) and KC as an additive.c) Grasping experiment of a kiwi that weighs 128.4 g (i: gripper in position, ii: approaching to kiwi, iii: grasping upon pressurization, iv: lifting the kiwi).d,e) Grasping experiment with different objects (d: plum, 96.8 g, e: clam, 15.2 g).
The actuator consists of an inflatable (chamber) part and a strain limiting layer for bending motion.Each part is bonded together to prevent delamination.The chamber part consists of c-KGM and the bottom part (strain limiting part) consists of c-KGM with an additive (see following paragraph) to ensure stiffness and bonding strength.In Figure 1a, the actuator is aligned vertically with the gravitational force.In an unpressurized state, it maintains a straight alignment (P = 0).Upon pressurization (P > 0), a liquid (distilled water) flows into the inner channel and expands the chamber.
Normally, c-KGM is fabricated by the gelatinization of KGM with an alkali additive, [27] such as sodium bicarbonate (SB). [30,42,43]However, the stretchability of previously developed SB-treated c-KGM was limited.In this study, we significantly improve the stretchability by changing the material composition ratio.As depicted in Figure 1b, heating the mixture of KGM and SB triggers gelation and leads to formation of KGM micelles (illustrated as purple balls) within a network of intertwined fibers (black curved lines). [44]For the bottom part of the actuator, kappa carrageenan (KC) was added to KGM and SB during crosslinking to afford KC/c-KGM.This procedure ensures secure bonding strength to c-KGM due to the hydrogen bonding with KGM molecules, which contributes to fiber formation (refer to Figure 1b). [45]The actuator was subsequently employed to assemble a robotic gripper, capable of grasping various objects, such as fruits and clams (Figure 1c-e).

Fabrication and Characterization of c-KGM and KC/c-KGM
The mechanical properties of c-KGM can be tuned by adjusting material ratios, [44,46,47] but to date there is no method to manufacture c-KGM that meets mechanical properties required for soft aquatic robots.Here, we study a method and material ratios needed to fabricate c-KGM and KC/c-KGM with suitable properties for soft robotic applications (Figure 2a).Although refs.[44,46,47] commonly used KGM as one of the material ingredients, all the other materials were different from our work.Please refer to c-KGM and KC/c-KGM gel fabrication in Section 4 for more details about the fabrication process.
The properties of c-KGM and KC/c-KGM were measured for water solubility, water absorption, and tensile strength.In the case of water solubility and absorption tests, c-KGM and KC/ c-KGM cuboid samples were prepared (refer to the water solubility and absorption test in Section 4).The samples were immersed from 1 to 7 days (refer to Figure 2b-c), and their water solubility and absorption properties were evaluated every 24 h.Both the cubed samples and resultant supernatant (i.e., the solution leftover after soaking and removal of solids) were dried in an oven to evaporate internal moisture.The water solubility index (WSI) was calculated as follows: [49][50][51] WSI ¼ M solids in supernatant M solids before immersion Â 100 Where M solids in supernatant is the mass of the solids (c-KGM or KC/ c-KGM) left over after evaporating water from the supernatant.M solids before immersion is the initial mass of the cubes of c-KGM or KC/c-KGM before immersion.The water absorption index (WAI) was calculated based on ISO 1817 [52] and expressed as follows: where M solids change by immersion is the mass difference between the cubes of c-KGM or KC/c-KGM before and right after immersion.In the case of c-KGM (Figure 2b), its WSI values are around 0.3% for all 7 days.This indicates that our c-KGM is almost insoluble in water.The WAI values measured over 7 days remain stable at 27% for c-KGM and 35% for KC/c-KGM (Figure 2c), which implies initial water absorption and likely volumetric swelling.
Although the KC/c-KGM was insoluble, it clearly absorbed more water than the c-KGM.With the addition of KC to KGM, the number of hydrogen bonding sites within the matrix structure of the gel is increased, which is likely to facilitate better water retention within the gel matrix.Next, we measured the effect of water immersion on the tensile properties of c-KGM and KC/c-KGM.From each of the three batches of dumbbell shaped specimens (Type 1A as per ISO 37:2017, [53] Figure S3, Supporting Information), one sample was immersed in water for different lengths of time (1, 2, and 3 days).Please refer to the Tensile test in Section 4 for more details about the specimen preparation.The average elongation at break (EB) of c-KGM (Figure 3a) decreased gradually over immersion time: 67.2 AE 2.2% (1 day), 63.0 AE 0.5% (2 days), and 60.3 AE 1.0% (3 days).Similarly, Young's modulus (E) slightly decreased with increasing immersion time: 77.4 AE 6.5 kPa (1 day), 76.4 AE 9.3 kPa (2 days), and 63.8 AE 10.2 kPa (3 days).The decrease in both EB and E is likely caused by the absorbed water molecules, which alter the intermolecular interactions around the cross-linked area.Over time, water molecules bind to the cross-linked structure and weaken its bonding strength.The same phenomenon is reported in other hydrogels. [54]The EB measured for the KC/c-KGM (Figure 3b) exhibited less variation without a clear trend: 54.9 AE 2.2% (1 day), 58.2 AE 3.0% (2 days), and 56.1 AE 2.8% (3 days).Moreover, the associated E varied irregularly across the days of immersion: 71.4 AE 4.5 kPa (1 day), 83.7 AE 4.5 kPa (2 days), and 73.9 AE 8.3 kPa (3 days).Overall, the stress versus strain data shows little variability, indicating that the mechanical properties of the studied materials were not significantly changed during water immersion.When comparing E of two types of c-KGM, E of KC/c-KGM was marginally higher than that of c-KGM once the sample was immersed at least 2 days.Furthermore, the properties of KC/c-KGM remain relatively stable throughout the immersion period.This suggests that KC/c-KGM might be more resistant to the penetration of water molecules, which could potentially weaken the material strength.Given this resistance, it can be inferred that KC/c-KGM is less pliable than c-KGM and more stable, making it a suitable material for the bottom part of the actuator, which requires more rigidity than the chamber part.
In order to assess operation durability of the materials, cyclic tensile tests were performed by using the same specimens (Figure S3, Supporting Information), which were fabricated by the same procedure and equipment described above.The specimens were repeatedly stretched and released 1000 times from the initial position to 30% strain (21 mm) at a speed of 420 mm min À1 (i.e., 10% strain per second).The results from c-KGM (Figure 3c) and KC/c-KGM (Figure 3d) showed that they were durable and not torn apart after 1000 repetitions.In addition, the measured tensile stress was consistent except the very first cycle.The drift error (Equation (3) in Section 4) at zero strain after 1000 cycles for c-KGM and KC/c-KGM was 8.39% and 3.66%, respectively, indicating that both materials were still elastic after repeated tensile stress.The 1000 tested cycles may initially appear to fall short when compared to conventional silicone actuators.For long-term and rigorous applications, conventional silicone might still be the preferred choice.However, in scenarios where tasks are short-term and require frequent replacements or in environments where the impact of nonbiodegradable waste can be catastrophic, our material offers a significant advantage.
The specific innovation of this article is optimizing the mechanical properties (i.e., higher elongation at break, lower Young's modulus) of KGM gel to develop an edible soft actuator working underwater.In Figure 3e-f, EB and E are compared with edible and c-KGM-based materials tested in previous studies. [30]s shown in Figure 3e, the EB of our material is 26.9% larger than the previously largest EB (i.e., CA(OH) 2 /KGM).Meanwhile, the E of our material is within the order of kPa of Agar/KGM [55] and much smaller than E of other edible and c-KGM-based materials [48,49,[56][57][58] (Figure 3f ).Note that E of Ca(OH) 2 /KGM, [30] gelatin/KGM, [44] EC/KGM, [46] and Ag/KGM [57] have not been reported, and therefore no comparison is made.Although Agar/KGM is as soft as ours, it is water soluble and cannot be used for underwater operation.Based on these results, the c-KGM developed in this work is the most suitable for soft actuator applications, which generally require high stretchability and material compliance.

Edible Gripper for Underwater Operation
The same edible materials were used to manufacture multifingered pneumatic grippers for underwater manipulation.The step-by-step fabrication process of a single finger is illustrated in Figure 4a (please refer to Actuator fabrication in Section 4 for more details).
The finger properties were characterized by measuring its bending angle (Figure 4b) and blocked force (Figure 4c) upon pressurization.Note that the actuator was pressurized by liquid to avoid buoyancy effects.The gripper base was fixed to the experimental setup (Figure S4, Supporting Information) and immersed in water for 1, 2, and 3 days.Pressure was applied to the actuator starting from 3 to 20 kPa (gauge pressure) with 1 kPa stepwise increments.During the experiment, characterized samples exhibited repeatable operation underwater over the course of 3 days.The difference between the tip angle of the actuator in its pressurized and unpressurized state was used to calculate the bending angle.The maximum bending angle resulted at 47.1°AE 1.9°upon 20 kPa of liquid input.The bending angles of the 2 and 3 days immersed samples were slightly smaller than those for the 1-day sample.Since the bending angles for the 2 and 3 days immersions were identical, it can be assumed that water absorption within the actuator reached its saturation in 2 days.As the surface area/volume of the actuator is only 0.3 cm À1 , which is much smaller than the cuboid sample (6 cm À1 ), it would take more time to reach the water absorption saturation than the cuboid (1 day, refer to Figure 2b,c).Linear trends were predominant in the experimental data from the actuator, and they are consistent regardless of immersion days.Such a linear trend was observed also in other PneuNet actuators with a geometry similar to that used here but made of non-edible elastomers. [59,60]he blocked force was measured using a load cell connected to a moment arm mechanism that has a low friction bearing at the center (Figure 4c and S5, Supporting Information).The maximum force output was 0.1 N from the 1-day immersed actuator, which was comparable to a gelatin-based edible actuator. [21]ust like the bending angle test, the force output from the actuator that was immersed for 1 day was larger than that of the actuators that were immersed for 2 and 3 days, due to the decrease of Young's modulus of c-KGM during prolonged water submersion.Again, the linear trend between pressure input and force output was consistent with previously developed PneuNet actuators.Overall, the bending angle and force output of our c-KGMbased actuator was unchanged during underwater operation.
A number of grippers with different number of fingers were fabricated and used for grasping various submerged objects (Figure 1c-e, and Supplementary Video S2, Supporting Information).Three fingers were used for grasping objects such as kiwis and plums.However, four fingers were utilized when grasping smaller objects, like clams.The rationale behind this approach is that smaller objects can be more easily stabilized and lifted when a larger number of fingers are used.The heaviest object that could be lifted by the gripper was a kiwi (128.4 g) followed by a plum (96.8 g) and a clam (15.2 g).The four-fingered gripper could also grasp a thin object, such as seaweed (Figure S6, Supporting Information).

Conclusion
We have described a method and material composition for the fabrication of edible underwater materials whose mechanical performance is comparable to other non-edible elastomers and biodegradable elastomers that use toxic materials, and validated it by manufacturing the edible gripping actuator that can be operated in aquatic environments.Its elongation at break was 67.2%, greater than any other konjac-based material reported in the literature.Also, its Young's modulus was 77.4 kPa, which is comparable to conventional silicone materials used in soft robots.Furthermore, it could withstand more than 1000 repeated tensile elongations with small drift error (c-KGM: 8.39%, KC/c-KGM: 3.66%).These results suggest that the proposed method and material composition is not only suitable for soft grippers, but also for other edible robotic components that operate underwater or in very wet and humid environments.Although konjac provides limited calorie intake, it could be functionalized for specific nutritional purposes by including proteins, fat, and micronutrients.The validation prototype described here relies on an electric-powered pump.Failure to retrieve the electronics indeed poses environmental challenges.However, the increase in soft robot applications accentuates the importance of creating edible soft robots.If robots are lost or abandoned, then firstly their biodegradability ensures they don't contribute to aquatic pollution, and secondly their edibility ensures the safety of aquatic animals should the materials be ingested, unlike their silicone-bodied counterparts.In the future, integrating food-grade additives could be envisioned, allowing edible actuators to even offer nutritional or medicinal benefits to aquatic life.Furthermore, electronic systems for soft actuators can be substituted with non-electronic power sources, such as CO 2 gas generated through a mixture of SB and citric acid (Supplementary Video S3, Supporting Information). [22]This opens the possibility of creating an almost entirely biodegradable/edible soft robot, ensuring a minimal environmental footprint at the end of its lifecycle. [22,62]

Experimental Section
Materials: KGM (E number: E425, a code indicating its approval for use as a thickener, stabilizer, or emulsifier within the European Union (EU) food industry) was sourced from Now Foods.KC (E number: E407, recognized within the EU as a commonly used gelling, thickening, and stabilizing agent in a variety of food products) was obtained from a supplier.SB (E number: E500, known within the EU as a leavening agent and pH regulator in foods) was purchased from Sigma-Aldrich.Distilled water was provided by École Polytechnique Fédérale de Lausanne.All the relative mass fractions (wt%) are calculated based on the total mass of distilled water (i.e., 250 g) in solution.
c-KGM and KC/c-KGM gel Fabrication: The first step of c-KGM fabrication, a mixture of 13.02 g (5.208 wt%) KGM and 200 Â g (80 wt%) distilled water was prepared.The mixture was stirred using a planetary centrifugal mixer (THINKY, ARE-250) at 1000 rpm for 2 min and defoamed at 400 rpm for 2 min.It was then left to stand for 30 min, allowing KGM to absorb the water.The gel was softened by stirring it with the mixer at 1000 rpm for 10 min and then defoamed at 400 rpm for 2 min.Next, the softened gel was mixed with 1.53 g (0.612 wt%) sodium bicarbonate, and 50 g (20 wt%) distilled water for 1 min, and split into small portions.Finally, the gel was stirred with a mixer at 2000 rpm for 10 min to mix it completely and defoamed at 400 rpm for 2 min.In case of fabrication of KC/c-KGM gel, 0.688 g (0.275 wt%) KC was also mixed with KGM and distilled water.Except for this step, all the other steps are the same as those used for c-KGM fabrication.
Mold Fabrication for Water Solubility and Absorption Tests: The molds consist of a container and a top cover.The container has a space to cast c-KGM or KC/c-KGM gel.The top cover is used to seal the container after casting the gel inside.The molds were made of polycarbonate (PC), a heatresistant material capable of withstanding 90 °C.The PC filament (Ultimaker) was printed by an Ultimaker S5 3D printer.The overall size of the mold was 30 mm Â 30 mm Â 100 mm.Due to the 5 mm of wall thickness, the internal space for the gel had dimension of 20 mm Â 20 mm Â 90 mm.
Water Solubility and Absorption Test: c-KGM and KC/c-KGM gels were cast in separate 3D-printed molds, and firmly sealed with covers by using bolts and nuts.These assemblies were heated in distilled water (90 °C) for 1 h using an induction heater (ProfiCook, PC-DKI 1067); the water temperature was monitored using a digital thermometer (TFA Dostmann, 30.1058).The gel was then removed from the mold and immersed in 70% ethanol solution (Reactolab, Ethanol 70%) for 24 h to further agglomerate the gels.This procedure greatly shrunk porous holes in c-KGM and KC/c-KGM gels, as shown in Figure S7, Supporting Information.The konjac was cut into cubes (size: 10 mm on each side), and all the surfaces were wiped with a powder-free towel (Kimberly-Clark, Kimtech Science Precision Wipes).They were kept in a conical tube filled with distilled water, which was at least 10 times heavier than the weight summation of all the c-KGM and KC/c-KGM samples, for different durations (1 day to 7 days).Four samples were taken out from the conical tube on each day and dried in an oven for 24 h (50 °C) to remove moisture.The WSI and WAI were calculated by Equation ( 1) and (2), respectively.The sample mass was measured using an electronic balance (Denver Instrument, MXX-123) that has 1 mg of resolution.
Tensile Test: A 110 mm Â 60 mm Â 3 mm sized PC mold was printed using the same 3D printer.The mold consists of a container part and a top cover.The c-KGM gel and KC/c-KGM gel were cast in the mold and heated in distilled water (90 °C) for 1 h.Then, they were immersed in a 70% ethanol solution for 24 h, wiped with a powder-free towel, and immersed in distilled water at room temperature for 1, 2, and 3 days.After that, the material was cut into a dumbbell type 1A by a laser cutting machine (Trotec', Speedy 400).The exact dimension is shown in Figure S3, Supporting Information.Thickness was measured at the center of each specimen and 10 mm away from the center on both sides, using a micrometer (Vogel, DIN 863).The specimen was fixed to a tensile testing machine (Instron, 5695) and pulled at a speed of 50 mm min À1 until ruptured.The tensile force was measured every 100 ms and converted to tensile stress by dividing the cross-sectional area of the specimen (thickness: 3.1 AE 0.4 mm, width: 5 mm).The Young's modulus is determined by using the Yeoh's hyperelastic model. [61]yclic Tensile Test: c-KGM and KC/c-KGM specimens were prepared by following the same procedure as above and immersed in distilled water for 1 day.The specimen was fixed to the same tensile testing machine and repetitively stretched for 1000 times at a speed of 420 mm min À1 from its initial position to 30% strain (21 mm distance).Drift error (D e ) at 1000th cycle is obtained as where TS 1000 and TS 1 is tensile stress at 1000th cycle and first cycle, respectively.Actuator Fabrication: A mold was 3D printed with a PC filament (refer Figure 4a), and a silicone tube (inner diameter: 2 mm) was connected to it thereafter.The mold design was based on the PneuNet-type soft actuators [21] with five internal chambers, and the actuator size was 90 mm Â 20 mm Â 17.5 mm.The thickness of chamber ceiling and side wall was 2.5 and 2.75 mm, respectively.The mold was filled with c-KGM gel and then sealed with a cover.Here, this entire assembly was referred as "top part".It was then heated in distilled water (90 °C) for 30 min, followed by the removal of the bottom part of the mold.The base part mold, which was designed to hold 90 Â 20 Â 4 mm 3 of gel, was filled with KC/c-KGM gel.The remaining top part was assembled with the base part, and the entire mold was heated again in distilled water (90 °C) for 1 h.All the molded parts were removed and immersed in a 70% ethanol solution for 24 h.Finally, they were removed from the ethanol solution and stored in distilled water at room temperature.
Bending Angle Characterization: An illustration of the experimental setup is shown in Figure S4, Supporting Information.Actuators were immersed in distilled water for 1, 2, and 3 days before the test.Three samples were prepared for each immersion day.The actuator was firmly fixed to the acrylic mount with hook and loop fasteners.The acrylic mount was fixed to a 12.7 mm thick aluminum plate (Thorlabs, MB3045/M) using a vertical mounting jig (Thorlabs, AP90/M).The pressure applied to the actuator was controlled by an electro-pneumatic regulator (SMC, ITV1050-31F2L).The electro-pneumatic regulator was connected to a pump (Metabo, Basic 250-24W) via a polyurethane tube, which served as the fluid supply source.Here, the actuator was driven by liquid to remove any effect caused by buoyant force.
The cross-sectional view of the silicone tube shows that the fluid is divided into air area and distilled water area (in Figure S4, Supporting Information).Since there is no pressure difference between the air and liquid areas, the pressure regulated by the electro-pneumatic regulator is equal to the pressure applied to the actuator.The pressure was applied from 3 to 20 kPa in gauge pressure, with 1 kPa of step increment.A DC power supply (Meanwell, LRS-100-24) and a signal power supply (B&K Precision, BK9141) were used to control the pressure in the electropneumatic regulator.A camera (Canon, EOS 600D) was used to record the bending motion, and the bending angle was measured (after 5 s upon pressurization) using an image processing software (Tracker).Here, the bending angle was defined by the tip angle difference of the actuator before and after applying the pressure.
Blocked Force Characterization: The actuator was operated using the same setup as above.The raw blocked force was measured using an Instron load cell (model EX2580-500N) connected to a moment arm mechanism with a low friction bearing.The increment of applied pressure was also the same as above.The blocked force was then determined using the moment arm principle (refer Figure S5 and Equation (S1) in Supporting Information).
Gripper Demonstration: A gripper was developed by combining either three or four actuators.This gripper was used to lift a kiwi (128.4 g), a plum (96.8 g), and a clam (15.2 g) that were submerged in water.The kiwi and plum were lifted using three actuators, while the clam was lifted with four actuators.All items were grasped underwater, lifted to the surface of the water, and held for at least 2 s.All the food items were purchased from a local supermarket.
Biodegradation Test: c-KGM samples were immersed in the natural seawater collected from Tokyo Bay, Japan.

Figure 2 .
Figure 2. Fabrication process of the c-KGM actuator and its water-resistance tests.a-i) 13.02 g (5.208 wt%) of KGM is dissolved in 200 g (80 wt%) of distilled water, and the mixture is stirred by a centrifugal mixer.a-ii) 1.53 g (0.612 wt%) of SB and 50 g (20 wt%) of distilled water are added and stirred.aiii) c-KGM gel is cast in a mold, followed by heating.a-iv) the c-KGM is immersed in 70% ethanol solution .a-v) The c-KGM or KC/c-KGM is obtained.b,c) Water solubility index (WSI) and water absorption index (WAI) as a function of immersion time (b: c-KGM, c: KC/c-KGM).

Figure 4 .
Figure 4. Fabrication process and characterization results of the liquid-driven soft actuator.a-i) The preparation of mold and silicone tube.a-ii) KGM gel is cast.a-iii) The mold is fully assembled and heated.a-iv) The mold is cooled down to room temperature.a-v) The base part is removed.a-vi) KC/c-KGM gel is cast.a-vii) Molds are connected and heated.a-viii) The mold is removed, and the sample is immersed in ethanol.b) Measured bending angle as a function of hydraulic pressure.Three actuators were measured, and the average values are plotted.c) Measured blocked force as a function of hydraulic pressure.Three actuators were measured, and the average values are plotted.