Programmable somatosensory soft robots

Programmable somatosensory soft robots | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Programmable somatosensory soft robots Antonia Georgopoulou, Malena Aguiriano Calvo, Lorenzo Lucherini, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7149271/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Mar, 2026 Read the published version in npj Flexible Electronics → Version 1 posted 11 You are reading this latest preprint version Abstract Robotic intelligence has advanced greatly in the past decade. Nevertheless, integrating embodied intelligent and responsive behaviour into soft robotic systems remains challenging because it typically requires bulky hardware for environmental feedback and decision-making. While soft materials like poly(N-isopropylacrylamide) (PNIPAM) offer potential for simplified material-based actuation through temperature-responsive motion, their slow response and high energy demands limit their use in closed-loop control systems. To overcome this limitation, we present soft PNIPAM-based actuators with integrated hydrogel-based Joule heating, enabling localized actuation without significantly altering the temperature within 1 cm of the actuator. The potential of the material is demonstrated by processing it in into a soft gripper that can lift up to three-fold its own weight with integrated capability to adjust its actuation in response to the gripped object. This design is well-suited for energy-efficient manipulation and sorting of delicate items, such as those found in automated packaging systems. Physical sciences/Engineering Physical sciences/Materials science smart materials stimuli-responsive hydrogels 3D printing soft robotics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Significant advances in intelligent behaviors of soft robotics were achieved by embedded sensing and model-free control systems. Unfortunately, the physical capacity of soft robots to sense and actuate remains limited. 1 – 3 The ability to collect and process information from the robot’s environment for feedback and decision-making often depends on complex algorithms and hardware components. 4 – 7 Responsive materials offer a promising alternative to actuate robotic systems without the need for complicated hardware and control methods by changing environmental conditions. 8 – 10 A prominent example is the thermo-responsive hydrogel poly(N-isopropylacrylamide) (PNIPAM), which possesses a lower critical solution temperature (LCST) around 32°C. 11 , 12 The reversible swelling behaviour of PNIPAM can be harnessed to induce repeated motion, like bending. 13 – 18 Responsive materials can act as a link between the robot’s body and its environment, thereby offering the potential to actuate in a closed-loop. Closed-loop actuation of soft robots requires a link between body and control architecture. 4 , 19 , 20 A key limitation to establish this link with PNIPAM lies in its lack of electronic programmability, as conventionally the entire environment must be heated to trigger actuation, leading to long response times and high energy demands. 21 As a result of this limitation, PNIPAM is rarely used in soft robotics. 22 , 23 PNIPAM actuators can also be driven by exploiting heat radiation. 24 , 25 Yet, PNIPAM response to temperature changes is slow, making its implementation into closed-loop control systems challenging. Indeed, soft hydrogel actuators that enable seamless closed-loop control between sensing and actuation are yet to be realized. Electrical energy can be converted into heat through resistive losses, called Joule heating. 26 – 28 Joule heating elements can be integrated in closed-loop control setups in line with resistive sensors. Traditional Joule heaters are composed of rigid metals or alloys, which are suboptimal for soft actuators because they stiffen them. Flexible Joule heating elements have been fabricated from polymers and conductive fillers. However, these elements are often composed of rigid polymers or thermoplastic elastomers with a Shore hardness of 85A, which stiffen the overall structures. 29 – 32 Hydrogel-based flexible Joule heaters used as anti-freeze devices that restore the temperature of e-skins from subzero to room temperature have been introduced. However, they have never been employed to regulate thermally responsive actuators. 33 In this study, we introduce soft, thermo-responsive hydrogel-based actuators with incorporated Joule heating elements, which enable an energy-efficient closed-loop control of the actuation. We demonstrate the potential of these materials by 3D printing them into a soft robotic gripper, which can size-selectively lift desired fruits. Joule heaters are composed of an electronically and ionically conductive hydrogels. The localized heat, generated if currents flow through these conductive tracks, raises the temperature locally above the LCST of PNIPAM, thereby initiating its actuation within 35 s. Importantly, the Shore hardness of the Joule heater is similar to that of PNIPAM, preventing any hardness mismatch within our actuator that could negatively impact its lifespan. The deformation of the actuator is transferred to a sensor that transmits different somatosensory information, like the position of the actuator and contact with external objects to control the heating element, thereby enabling a closed-loop control, as shown schematically in Fig. 1 . We foresee the combination of electronic programmability, somatosensory feedback and the flexibility in the processing of these materials provided by their 3D printability to enable the fabrication of closed-loop actuators that are more flexible, energy efficient and can be used in temperature-sensitive environments, including biological milieus. Results and Discussion 2.1 The Effect of Actuator Stiffness on the Bending Angle 2.1.1 Tensile Properties To create a somatosensory thermally responsive actuator, we fabricate PNIPAM microgels by fragmenting single-network PNIPAM hydrogels. The fragments, with an average diagonal of 11 µm, are soaked in an aqueous solution containing NIPAM. The precursor-loaded PNIPAM microgels are jammed before they are cast or 3D printed into the desired shape. The structures are rigidified through polymerization of the NIPAM precursors to form a 2nd percolating hydrogel that interpenetrates the microgels and covalently connects them, resulting in thermally responsive DNGHs, which we use as an actuator (DNGH-A), as shown in Fig. 2 A. To assess the influence of the microstructure of the DNGH-A on its mechanical properties, we perform tensile tests on DNGHs and bulk double-network gels of the same composition but lacking any microstructure, as schematically shown in Fig. S1 A. Bulk double network PNIPAM and DNGH-A exhibit an ultimate strength of 0.1 MPa, elongation at break of 75% and a Young’s modulus of 0.12 MPa, as shown in Fig. 2 B. Because the 2nd network has the same crosslinker concentration as the 1st counterpart, the microstructure of DNGH-A does not affect the stiffness or work of fracture of the actuator. We select identical crosslinker concentrations for both phases because higher crosslinker concentrations in the microfragments resulted in the loss of thermoresponsivity in the DNGOGs and a reduced strain at break, as shown in Fig. S2 . Lower crosslinker concentrations in the 2nd network compromised the integrity of DNGOGs. To fabricate piezoresistive sensors, we impart the DNGHs electrical conductivity by incubating PAMPS microgels in an aqueous solution containing AAm and PEDOT:PSS, as shown in Fig. 2 C. We chose PAMPS microgels because they have a higher degree of swelling in water compared to PNIPAM such that they take up a larger volume of precursors, translating into superior work of fracture and strain at break of the resulting DNGHs, as shown in Fig. S3 . Precursor-loaded PNIPAM fragments are jammed and processed into the desired shape. The shape is rigidified by polymerizing the AAm to form a 2nd hydrogel network, thereby resulting in a piezoresistive DNGH Sensor (DNGH-S). To endow electronic programmability to our DNGH-A, we increase the conductivity of the DNGH-S to use it as a flexible Joule Heating element (DNGH-H). To achieve this goal, we add zinc chloride (ZnCl 2 ) to the precursor solution of the 2nd network. To test the influence of the microstructure of the DNGH-H on its mechanical properties, we perform tensile tests on it and compare the results to those of bulk samples possessing the identical overall composition. The Young’s modulus of ZnCl 2 -free DNGH-A is 0.7 MPa. The addition of ZnCl 2 , which converts DNGH-As into DNGH-Hs, increases the Young’s modulus to 1 MPa. The ultimate strength and strain at break do not significantly change upon addition of ZnCl 2 remaining at 0.78 MPa and 170%, respectively, as shown in Fig. 2 D. These results indicate that the DNGH-H maintain softness and extensibility, despite the added ionic conductivity. Because the 2nd network is relatively soft, the Young’s modulus of bulk hydrogels is much higher than that of DNGHs, 2.7 MPa. These results indicate that the microstructure of the heating element significantly softens it, such that its stiffness is within the same order of magnitude as that of DNGH-As. 2.1.2 Actuation Bending Angle To transform the volumetric changes of PNIPAM into a bending motion required for gripping, we formulate bilayer structures by combining an active layer made of DNGH-A with a passive layer made of DNGH-S, as shown in Fig. S1 B. The DNGH-S and DNGH-A are simultaneously solidified through the free radical polymerization of acrylamide-based precursors to form a 2nd network. Hence, we expect the two types of 2nd networks to covalently connect, thereby forming strong material interfaces. To maximize the bending angle, a large volumetric change of the DNGH-A is required. We expect the volumetric change to depend on the stiffness of the DNGH-A. To evaluate the effect of the stiffness of the DNGH-A on its actuation, we vary the MBA crosslinker concentration within the PNIPAM microgels of the DNGH-A and quantify its volumetric change if heated above its LCST. When we increase the MBA concentration from 1 wt% to 2 wt%, the Young’s modulus of the DNGH-A increases from 0.09 MPa to 0.16 MPa, as shown in Fig. S3 . This increase in stiffness minimally reduces the volumetric change from 28–25%, as shown in Fig. 2 E. Similar results are obtained for bulk actuators of the same composition, as shown in Fig. 2 E, suggesting that the microstructure does not influence the volumetric change of the DNGH-A. Any further increase of the MBA concentration reduces the volumetric change of the hydrogel in response to temperature changes, rendering it unsuitable for actuation. Therefore, we fix the MBA concentration to 2 wt% for the remainder of this study. To assess the influence of the volumetric change of the actuator on the bending of the bilayer, we quantify the time-dependent bending angle. A bilayer composed of a 1.5 mm thick DNGH-A and a 1.5 mm thick passive DNGH-S bends 150°. The same bending angle can be achieved by single network PNIPAM actuators of the same thickness. This is in stark contrast to a bilayer composed of double network bulk counterparts, that only bends to 52°, as shown in Fig. 2 F. We assign the much lower bending angle of bulk double network hydrogels to the increased stiffness of the passive layer, as shown in Fig. 2 G. Remarkably, DNGH-based actuators maintain a horizontal position at rest, in stark contrast to bilayers based on single network bulk hydrogels despite their similar stiffnesses, as shown in Fig. 2 G, possibly due to the higher polymer density of the DNGH-S compare to its single network counterpart. This comparison highlights an important advantage of DNGHs: They can maintain a horizontal position at rest while reaching a high bending angle if actuated, a combination that cannot be achieved with bulk samples of similar compositions. The performance of many actuators depends on their response time. Single network bulk bilayers reach the maximum bending angle after 215 s. DNGH-based bilayers reach the maximum bending angle much faster: within 95 s, as shown in Fig. 2 F. We assign the faster response of DNGH-A to their initially straight form, as shown in Fig. 2 H and 2 I, in stark contrast to the single network bulk bilayers, which initially attain a negative deflection, as shown in Fig. 2 J. This negative deflection must be compensated before the actual actuation takes place, prolonging the actuation time. 2.1.3 Reversibility of the Actuation To assess the influence of the stiffness of the passive layer on the bending angle of the bilayer, we increase the stiffness of the DNGH-S by increasing the MBA concentration in the PAAm network, while the stiffness of the DNGH-A is kept constant. We do not vary the stiffness of the microgels in the DNGH-S, because softer particles lead to undesirable signal drift in the sensor response due to large stress-strain hysteresis in DNGH-S. 34 As expected, the maximum bending angle decreases from 250° to 125° if the stiffness of the DNGH-S is increased from 0.5 MPa to 1.2 MPa, translating into an increase in Shore hardness from 8A to 10A, as shown in Fig. 2 G. A further increase in stiffness to 2.5 MPa decreases the bending angle to 51°. This result suggests that a soft passive layer is better suited for bending applications. However, too soft passive layers prevent the recovery of the initial shape if cooled to room temperature. Indeed, bilayers containing a passive DNGH-S layer with a stiffness of 0.47 MPa only recover 70% of the horizontal position, as shown in Fig. S4 and Movie 1. By contrast, the bilayer containing DNGH-Ss with a stiffness 1.2 MPa or higher return to their original shape if cooled to 25°C. Based on these results, we fix the stiffness of the passive DNGH-S to 1.2 MPa. 2.2 Flexible Joule Heating Element To minimize the amount of energy needed for actuation and enhance the applicability of our DNGH-A in environments that cannot be easily heated, we generate heat near the thermo-responsive PNIPAM-based DNGH-A. According to Joule’s first law, the heat generation in resistive heaters is influenced by their electrical conductivity. To investigate how conductivity affects the heat generation of our DNGH-H, we vary the ion concentration in the acrylamide solution and measure the temperature of the resulting DNGH-H when 5 V is applied. The surface temperature of 2 cm thick DNGH-H gels increases during the first 30 s, whereafter it plateaus independent of the concentration of added ZnCl 2 , as shown in Fig. 3 A. However, the plateau temperature increases with increasing ZnCl 2 concentration: An increase in the ZnCl 2 concentration from 20 wt% to 40 wt% increases the plateau temperature from 25°C to 30.5°C. This increase is in stark contrast to Joule’s first law that states that the plateau temperature decreases with increasing conductivity. We attribute this contradiction to the increase in heat capacity of the hydrogel with increasing ZnCl 2 concentrations. 29 , 30 Unfortunately, even the highest plateau temperature we obtain, 30.5°C, is below the LCST of PNIPAM such that it cannot induce actuation of the DNGH-A. To enhance the Joule heating effect, we functionalize the DNGH-H with an electrically conductive polymer poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS). The thermal conductivity of PEDOT:PSS is low, preventing its use as a Joule heating element, as shown in Fig. S5. However, the addition of PEDOT:PSS to the ZnCl 2 functionalized DNGH-A increases the plateau temperature to 52°C if connected to a 5 V source, as shown in Fig. 3 B. This temperature is well above the LCST of PNIPAM such that the actuator readily shrinks if 5 V is applied, as shown in Fig. S6. We obtain a similar temperature evolution for bulk hydrogels possessing the identical composition but lacking any microstructure, as shown in Fig. 3 B. These results indicate that the microstructure does not significantly affect the Joule heating efficiency. To verify that ionic and electronic conductivity are present in our system, we perform electrical impedance measurements on the DNGH-H. 35 The low PEDOT:PSS concentration present in DNGH-Hs is sufficient to impart them electrical conductivity because the polymer accumulates in their interstitial spaces. 36 To differentiate between effects caused by PEDOT:PSS and ZnCl 2 , we perform the same test on the sensor DNGH-S that contains 0.027 wt% PEDOT:PSS but lacks ZnCl 2 and DNGH lacking any functionalization. DNGHs and DNGH-Ss are resistors of low conductivity, as indicated by the straight line in the Nyquist plots in Fig. 3 C. By contrast, DNGH-Hs exhibit a semi-circle at low frequencies indicative of a resistive behaviour with an additional time constant, which is associated with ionic conductivity. 35 We observe a similar profile of the Nyquist plots in double network bulk materials, even though the values of the electrical impedance are two orders of magnitude higher than those of their granular counterparts, as shown in Fig. S9. This difference is possibly due to the reduced polymer density within the interstitial spaces of DNGH-Hs, which increases the diffusivity of ions within these locations and hence, their conductivity. To calculate the electronic and ionic conductivity of our DNGH, we model the Nyquist plots with the equivalent electronic circuit indicated in Fig. 3 D. The circuit contains elements that simulate the electronic, ionic conductivity of the DNGH-H, the resistivity and capacity of the gold electrodes and interfaces between sample and gold electrodes. 37 The addition of 40 wt% Zn 2+ increases the ionic conductivity by two orders of magnitude from 6.2∙10 − 3 S/m to 4.4∙10 − 1 S/m, as shown in Fig. 3 D. Interestingly, the addition of 40 wt% Zn 2+ also increases the electrical conductivity from 6.4∙10 − 4 S/m to 1.2∙10 − 2 S/m. The increase in electrical conductivity might be caused by the Zn 2+ induced dissociation of PEDOT from PSS. This dissociation leads to a precipitation of PEDOT that increases its concentration within the interstitial spaces of DNGHs and facilitates the formation of a percolating network, leading to a higher electrical conductivity. To test this hypothesis, we perform UV-VIS spectroscopy on the supernatant of jammed microgels. Indeed, the supernatant of jammed microgels encompassing PEDOT:PSS and Zn 2+ ions exhibits an absorption peak at 240 nm, indicative of PSS, 38 and one at 349 nm indicative of Zn 2+ ions, 39 as shown in Fig. S10a. By contrast, the supernatant of jammed microgels encompassing PEDOT:PSS exhibits the peak at 224 nm, indicative of PEDOT and 240 nm, indicative of PSS, as shown in Fig. S10a. The hypothesis that Zn 2+ ions dissociate PEDOT:PSS complexes is further confirmed by the visual inspection of the supernatant of DNGH-S, which exhibits blue color indicative of a high PEDOT concentration. In contrast, the supernatant of DNGH-H is transparent, indicating that the vast majority of PEDOT has been incorporated in the DNGH-H, as shown in Fig. S10b. 2.3 Joule Heating-induced Actuation To evaluate the capability of the DNGH-H Joule heating element to induce bending of the composite actuator strips, we assess the evolution of the bending angle of bilayer structures consisting of DNGH-A and DNGH-S. We immerse the bilayer in 500 mL aqueous solution at 25°C and apply 5 V. The bilayer with a total thickness of 3 cm reaches a maximum bending angle of 140° within 95 s when immersed in 55°C water. When we use a commercial metallic heating wire, the bending angle decreases to 109°. Nonetheless, the metallic wire is stiff, leading to local failure during bending. We attribute this structural failure to stiffness mismatch at the interface between soft hydrogel actuator and metallic Joule heater that can be avoided if we employ the DNGH-H. Remarkably, we achieve a similar bending angle, 105°, if we use the flexible DNGH-H as Joule heater, as shown in Fig. 4 A. Note that the maximum bending angle is smaller, 105° if the actuation is caused by Joule heating. Yet, the heating elements enable programmed closed loop control of the actuation without significant changes in the temperature of the surrounding. Indeed, the temperature of the water measured 10 mm apart from the active actuator reaches only 27°C, as shown in Fig. 4 B. The localized heating significantly reduces the power consumption of the actuator. The localized heating adds another benefit: these actuators can be used for biomedical applications without risking temperature-induced damage to biological tissue, in stark contrast to the PNIPAM-based counterparts that require temperature changes of their surroundings. Adding more heating elements could help achieve more homogeneous heating throughout the actuation length to achieve a bending angle as high as 140°, but this would significantly increase power consumption. 2.4 Rheological characterization and 3D printing 3D printing enables cost-effective fabrication of customized structures and heating paths, which allow for more energy-efficient closed-loop actuation. To assess if our material can be 3D printed through direct ink writing, we characterize its rheological properties. All tested granular inks are shear thinning, as shown in Fig. 5 A. Moreover, they exhibit a low yield point, characterized as the crossover of the storage G’ and loss modulus G’’ in strain sweeps at 9% strain, as shown in Fig. 5 B. In addition, they all display a fast shear stress recovery, as shown in Fig. 5 C. These results indicate that the granular material fulfills the rheological requirements for direct ink writing. To leverage this feature, we direct ink write a dm scale gripper composed of a DNGH-A layer and a DNGH-S layer, as shown in Fig. 5 D. The total fabrication time of this gripper is only 15 min. To enable closed-loop actuation, we 3D print the DNGH-H as a single line in the middle of the DNGH-A layer, as shown in Fig. 5 E. Because both types of inks contain acrylamide-based monomers within their microparticles, we expect the free radical polymerization of acrylamides to proceed across the material interfaces, yielding strong interfaces. The actuation speed of PNIPAM-based materials is limited by the water diffusion. Water diffusion within hydrogels is significantly slower than in bulk water. 40 To test if we can accelerate the actuation by introducing open pores into our grippers, we vary the geometrical infill density, while we keep the thickness of DNGH-Ss and DNGH-As constant at 1 mm. Indeed, the response time to achieve the maximum bending angle of 105°C of our gripper increases from 160 s to 190 s if we decrease the infill density from 90–50%, as shown in Fig. 5 D, well in agreement with previous reports. 41 , 42 The smaller pores achieved with the higher infill enhance the heat transfer efficiency due to their higher surface area-to-volume ratio and improve the convective heat transfer, which accelerate heat transfer and hence response rate. Importantly, changes in the infill density do not affect the maximum bending angle, as shown in Fig. 5 D. Increasing the total length of the heating element leads to a decrease of the maximum temperature of the DNGH-H, compromising the activation of the PNIPAM such that we do not investigate geometries with longer heating paths. Grippers typically need to reach large bending angles. To assess if we can increase the maximum bending angle, we vary the thickness of our actuator printed with a 90% infill density while keeping the thickness of the DNGH-S constant at 1 mm. Indeed, the bending angle exponentially increases if the thickness of the actuator is increased beyond 0.6 mm, as shown in Fig. 5 E. We associate this increase to a larger actuation force generated by thicker DNGH-A layers. A 1.8 mm thick actuator exhibits a maximum bending angle of 180°, which is desirable for our soft robotic gripper application. A further increase of the thickness to 2 mm does not yield a significant increase in the bending angle, most likely because the entire material becomes too stiff. 2.5 Somatosensory Gripper and Closed-loop Control Somatosensory feedback of our soft robot gripper requires a reliable piezoresistive signal response. To verify the reliability of the DNGH-S sensor, we subject it to 20 cycles where the strain is varied between 0–30%. Within these cycles, we do not measure any signal drift, as shown in Fig. S11. Even if tested under quasi-static conditions by introducing a hold time of 30 s at maximum and minimum strain, the DNGH-S exhibits minimal signal drifts: In this case, the signal relaxes by 5% at maximum strain and not at all at minimum strain, as shown in Fig. S11. These results demonstrate the high reliability of the piezoresistive response of our sensor. The detection limit of the sensor is 0.5% strain, which is equivalent to 0.1° of bending. Soft grippers must be able to autonomously open and close. In addition, it is often advantageous if they can raise and lower their positions after gripping or releasing objects. To introduce this functionality into our gripper, we establish a closed-loop control between our gripper and a robot arm that relies on the somatosensory feedback from the DNGH-S to recognise the time point when the desired object has been gripped, as schematically shown in Fig. 6 A. We do not apply any voltage during the descent of the arm. When the robot arm is well positioned, voltage is supplied to the DNGH-H to activate the DNGH-A that then grips the object. The sensor signal from the DNGH-S records this motion, signalling the robot arm to lift the gripper with the object when the gripper reached its final gripping position. When the gripper reaches its release position, the controller withholds the voltage from the DNGH-H such that the DNGH-A slowly releases the object. During this action, the arm returns to its initial position where it releases the gripped object, as shown in Fig. 6 B. Somatosensory feedback can be exploited to classify gripped objects. It should be noted that a 35 g DNGH gripper can lift up to three-fold its own weight, up to 110 g. To evaluate the possibility to recognize objects, we aim to classify strawberries among a selection of strawberries, oranges and grapes. When a 3 cm strawberry is gripped, the change in relative resistance, caused by an increased strain of the DNGH-S, is 0.18, as shown in Fig. 6 C. Lifting larger fruits results in smaller changes in strain of the DNGH-S and hence, smaller variations in the relative resistance. Indeed, the relative resistance changes only by 0.08 if a 4.5 cm diameter cluster of grapes is gripped and by 0.05 if a 6 cm diameter orange is gripped. Based on those values, we program the controller to deactivate the heating element if the relative resistance is lower than 0.18 such that the gripper releases the object. By contrast, if the strawberry is gripped, the robot arm lifts, as shown in Fig. 6 D and Movie 2. Such object recognition features enable sorting and packaging applications. Soft robotic grippers with closed loop control and integrated somatosensory sensing have been previously reported. However, these grippers are traditionally based on tendon and pneumatic-based actuation such that they cannot be 3D printed. Moreover, these grippers typically exhibit Shore hardnesses around 40A. Our DNGH gripper for the first time combines closed loop control, softness and 3D printability, as summarized in Fig. 6 E. Furthermore, our DNGH gripper requires only 5 V to controllably grip objects, such that it can be operated with small batteries. This voltage is lower than that required to actuate conventional actuators, like pneumatic systems that rely on pumps, which typically require 24 V. In addition, the power we apply to the Joule heater to activate the gripper is only 2 W, which is significantly lower that the 12 W required for powering pneumatic based grippers. We foresee the energy-efficient actuation combined with the closed-loop control and processing flexibility to open up new opportunities in the design of intelligent soft grippers and robots. Conclusion We introduce soft grippers that can lift up to three times their own weight from double network granular hydrogels. These materials display Shore hardnesses below 10A while reaching a bending angle of 180° withing 90 s. Importantly, this bending is reached without significantly changing the surrounding temperature using a voltage as low as 5 V and a power as low as 2 W. We actuate the gripper by exploiting Joule heating, caused by a conductive double network granular hydrogel-based material (DNGH-H) that transforms electricity into localized heat. The localized heat causes PNIPAM to partially collapse, thereby inducing a bending of the gripper without excessively heating the surroundings. This feature clearly sets apart our DNGH-based actuator from more conventional PNIPAM-based counterparts that require heating of the bulk surrounding for actuation, rendering them less energy efficient and slower. Furthermore, integrated soft Joule heaters attribute electronic programmability to the gripper, a feature necessary for closed-loop control in soft robotics. The granular nature of the ink enables its direct ink writing into intricate multi-material 3D structures such as a 15 cm sized gripper capable of lifting objects as heavy as 75 g if immersed in a water bath. Leveraging the piezoresistivity of the passive layer, we design a gripper that operates with a closed-loop control, which enables picking up a pre-defined element, such as a strawberry from a fruit mixture. We foresee the combination of multi-functionality, programmability, softness and processibility of our DNGH to open up new applications in energy-efficient autonomous soft robots. Experimental Section Materials N-isopropylacrylamide (NIPAM) (Carl Roth, Germany), Acrylamido-2-methylpropane sulfonic acid (AMPS) (Sigma-Aldrich, USA), an aqueous solution containing 30 wt % and 40% acrylamide (Fisher Scientific, USA), N,N-methylene bisacrylamide (MBA) (Carl Roth, Germany), 2-hydroxy-2-methylpropriophenone (Sigma-Aldrich, USA), zinc chloride (Sigma-Aldrich, USA), PEDOT:PSS (Sigma-Aldrich, USA) were used as received. Microgel and DNGH Preparation Bulk PAMPS samples were cast from an aqueous precursor solution containing 25 wt% AMPS, 5 wt% MBA relative to the total AMPS concentration and 0.5 µl ml − 1 , 2-Hydroxy-2-methylpropiophenone that we use as a photoinitiator. To initiate the free radical polymerization, we expose the precursor to 1 mW cm − 2 of 364 nm UV light for 5 min. PAMPS microgels were produced by fragmenting bulk hydrogels at room temperature using a grinder equipped with steel blades. To produce the double network granular hydrogel sensor (DNGH-S), PAMPS microgels were incubated overnight in an aqueous solution containing 30 wt% acrylamide, 2% MBA relative to the total polyacrylamide, 0.5 µl ml − 1 PI, 0.027 wt% PEDOT:PSS. To produce the double network granular hydrogel joule heating elements (DNGH-Hs), PAMPS microgels were incubated overnight in an aqueous solution containing 30 wt% acrylamide, 2% MBA relative to the total polyacrylamide, 0.5 µl ml − 1 PI, 0.027 wt% PEDOT:PSS and 40 wt% ZnCl 2 . Bulk PNIPAM samples were cast from an aqueous precursor solution containing 20 wt% NIPAM, 2 wt% MBA relative to the total NIPAM concentration and 0.5 µl ml − 1 2-Hydroxy-2-methylpropiophenone PI. The cast samples were solidified by exposing them to 1 mW cm − 2 of 364 nm UV light for 5 min to initiate the free radical polymerization of the precursors. PNIPAM microgels were produced by fragmentation at room temperature using a grinder equipped with steel blades. To produce the double network granular actuator (DNGH-A), PAMPS microgels were incubated overnight in an aqueous solution containing 20 wt% NIPAM, 2% MBA relative to the total NIPAM concentration and 0.5 µl ml − 1 PI. Microgels were jammed through centrifugation at 4500 g for 10 min. The supernatant was discarded. The resulting DNGH ink was cast or 3D printed before the samples were exposed to UV 364 nm, 1 mW cm − 1 for 5 minutes to initiate the free radical polymerization of the precursors for the 2nd network. Double Network Bulk Preparation Double network bulk samples for the sensor and heating element were prepared by casting an aqueous solution containing 25 wt% AMPS, 5 wt% MBA relative to the total AMPS concentration and 0.5 µl ml − 1 PI. The films were polymerized by exposure to UV 366 nm, 1 mW cm − 1 for 10 minutes. For the sensor, the films were swollen in an aqueous solution containing 25 wt% AAm monomer, 0.3% mol MBA, 0.5 µl ml − 1 PI in relation to the AAm and 0.027 wt% PEDOT:PSS. For the heating element, samples were swollen in an aqueous solution containing AAm and 40 wt% ZnCl 2 . After overnight swelling of the microgels, the films were exposed to UV 366 nm, 1 mW cm − 1 for 10 minutes to initiate the free radical polymerization. Double network bulk actuators were prepared by casting an aqueous solution containing 20 wt% NIPAM, 2 wt% MBA and 0.5 µl ml − 1 PI. The films were polymerized by exposure to UV 366 nm, 1 mW cm − 1 for 10 minutes. The films were incubated in an aqueous solution containing 20 wt% NIPAM, 2 wt% MBA and 0.5 µl ml − 1 PI. After overnight swelling, the films were exposed to UV 366 nm, 1 mW cm − 1 for 10 minutes to initiate the free radical polymerization of the 2nd network. Joule Heating Effect To investigate the joule heating effect, a power source from PeakTech (PeakTech, Germany) was used to apply a voltage of 5V. The temperature was measured with a DHT11(Adafruit Industries, New York, USA) and an Arduino microcontroller. The temperature of the surrounding water was measured with an infrared camera. Impedance Spectroscopy Two-probes electrical impedance spectroscopy was performed with circular samples of 1 cm diameter and 2 mm thickness, using a Gamry potentiostat (Gamry Instruments, USA). Two gold-plated glass slides were used as electrodes. The frequency of the alternating current (AC) was varied from 0.1 Hz to 0.1 MHz using 5 measurement points per decade. The amplitude of the signal is 10 mV. The equivalent circuit was fitted to the experimental data using the Gamry software (EChem Analyst 2, Gamry). The samples were tested at swelling equilibrium in deionized water. Three samples were measured per sample type and plots represent the average of those three measurements. Rheological Characterization To characterize the rheology of the ink, shear rate sweeps, strain sweeps, and shear recovery tests were performed with a DHR-3TA Discovery instrument (TA Instruments, USA). An 8 mm parallel plate geometry was used with a gap of 0.8 mm. The TRIOS software from TA Instruments was used for recording the data. Amplitude sweep measurements were performed at 1.0 rad s − 1 and subjected to strains over a range of 0.01 to 300%. Shear thinning measurements were performed in the rotational mode with a strain rate between 1-100 s − 1 . Shear recovery measurements were performed at 1.0 rad s − 1 and strains of 1 and 30% were applied alternatively for 60 s. The cycle was repeated two times. Tensile testing and piezoresistive sensor response characterization To fabricate samples for tensile testing, the DNGH-S hydrogel was cast into dog-bone shaped Teflon molds with a cross-section of 4.6 mm 2 . Tensile tests were performed with a Zwick Roell Z005 universal testing machine (Zwick Roell, Ulm, Germany) with a strain rate of 200 mm/min. To characterize the piezoresistive response, ten cycles of 0–30% strain and 0-100% strain were performed while simultaneously recording the electrical resistance with an Arduino microcontroller equipped with an analogue to digital converter (ADC) unit. Quasi-static measurements were performed by applying five cycles of 0–30% strain with a dwell time of 30 s applied at maximum and minimum strains. The drift was calculated as the percentage difference of the relative resistance value at maximum strain between the second and last cycle. The relaxation was calculated as the percentage difference of the electrical resistance at the beginning and end of the dwell time, during the quasi-static measurements. 3D Printing A BIO X bioprinter (CellInk, USA) equipped with three extruders was used to fabricate actuator strips and the soft robotic gripper in a one-step process. The ink was extruded with a 20G conical nozzle (0.603 mm diameter) through a pressure driven piston with a speed of 10 mm s − 1 . UV-VIS spectroscopy UV–Vis spectroscopy of the supernatant of jammed microgels was performed between 200–700 nm with Perkin Elmer Lambda 365 spectrophotometer (Perkin Elmer,USA). Tests with Gripper and Robot Arm A Universal Robots UR5 robotic arm was used for lowering and lifting the gripper with a speed of 10 mm/s. The closed loop control was established linking the sensor response of the gripper with the lifting and lowering function of the robot arm at control frequency 200 Hz. Declarations Acknowledgements This project was financially supported by the NCCR Bioinspired Materials (51NF40-205603) through the fellowship program Women in Science. Author Contributions Antonia Georgopoulou : Conceptualization, Validation, Formal Analysis, Investigation, Resources, Data curation, Writing – Original Draft, Visualization, Supervision. Malena Aguiriano Calvo : Formal Analysis, Investigation, Data curation, Visualization Writing – Review and Editing. Lorenzo Lucherini : Validation, Investigation, Writing – Review and Editing. Sudong Lee : Investigation, Data Curation, Writing – Review and Editing. Josie Hughes : Validation, Writing – Review and Editing, Supervision. Esther Amstad : Conceptualization, Methodology, Validation, Formal Analysis, Writing – Review and Editing, Supervision, Project administration. Data Availability Statement Data for this article, including electromechanical characterization data, rheological characterization data, impedance spectra, bending angles during actuation and sensor signals are available at Zenodo: https://doi.org/10.5281/zenodo.15922612 References Hughes, J., Abdulali, A., Hashem, R. & Iida, F. Embodied Artificial Intelligence: Enabling the Next Intelligence Revolution. IOP Conf. Ser.: Mater. Sci. Eng. 1261 , 012001 (2022). Mengaldo, G. et al. A concise guide to modelling the physics of embodied intelligence in soft robotics. Nat Rev Phys 4 , 595–610 (2022). Laschi, C. The multifaceted approach to embodied intelligence in robotics. Science Robotics 10 , eadx2731 (2025). Zhao, Z. et al. Exploring Embodied Intelligence in Soft Robotics: A Review. Biomimetics 9 , 248 (2024). Loutfi, A. & Coradeschi, S. Smell, think and act: A cognitive robot discriminating odours. Auton Robot 20 , 239–249 (2006). Ma, B., Xu, C., Cui, L., Zhao, C. & Liu, H. Magnetic Printing of Liquid Metal for Perceptive Soft Actuators with Embodied Intelligence. ACS Appl. Mater. Interfaces 13 , 5574–5582 (2021). Georgopoulou, A. et al. 3D printing of self-healing longevous multi-sensory e-skin. Commun Mater 6 , 121 (2025). Dong, Y. et al. Multi-stimuli-responsive programmable biomimetic actuator. Nat Commun 10 , 4087 (2019). Zhang, X. et al. The Pathway to Intelligence: Using Stimuli-Responsive Materials as Building Blocks for Constructing Smart and Functional Systems. Advanced Materials 31 , 1804540 (2019). Liu, Z., Wang, W., Xie, R., Ju, X.-J. & Chu, L.-Y. Stimuli-responsive smart gating membranes. Chemical Society Reviews 45 , 460–475 (2016). Kim, Y. S. et al. Thermoresponsive actuation enabled by permittivity switching in an electrostatically anisotropic hydrogel. Nature Mater 14 , 1002–1007 (2015). Lee, H., Choi, H., Lee, M. & Park, S. Preliminary study on alginate/NIPAM hydrogel-based soft microrobot for controlled drug delivery using electromagnetic actuation and near-infrared stimulus. Biomed Microdevices 20 , 103 (2018). Li, X., Cai, X., Gao, Y. & J. Serpe, M. Reversible bidirectional bending of hydrogel-based bilayer actuators. Journal of Materials Chemistry B 5 , 2804–2812 (2017). Tan, Y. et al. A Fast, Reversible, and Robust Gradient Nanocomposite Hydrogel Actuator with Water-Promoted Thermal Response. Macromolecular Rapid Communications 39 , 1700863 (2018). Haraguchi, K., Kimura, Y. & Shimizu, S. Reversible generation of large retractive tensile forces in isometric chemo-mechanical actuators composed of nanocomposite hydrogels and aqueous NaCl solutions. Soft Matter 14 , 927–933 (2018). Liu, J. et al. Gradient porous PNIPAM-based hydrogel actuators with rapid response and flexibly controllable deformation. Journal of Materials Chemistry C 8 , 12092–12099 (2020). Spratte, T. et al. Increasing the Efficiency of Thermoresponsive Actuation at the Microscale by Direct Laser Writing of pNIPAM. Advanced Materials Technologies 8 , 2200714 (2023). Yao, C. et al. Poly(N-isopropylacrylamide)-Clay Nanocomposite Hydrogels with Responsive Bending Property as Temperature-Controlled Manipulators. Advanced Functional Materials 25 , 2980–2991 (2015). A. Baulin, V. et al. Intelligent soft matter: towards embodied intelligence. Soft Matter 21 , 4129–4145 (2025). Pfeifer, R., Iida, F. & Lungarella, M. Cognition from the bottom up: on biological inspiration, body morphology, and soft materials. Trends in cognitive sciences 18 , 404–413 (2014). Liu, J., Jiang, L., He, S., Zhang, J. & Shao, W. Recent progress in PNIPAM-based multi-responsive actuators: A mini-review. Chemical Engineering Journal 433 , 133496 (2022). Stuart, M. A. C. et al. Emerging applications of stimuli-responsive polymer materials. Nature Mater 9 , 101–113 (2010). Shen, Z., Chen, F., Zhu, X., Yong, K.-T. & Gu, G. Stimuli-responsive functional materials for soft robotics. Journal of Materials Chemistry B 8 , 8972–8991 (2020). Wang, H.-X. et al. Thermal-Responsive Hydrogel Actuators with Photo-Programmable Shapes and Actuating Trajectories. ACS Appl. Mater. Interfaces 14 , 51244–51252 (2022). Feng, X.-F. et al. A multi-stimuli-responsive actuator for efficient thermal management and various biomimetic locomotion. Cell Reports Physical Science 4 , 101588 (2023). Kotikian, A. et al. Innervated, Self-Sensing Liquid Crystal Elastomer Actuators with Closed Loop Control. Advanced Materials 33 , 2101814 (2021). Tang, X. et al. Temperature self-sensing and closed-loop position control of twisted and coiled actuator. Sensors and Actuators A: Physical 285 , 319–328 (2019). van der Weijde, J., Vallery, H. & Babuška, R. Closed-Loop Control Through Self-Sensing of a Joule-Heated Twisted and Coiled Polymer Muscle. Soft Robotics 6 , 621–630 (2019). Weerathunga, H. et al. Washable and Flexible All Carbon Electrothermal Joule Heater for Electric Vehicles. Advanced Materials Technologies 8 , 2201538 (2023). Tembei, S. A. N., Hessein, A., Fath El-Bab, A. M. R. & El-Moneim, A. A. A low voltage, flexible, graphene-based electrothermal heater for wearable electronics and localized heating applications. Materials Today: Proceedings 33 , 1840–1844 (2020). Chan, Y. H. T. et al. Portable Flexible Spherical Origami Joule-Heaters with Aerogel. Advanced Materials Interfaces n/a , 2400544. Georgopoulou, A., Diethelm, P., Wagner, M., Spolenak, R. & Clemens, F. Soft Self-Regulating Heating Elements for Thermoplastic Elastomer-Based Electronic Skin Applications. 3D Printing and Additive Manufacturing 11 , e828–e838 (2024). Huang, J. et al. Stretchable and Heat‐Resistant Protein‐Based Electronic Skin for Human Thermoregulation. Advanced Functional Materials 30 , 1910547 (2020). Georgopoulou, A. et al. Mechanoreceptive soft robotic molluscoids made of granular hydrogel-based organoelectronics. Materials & Design 114297 (2025). Cheng, H., Yue, S., Le, Q., Qian, Q. & Ouyang, J. A mixed ion-electron conducting carbon nanotube ionogel to efficiently harvest heat from both a temperature gradient and temperature fluctuation. J. Mater. Chem. A 9 , 13588–13596 (2021). Amstad, E. et al. Mechanoreceptive soft robotic molluscoids made of granular hydrogel-based organoelectronics. Preprint at https://doi.org/10.21203/rs.3.rs-5196882/v1 (2024). Yuk, H. et al. 3D printing of conducting polymers. Nat Commun 11 , 1604 (2020). Yoo, D. et al. Gradual thickness-dependent enhancement of the thermoelectric properties of PEDOT:PSS nanofilms. RSC Advances 4 , 58924–58929 (2014). Khan, Z. & Tanoli, M. A. Synthesis, Characterization, and In-Vitro Anti-Microbial and Anti-Oxidant Activities of Co+2, Ni+2, Cu+2 and Zn+2 Complexes of 5-chloro-2-hydroxybenzaldehyde-N-(2-oxo-1,2-dihydro-3H-indol-3-ylidene)hydrazone. Pakistan Journal of Chemistry 5 , 143–149 (2015). Masaro, L. & Zhu, X. X. Physical models of diffusion for polymer solutions, gels and solids. Progress in Polymer Science 24 , 731–775 (1999). Yilmaz, R. B., Chaabane, Y. & Mansard, V. Development of a Soft Actuator from Fast Swelling Macroporous PNIPAM Gels for Smart Braille Device Applications in Haptic Technology. ACS Appl. Mater. Interfaces 15 , 7340–7352 (2023). Strachota, B., Oleksyuk, K., Strachota, A. & Šlouf, M. Porous hybrid poly(N-isopropylacrylamide) hydrogels with very fast volume response to temperature and pH. European Polymer Journal 120 , 109213 (2019). Kim, S. Y. et al. Sustainable manufacturing of sensors onto soft systems using self-coagulating conductive Pickering emulsions. Science Robotics 5 , eaay3604 (2020). Koivikko, A. et al. Integrated stretchable pneumatic strain gauges for electronics-free soft robots. Commun Eng 1 , 1–10 (2022). Ham, J., Han, A. K., Cutkosky, M. R. & Bao, Z. UV-laser-machined stretchable multi-modal sensor network for soft robot interaction. npj Flex Electron 6 , 1–9 (2022). Hoang, T. T. et al. Soft robotic fabric gripper with gecko adhesion and variable stiffness. Sensors and Actuators A: Physical 323 , 112673 (2021). Georgopoulou, A., Vanderborght, B. & Clemens, F. Fabrication of a soft robotic gripper with integrated strain sensing elements using multi-material additive manufacturing. Frontiers in Robotics and AI 8 , 615991 (2021). Truby, R. L. et al. Soft Somatosensitive Actuators via Embedded 3D Printing. Advanced Materials 30 , 1706383 (2018). Spina, F., Pouryazdan, A., Costa, J. C., Cuspinera, L. P. & Münzenrieder, N. Directly 3D-printed monolithic soft robotic gripper with liquid metal microchannels for tactile sensing. Flex. Print. Electron. 4 , 035001 (2019). Shih, B. et al. Design Considerations for 3D Printed, Soft, Multimaterial Resistive Sensors for Soft Robotics. Frontiers in Robotics and AI 6 , (2019). Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformation.docx grippermontageorangestrawberry.mp4 SoftMediumHardstrips.mp4 Cite Share Download PDF Status: Published Journal Publication published 07 Mar, 2026 Read the published version in npj Flexible Electronics → Version 1 posted Editorial decision: Revision requested 03 Dec, 2025 Reviews received at journal 10 Nov, 2025 Reviews received at journal 09 Nov, 2025 Reviews received at journal 09 Nov, 2025 Reviewers agreed at journal 28 Oct, 2025 Reviewers agreed at journal 28 Oct, 2025 Reviewers agreed at journal 27 Oct, 2025 Reviewers invited by journal 17 Aug, 2025 Editor assigned by journal 21 Jul, 2025 Submission checks completed at journal 21 Jul, 2025 First submitted to journal 17 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7149271","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":502882840,"identity":"09cc2f6d-6d11-4fd6-ab81-e3876f3dbde3","order_by":0,"name":"Antonia Georgopoulou","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Antonia","middleName":"","lastName":"Georgopoulou","suffix":""},{"id":502882842,"identity":"37ba865c-2939-418b-88dc-22fdfd01916b","order_by":1,"name":"Malena Aguiriano Calvo","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Malena","middleName":"Aguiriano","lastName":"Calvo","suffix":""},{"id":502882844,"identity":"793a7298-b6df-4b7a-8fb6-436667178341","order_by":2,"name":"Lorenzo Lucherini","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Lorenzo","middleName":"","lastName":"Lucherini","suffix":""},{"id":502882846,"identity":"48d4492f-1372-4ad5-a468-135a2c15b765","order_by":3,"name":"Sudong Lee","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Sudong","middleName":"","lastName":"Lee","suffix":""},{"id":502882848,"identity":"8edc536d-1fab-4f66-9a56-a98f0f93569f","order_by":4,"name":"Josie Hughes","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Josie","middleName":"","lastName":"Hughes","suffix":""},{"id":502882849,"identity":"a827b8b1-e529-4fdf-879e-85f9df2075ad","order_by":5,"name":"Esther Amstad","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIiWNgGAWjYHACNgSTp4KBgb2dgYEZvw5mmBagOp4zQHyYJC28bURo4ec/f+zBxz0M8rr95w9+eDvPJo+HmYHxcwEeLZIzktkNZzxjMNx2I5lZcu62tGKgFmbpGXi0GNxgZpPmOcDAuO0GM4M077bDifuZGdiYefBosT9/mE36zwEG+23nDzP/5p3zP7GHkBYDhmQ2aYYDDInbDgAZvA0HCGuRuJFsJtlzQCIZ6BczyznHkoF+YWyWxqeFv//gM4kfB2xst50/+PjGmxq7PB725oOf8WmBWQZnJTAwMDYQ1oAMEkhTPgpGwSgYBSMBAAAbMETwa2Yq4gAAAABJRU5ErkJggg==","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":true,"prefix":"","firstName":"Esther","middleName":"","lastName":"Amstad","suffix":""}],"badges":[],"createdAt":"2025-07-17 12:53:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7149271/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7149271/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41528-026-00558-0","type":"published","date":"2026-03-07T15:57:29+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":89851950,"identity":"e9e455c4-f06a-48f7-af97-98364de29120","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":702904,"visible":true,"origin":"","legend":"\u003cp\u003eSoft robotic gripper exhibiting a closed loop between sensing and actuation. (i) The gripper is composed of two types of double network granular hydrogels (DNGHs): The actuator is made of Poly-N-isopropylacrylamide (PNIPAM) microgels connected through a PNIPAM secondary network. The sensor is composed of poly(2-acrylamido-2-methyl-1-propanesulfonic) acid (PAMPS) microgels connected through a polyacrylamide (PAAm) secondary network blended with poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) PEDOT:PSS. To convert certain locations of the sensor into a Joule heater, they are functionalized with PEDOT:PSS and Zn\u003csup\u003e2+\u003c/sup\u003e to impart them electronic and ionic conductivity. (ii) The sensor and actuator are covalently connected, resulting in firm material interfaces. (iii) Leveraging the granular structure of the material, it can be direct ink written into intricate structures with locally varying compositions.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/de82d0dadf9060897f7d7ce8.png"},{"id":89852433,"identity":"6843d7c5-3aa9-4a6e-a445-be26c93abc86","added_by":"auto","created_at":"2025-08-25 17:55:10","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":378910,"visible":true,"origin":"","legend":"\u003cp\u003eActuation of DNGHs and bulk counterparts. \u003cstrong\u003eA-C.\u003c/strong\u003e Schematic illustration of the double network bulk A. actuator and C. sensor with their DNGH counterpart \u003cstrong\u003eB\u003c/strong\u003e-\u003cstrong\u003eD\u003c/strong\u003e. The corresponding stress-strain curves of double network bulk (dotted line) and DNGH (straight line) based B. actuators, D. sensors (blue) and heating elements (red). \u003cstrong\u003eE.\u003c/strong\u003e Volumetric change of single-network bulk (orange), double network bulk (purple) and DNGH (violet) actuators as a function of the crosslinker concentration within PNIPAM. The crosslinker concentration within microgels and the 2\u003csup\u003end\u003c/sup\u003e network is identical. \u003cstrong\u003eF. \u003c/strong\u003eActuation kinetics of\u003cstrong\u003e \u003c/strong\u003esingle-network, double-network and DNGH bilayers. \u003cstrong\u003eG.\u003c/strong\u003e Maximum bending angle of DNGH bilayers at 55°C as a function of the crosslinking density within PNIPAM (green) and recovery of their initial position (purple) at 25°C. Timelapse of bilayers made of:\u003cstrong\u003e H\u003c/strong\u003e. Single-network PNIPAM bulk (scale bar 1cm), \u003cstrong\u003eJ.\u003c/strong\u003e Double network PNIPAM bulk (scale bar 1 cm)\u003cstrong\u003e \u003c/strong\u003eand\u003cstrong\u003e I.\u003c/strong\u003e DNGH PNIPAM (scale bar 1 cm) if heated to 55°C.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/c90582fa917896dd1f8d254b.png"},{"id":89851953,"identity":"c225ffdb-c6ae-4882-bc2f-49c84ba6fe01","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":205617,"visible":true,"origin":"","legend":"\u003cp\u003eComposition and characterization of the Joule heating element \u003cstrong\u003eA.\u003c/strong\u003e Temperature evolution of PAAm gels functionalized with 5 wt% (very light), 10 wt% (light), 20 wt% (medium), 40 wt% (dark) ZnCl\u003csub\u003e2\u003c/sub\u003e if connected to a 5 V source. \u003cstrong\u003eB.\u003c/strong\u003e Temperature of double network PAMPS/PAAm bulk gels (white) and PAMPS/PAAm DNGH-H (magenta) functionalized with 40 wt% ZnCl\u003csub\u003e2\u003c/sub\u003e and 0.027 wt% PEDOT:PSS measured at the surface of the gels. \u003cstrong\u003eC\u003c/strong\u003e. Nyquist plots for DNGH-H containing 40 wt% ZnCl\u003csub\u003e2\u003c/sub\u003e and 0.027 wt% PEDOT:PSS and DNGH-S containing 0.027 wt% PEDOT:PSS. \u003cstrong\u003eD\u003c/strong\u003e. Schematic and equivalent electric circuit of the DNGH, electrodes (R\u003csub\u003eAu\u003c/sub\u003e) and interface (R\u003csub\u003eint\u003c/sub\u003e and C\u003csub\u003eEDL\u003c/sub\u003e) measured with impedance spectroscopy. In the equivalent circuit models, R\u003csub\u003ee\u003c/sub\u003e represents electronic resistance, R\u003csub\u003eion\u003c/sub\u003e ionic resistance, C\u003csub\u003edif\u003c/sub\u003e the diffusion limited properties of the ionic conduction, constant phase element (CPE) accounts for geometric inhomogeneous or imperfect capacitance of the system. The corresponding Bode plots are shown in Fig. S8. \u0026nbsp;\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/0f7c335b766df3e5f38e14db.png"},{"id":89851958,"identity":"7231e2a4-8745-4978-bd27-73290a514362","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":89870,"visible":true,"origin":"","legend":"\u003cp\u003eInfluence of the heating element on actuation. \u003cstrong\u003eA.\u003c/strong\u003e DNGH-As activated by immersing them into 55°C water (blue), through an integrated metallic heating wire (orange) or DNGH-H (red) at a solution temperature of 25°C. \u003cstrong\u003eB.\u003c/strong\u003e Variation of temperature with the distance from the surface of the DNGH-A with integrated DNGH-H immersed in 25°C water measured with an infrared camera.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/55dde2dc7dbb781838db1686.png"},{"id":89851951,"identity":"06c12723-b094-4f44-896f-9e3d62796aba","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":182818,"visible":true,"origin":"","legend":"\u003cp\u003eRheological properties and co-3D Printing of DNGH actuator, sensor and heating element. Rheological characterization of inks used to fabricate the actuator (grey), sensor (blue), heating element (red) \u003cstrong\u003eA.\u003c/strong\u003e Amplitude sweep, \u003cstrong\u003eB.\u003c/strong\u003eshear rate and \u003cstrong\u003eC.\u003c/strong\u003e shear recovery tests. \u003cstrong\u003eD.\u003c/strong\u003e Photograph a DNGH gripper 3D printed with a 90% infill density and an integratedheating element (scale bar 1 cm). \u003cstrong\u003eE.\u003c/strong\u003e Close up of a single leg of the 3D printed gripper (scale bar 5 mm). \u003cstrong\u003eF\u003c/strong\u003e. Time required to reach the maximum bending (blue) and the maximum bending angle (pink) ofactuators as a function of the infill density used to print them \u003cstrong\u003eG\u003c/strong\u003e. Influence of layer thickness of actuators printed with a 90% infill density on the maximum bending angle.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/448fe128627896803b5d48a0.png"},{"id":89851964,"identity":"2e18ca4c-0c1d-410c-af35-1be522960a0f","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":321268,"visible":true,"origin":"","legend":"\u003cp\u003eSomatosensory feedback and closed-loop control for DNGH gripper. \u003cstrong\u003eA.\u003c/strong\u003e Schematic of the closed-loop control of the gripper relying on the somatosensory feedback \u003cstrong\u003eB. \u003c/strong\u003eClosed-loop control linking the heating, sensor response and positioning of the robot arm \u003cstrong\u003eC. \u003c/strong\u003eTesting the somatosensory capabilities of the gripper by picking up strawberries, grapes and an orange \u003cstrong\u003eD.\u003c/strong\u003e Photographs of the gripper with the integrated somatosensory feedback (i) before (ii) after the heating is activated such that a strawberry is gripped and (iii) lifted (scale bar 5 cm) . \u003cstrong\u003eE.\u003c/strong\u003e Radar chart of soft robotic grippers with integrated sensing depicting the soft actuator and soft sensor Shore hardness, their power consumption, 3D printability and closed-loop control. The grippers consist of silicone elastomer polydimethylsiloxane and carbon black PDMS-CB (olive),\u003csup\u003e43\u003c/sup\u003e silicone elastomer EcoFlex 00-50 (violet),\u003csup\u003e44\u003c/sup\u003e silicone elastomer Sylgard 184 (yellow),\u003csup\u003e45\u003c/sup\u003e silicone elastomer Sylgard 170 (magenta),\u003csup\u003e46\u003c/sup\u003e thermoplastic polyurethane TPU 80A (brown),\u003csup\u003e47\u003c/sup\u003e silicone elastomer EcoFlex 00-10/30 (blue),\u003csup\u003e48\u003c/sup\u003e rubber TangoBlack FLX 973 (teal),\u003csup\u003e49\u003c/sup\u003e rubber TangoBlack Plus (purple) \u003csup\u003e50\u003c/sup\u003e and the DNGH of this study (orange).\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/cc624927779071e91b042a9b.png"},{"id":104250598,"identity":"6d2da776-dedf-49a4-bbfb-7b1015b6e998","added_by":"auto","created_at":"2026-03-09 16:01:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2670024,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/371f4ffe-13ae-4dda-956c-1b6cc84991ca.pdf"},{"id":89851954,"identity":"4a201fe9-0535-4805-8673-b199ee5f82ea","added_by":"auto","created_at":"2025-08-25 17:47:10","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":2872669,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/9cb45b3e758a684bc2002c79.docx"},{"id":89851977,"identity":"7b38e8d7-5cdc-4afe-8b10-449fd09cd876","added_by":"auto","created_at":"2025-08-25 17:47:12","extension":"mp4","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":60088446,"visible":true,"origin":"","legend":"","description":"","filename":"grippermontageorangestrawberry.mp4","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/1e819b75be8ab0a6e6f6f4ab.mp4"},{"id":89851975,"identity":"1b0e046a-b317-41ad-b5ba-474030a863c1","added_by":"auto","created_at":"2025-08-25 17:47:11","extension":"mp4","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":22521951,"visible":true,"origin":"","legend":"","description":"","filename":"SoftMediumHardstrips.mp4","url":"https://assets-eu.researchsquare.com/files/rs-7149271/v1/a0bbff3988669742754618ff.mp4"}],"financialInterests":"No competing interests reported.","formattedTitle":"Programmable somatosensory soft robots","fulltext":[{"header":"Introduction","content":"\u003cp\u003eSignificant advances in intelligent behaviors of soft robotics were achieved by embedded sensing and model-free control systems. Unfortunately, the physical capacity of soft robots to sense and actuate remains limited.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e The ability to collect and process information from the robot\u0026rsquo;s environment for feedback and decision-making often depends on complex algorithms and hardware components.\u003csup\u003e\u003cspan additionalcitationids=\"CR5 CR6\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e Responsive materials offer a promising alternative to actuate robotic systems without the need for complicated hardware and control methods by changing environmental conditions.\u003csup\u003e\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e A prominent example is the thermo-responsive hydrogel poly(N-isopropylacrylamide) (PNIPAM), which possesses a lower critical solution temperature (LCST) around 32\u0026deg;C.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e The reversible swelling behaviour of PNIPAM can be harnessed to induce repeated motion, like bending.\u003csup\u003e\u003cspan additionalcitationids=\"CR14 CR15 CR16 CR17\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eResponsive materials can act as a link between the robot\u0026rsquo;s body and its environment, thereby offering the potential to actuate in a closed-loop. Closed-loop actuation of soft robots requires a link between body and control architecture.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e A key limitation to establish this link with PNIPAM lies in its lack of electronic programmability, as conventionally the entire environment must be heated to trigger actuation, leading to long response times and high energy demands.\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e As a result of this limitation, PNIPAM is rarely used in soft robotics.\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e PNIPAM actuators can also be driven by exploiting heat radiation.\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e Yet, PNIPAM response to temperature changes is slow, making its implementation into closed-loop control systems challenging. Indeed, soft hydrogel actuators that enable seamless closed-loop control between sensing and actuation are yet to be realized.\u003c/p\u003e\u003cp\u003eElectrical energy can be converted into heat through resistive losses, called Joule heating.\u003csup\u003e\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e Joule heating elements can be integrated in closed-loop control setups in line with resistive sensors. Traditional Joule heaters are composed of rigid metals or alloys, which are suboptimal for soft actuators because they stiffen them. Flexible Joule heating elements have been fabricated from polymers and conductive fillers. However, these elements are often composed of rigid polymers or thermoplastic elastomers with a Shore hardness of 85A, which stiffen the overall structures.\u003csup\u003e\u003cspan additionalcitationids=\"CR30 CR31\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e Hydrogel-based flexible Joule heaters used as anti-freeze devices that restore the temperature of e-skins from subzero to room temperature have been introduced. However, they have never been employed to regulate thermally responsive actuators.\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eIn this study, we introduce soft, thermo-responsive hydrogel-based actuators with incorporated Joule heating elements, which enable an energy-efficient closed-loop control of the actuation. We demonstrate the potential of these materials by 3D printing them into a soft robotic gripper, which can size-selectively lift desired fruits. Joule heaters are composed of an electronically and ionically conductive hydrogels. The localized heat, generated if currents flow through these conductive tracks, raises the temperature locally above the LCST of PNIPAM, thereby initiating its actuation within\u003c/p\u003e\u003cp\u003e35 s. Importantly, the Shore hardness of the Joule heater is similar to that of PNIPAM, preventing any hardness mismatch within our actuator that could negatively impact its lifespan. The deformation of the actuator is transferred to a sensor that transmits different somatosensory information, like the position of the actuator and contact with external objects to control the heating element, thereby enabling a closed-loop control, as shown schematically in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. We foresee the combination of electronic programmability, somatosensory feedback and the flexibility in the processing of these materials provided by their 3D printability to enable the fabrication of closed-loop actuators that are more flexible, energy efficient and can be used in temperature-sensitive environments, including biological milieus.\u003c/p\u003e"},{"header":"Results and Discussion","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 The Effect of Actuator Stiffness on the Bending Angle\u003c/h2\u003e\u003cdiv id=\"Sec4\" class=\"Section3\"\u003e\u003ch2\u003e2.1.1 Tensile Properties\u003c/h2\u003e\u003cp\u003eTo create a somatosensory thermally responsive actuator, we fabricate PNIPAM microgels by fragmenting single-network PNIPAM hydrogels. The fragments, with an average diagonal of 11 \u0026micro;m, are soaked in an aqueous solution containing NIPAM. The precursor-loaded PNIPAM microgels are jammed before they are cast or 3D printed into the desired shape. The structures are rigidified through polymerization of the NIPAM precursors to form a 2nd percolating hydrogel that interpenetrates the microgels and covalently connects them, resulting in thermally responsive DNGHs, which we use as an actuator (DNGH-A), as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA. To assess the influence of the microstructure of the DNGH-A on its mechanical properties, we perform tensile tests on DNGHs and bulk double-network gels of the same composition but lacking any microstructure, as schematically shown in Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA. Bulk double network PNIPAM and DNGH-A exhibit an ultimate strength of 0.1 MPa, elongation at break of 75% and a Young\u0026rsquo;s modulus of 0.12 MPa, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB. Because the 2nd network has the same crosslinker concentration as the 1st counterpart, the microstructure of DNGH-A does not affect the stiffness or work of fracture of the actuator. We select identical crosslinker concentrations for both phases because higher crosslinker concentrations in the microfragments resulted in the loss of thermoresponsivity in the DNGOGs and a reduced strain at break, as shown in Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e. Lower crosslinker concentrations in the 2nd network compromised the integrity of DNGOGs.\u003c/p\u003e\u003cp\u003eTo fabricate piezoresistive sensors, we impart the DNGHs electrical conductivity by incubating PAMPS microgels in an aqueous solution containing AAm and PEDOT:PSS, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC. We chose PAMPS microgels because they have a higher degree of swelling in water compared to PNIPAM such that they take up a larger volume of precursors, translating into superior work of fracture and strain at break of the resulting DNGHs, as shown in Fig. \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e. Precursor-loaded PNIPAM fragments are jammed and processed into the desired shape. The shape is rigidified by polymerizing the AAm to form a 2nd hydrogel network, thereby resulting in a piezoresistive DNGH Sensor (DNGH-S).\u003c/p\u003e\u003cp\u003eTo endow electronic programmability to our DNGH-A, we increase the conductivity of the DNGH-S to use it as a flexible Joule Heating element (DNGH-H). To achieve this goal, we add zinc chloride (ZnCl\u003csub\u003e2\u003c/sub\u003e) to the precursor solution of the 2nd network. To test the influence of the microstructure of the DNGH-H on its mechanical properties, we perform tensile tests on it and compare the results to those of bulk samples possessing the identical overall composition. The Young\u0026rsquo;s modulus of ZnCl\u003csub\u003e2\u003c/sub\u003e-free DNGH-A is 0.7 MPa. The addition of ZnCl\u003csub\u003e2\u003c/sub\u003e, which converts DNGH-As into DNGH-Hs, increases the Young\u0026rsquo;s modulus to 1 MPa. The ultimate strength and strain at break do not significantly change upon addition of ZnCl\u003csub\u003e2\u003c/sub\u003e remaining at 0.78 MPa and 170%, respectively, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD. These results indicate that the DNGH-H maintain softness and extensibility, despite the added ionic conductivity. Because the 2nd network is relatively soft, the Young\u0026rsquo;s modulus of bulk hydrogels is much higher than that of DNGHs, 2.7 MPa. These results indicate that the microstructure of the heating element significantly softens it, such that its stiffness is within the same order of magnitude as that of DNGH-As.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section3\"\u003e\u003ch2\u003e2.1.2 Actuation Bending Angle\u003c/h2\u003e\u003cp\u003eTo transform the volumetric changes of PNIPAM into a bending motion required for gripping, we formulate bilayer structures by combining an active layer made of DNGH-A with a passive layer made of DNGH-S, as shown in Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eB. The DNGH-S and DNGH-A are simultaneously solidified through the free radical polymerization of acrylamide-based precursors to form a 2nd network. Hence, we expect the two types of 2nd networks to covalently connect, thereby forming strong material interfaces.\u003c/p\u003e\u003cp\u003eTo maximize the bending angle, a large volumetric change of the DNGH-A is required. We expect the volumetric change to depend on the stiffness of the DNGH-A. To evaluate the effect of the stiffness of the DNGH-A on its actuation, we vary the MBA crosslinker concentration within the PNIPAM microgels of the DNGH-A and quantify its volumetric change if heated above its LCST. When we increase the MBA concentration from 1 wt% to 2 wt%, the Young\u0026rsquo;s modulus of the DNGH-A increases from 0.09 MPa to 0.16 MPa, as shown in Fig. \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e. This increase in stiffness minimally reduces the volumetric change from 28\u0026ndash;25%, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE. Similar results are obtained for bulk actuators of the same composition, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE, suggesting that the microstructure does not influence the volumetric change of the DNGH-A. Any further increase of the MBA concentration reduces the volumetric change of the hydrogel in response to temperature changes, rendering it unsuitable for actuation. Therefore, we fix the MBA concentration to 2 wt% for the remainder of this study.\u003c/p\u003e\u003cp\u003eTo assess the influence of the volumetric change of the actuator on the bending of the bilayer, we quantify the time-dependent bending angle. A bilayer composed of a 1.5 mm thick DNGH-A and a 1.5 mm thick passive DNGH-S bends 150\u0026deg;. The same bending angle can be achieved by single network PNIPAM actuators of the same thickness. This is in stark contrast to a bilayer composed of double network bulk counterparts, that only bends to 52\u0026deg;, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF. We assign the much lower bending angle of bulk double network hydrogels to the increased stiffness of the passive layer, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG. Remarkably, DNGH-based actuators maintain a horizontal position at rest, in stark contrast to bilayers based on single network bulk hydrogels despite their similar stiffnesses, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG, possibly due to the higher polymer density of the DNGH-S compare to its single network counterpart. This comparison highlights an important advantage of DNGHs: They can maintain a horizontal position at rest while reaching a high bending angle if actuated, a combination that cannot be achieved with bulk samples of similar compositions.\u003c/p\u003e\u003cp\u003eThe performance of many actuators depends on their response time. Single network bulk bilayers reach the maximum bending angle after 215 s. DNGH-based bilayers reach the maximum bending angle much faster: within 95 s, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF. We assign the faster response of DNGH-A to their initially straight form, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eH and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eI, in stark contrast to the single network bulk bilayers, which initially attain a negative deflection, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eJ. This negative deflection must be compensated before the actual actuation takes place, prolonging the actuation time.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section3\"\u003e\u003ch2\u003e2.1.3 Reversibility of the Actuation\u003c/h2\u003e\u003cp\u003eTo assess the influence of the stiffness of the passive layer on the bending angle of the bilayer, we increase the stiffness of the DNGH-S by increasing the MBA concentration in the PAAm network, while the stiffness of the DNGH-A is kept constant. We do not vary the stiffness of the microgels in the DNGH-S, because softer particles lead to undesirable signal drift in the sensor response due to large stress-strain hysteresis in DNGH-S.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e As expected, the maximum bending angle decreases from 250\u0026deg; to 125\u0026deg; if the stiffness of the DNGH-S is increased from 0.5 MPa to 1.2 MPa, translating into an increase in Shore hardness from 8A to 10A, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG. A further increase in stiffness to 2.5 MPa decreases the bending angle to 51\u0026deg;. This result suggests that a soft passive layer is better suited for bending applications. However, too soft passive layers prevent the recovery of the initial shape if cooled to room temperature. Indeed, bilayers containing a passive DNGH-S layer with a stiffness of 0.47 MPa only recover 70% of the horizontal position, as shown in Fig. S4 and Movie 1. By contrast, the bilayer containing DNGH-Ss with a stiffness 1.2 MPa or higher return to their original shape if cooled to 25\u0026deg;C. Based on these results, we fix the stiffness of the passive DNGH-S to 1.2 MPa.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Flexible Joule Heating Element\u003c/h2\u003e\u003cp\u003eTo minimize the amount of energy needed for actuation and enhance the applicability of our DNGH-A in environments that cannot be easily heated, we generate heat near the thermo-responsive PNIPAM-based DNGH-A. According to Joule\u0026rsquo;s first law, the heat generation in resistive heaters is influenced by their electrical conductivity. To investigate how conductivity affects the heat generation of our DNGH-H, we vary the ion concentration in the acrylamide solution and measure the temperature of the resulting DNGH-H when 5 V is applied. The surface temperature of 2 cm thick DNGH-H gels increases during the first 30 s, whereafter it plateaus independent of the concentration of added ZnCl\u003csub\u003e2\u003c/sub\u003e, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA. However, the plateau temperature increases with increasing ZnCl\u003csub\u003e2\u003c/sub\u003e concentration: An increase in the ZnCl\u003csub\u003e2\u003c/sub\u003e concentration from 20 wt% to 40 wt% increases the plateau temperature from 25\u0026deg;C to 30.5\u0026deg;C. This increase is in stark contrast to Joule\u0026rsquo;s first law that states that the plateau temperature decreases with increasing conductivity. We attribute this contradiction to the increase in heat capacity of the hydrogel with increasing ZnCl\u003csub\u003e2\u003c/sub\u003e concentrations.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e Unfortunately, even the highest plateau temperature we obtain, 30.5\u0026deg;C, is below the LCST of PNIPAM such that it cannot induce actuation of the DNGH-A.\u003c/p\u003e\u003cp\u003eTo enhance the Joule heating effect, we functionalize the DNGH-H with an electrically conductive polymer poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS). The thermal conductivity of PEDOT:PSS is low, preventing its use as a Joule heating element, as shown in Fig. S5. However, the addition of PEDOT:PSS to the ZnCl\u003csub\u003e2\u003c/sub\u003e functionalized DNGH-A increases the plateau temperature to 52\u0026deg;C if connected to a 5 V source, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB. This temperature is well above the LCST of PNIPAM such that the actuator readily shrinks if 5 V is applied, as shown in Fig. S6. We obtain a similar temperature evolution for bulk hydrogels possessing the identical composition but lacking any microstructure, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB. These results indicate that the microstructure does not significantly affect the Joule heating efficiency.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo verify that ionic and electronic conductivity are present in our system, we perform electrical impedance measurements on the DNGH-H.\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e The low PEDOT:PSS concentration present in DNGH-Hs is sufficient to impart them electrical conductivity because the polymer accumulates in their interstitial spaces.\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e To differentiate between effects caused by PEDOT:PSS and ZnCl\u003csub\u003e2\u003c/sub\u003e, we perform the same test on the sensor DNGH-S that contains 0.027 wt% PEDOT:PSS but lacks ZnCl\u003csub\u003e2\u003c/sub\u003e and DNGH lacking any functionalization. DNGHs and DNGH-Ss are resistors of low conductivity, as indicated by the straight line in the Nyquist plots in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC. By contrast, DNGH-Hs exhibit a semi-circle at low frequencies indicative of a resistive behaviour with an additional time constant, which is associated with ionic conductivity.\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e We observe a similar profile of the Nyquist plots in double network bulk materials, even though the values of the electrical impedance are two orders of magnitude higher than those of their granular counterparts, as shown in Fig. S9. This difference is possibly due to the reduced polymer density within the interstitial spaces of DNGH-Hs, which increases the diffusivity of ions within these locations and hence, their conductivity.\u003c/p\u003e\u003cp\u003eTo calculate the electronic and ionic conductivity of our DNGH, we model the Nyquist plots with the equivalent electronic circuit indicated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD. The circuit contains elements that simulate the electronic, ionic conductivity of the DNGH-H, the resistivity and capacity of the gold electrodes and interfaces between sample and gold electrodes.\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e The addition of 40 wt% Zn\u003csup\u003e2+\u003c/sup\u003e increases the ionic conductivity by two orders of magnitude from 6.2∙10\u003csup\u003e\u0026minus;\u0026thinsp;3\u003c/sup\u003e S/m to 4.4∙10\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e S/m, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD. Interestingly, the addition of 40 wt% Zn\u003csup\u003e2+\u003c/sup\u003e also increases the electrical conductivity from 6.4∙10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e S/m to 1.2∙10\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e S/m. The increase in electrical conductivity might be caused by the Zn\u003csup\u003e2+\u003c/sup\u003e induced dissociation of PEDOT from PSS. This dissociation leads to a precipitation of PEDOT that increases its concentration within the interstitial spaces of DNGHs and facilitates the formation of a percolating network, leading to a higher electrical conductivity. To test this hypothesis, we perform UV-VIS spectroscopy on the supernatant of jammed microgels. Indeed, the supernatant of jammed microgels encompassing PEDOT:PSS and Zn\u003csup\u003e2+\u003c/sup\u003e ions exhibits an absorption peak at 240 nm, indicative of PSS,\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e and one at 349 nm indicative of Zn\u003csup\u003e2+\u003c/sup\u003e ions,\u003csup\u003e39\u003c/sup\u003e as shown in Fig. S10a. By contrast, the supernatant of jammed microgels encompassing PEDOT:PSS exhibits the peak at 224 nm, indicative of PEDOT and 240 nm, indicative of PSS, as shown in Fig. S10a. The hypothesis that Zn\u003csup\u003e2+\u003c/sup\u003e ions dissociate PEDOT:PSS complexes is further confirmed by the visual inspection of the supernatant of DNGH-S, which exhibits blue color indicative of a high PEDOT concentration. In contrast, the supernatant of DNGH-H is transparent, indicating that the vast majority of PEDOT has been incorporated in the DNGH-H, as shown in Fig. S10b.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Joule Heating-induced Actuation\u003c/h2\u003e\u003cp\u003eTo evaluate the capability of the DNGH-H Joule heating element to induce bending of the composite actuator strips, we assess the evolution of the bending angle of bilayer structures consisting of DNGH-A and DNGH-S. We immerse the bilayer in 500 mL aqueous solution at 25\u0026deg;C and apply 5 V. The bilayer with a total thickness of 3 cm reaches a maximum bending angle of 140\u0026deg; within 95 s when immersed in 55\u0026deg;C water. When we use a commercial metallic heating wire, the bending angle decreases to 109\u0026deg;. Nonetheless, the metallic wire is stiff, leading to local failure during bending. We attribute this structural failure to stiffness mismatch at the interface between soft hydrogel actuator and metallic Joule heater that can be avoided if we employ the DNGH-H. Remarkably, we achieve a similar bending angle, 105\u0026deg;, if we use the flexible DNGH-H as Joule heater, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA. Note that the maximum bending angle is smaller, 105\u0026deg; if the actuation is caused by Joule heating. Yet, the heating elements enable programmed closed loop control of the actuation without significant changes in the temperature of the surrounding. Indeed, the temperature of the water measured 10 mm apart from the active actuator reaches only 27\u0026deg;C, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB. The localized heating significantly reduces the power consumption of the actuator. The localized heating adds another benefit: these actuators can be used for biomedical applications without risking temperature-induced damage to biological tissue, in stark contrast to the PNIPAM-based counterparts that require temperature changes of their surroundings. Adding more heating elements could help achieve more homogeneous heating throughout the actuation length to achieve a bending angle as high as 140\u0026deg;, but this would significantly increase power consumption.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Rheological characterization and 3D printing\u003c/h2\u003e\u003cp\u003e3D printing enables cost-effective fabrication of customized structures and heating paths, which allow for more energy-efficient closed-loop actuation. To assess if our material can be 3D printed through direct ink writing, we characterize its rheological properties. All tested granular inks are shear thinning, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA. Moreover, they exhibit a low yield point, characterized as the crossover of the storage G\u0026rsquo; and loss modulus G\u0026rsquo;\u0026rsquo; in strain sweeps at 9% strain, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB. In addition, they all display a fast shear stress recovery, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC. These results indicate that the granular material fulfills the rheological requirements for direct ink writing. To leverage this feature, we direct ink write a dm scale gripper composed of a DNGH-A layer and a DNGH-S layer, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD. The total fabrication time of this gripper is only 15 min. To enable closed-loop actuation, we 3D print the DNGH-H as a single line in the middle of the DNGH-A layer, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE. Because both types of inks contain acrylamide-based monomers within their microparticles, we expect the free radical polymerization of acrylamides to proceed across the material interfaces, yielding strong interfaces.\u003c/p\u003e\u003cp\u003eThe actuation speed of PNIPAM-based materials is limited by the water diffusion. Water diffusion within hydrogels is significantly slower than in bulk water.\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e To test if we can accelerate the actuation by introducing open pores into our grippers, we vary the geometrical infill density, while we keep the thickness of DNGH-Ss and DNGH-As constant at 1 mm. Indeed, the response time to achieve the maximum bending angle of 105\u0026deg;C of our gripper increases from 160 s to 190 s if we decrease the infill density from 90\u0026ndash;50%, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD, well in agreement with previous reports.\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e,\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e The smaller pores achieved with the higher infill enhance the heat transfer efficiency due to their higher surface area-to-volume ratio and improve the convective heat transfer, which accelerate heat transfer and hence response rate. Importantly, changes in the infill density do not affect the maximum bending angle, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD. Increasing the total length of the heating element leads to a decrease of the maximum temperature of the DNGH-H, compromising the activation of the PNIPAM such that we do not investigate geometries with longer heating paths.\u003c/p\u003e\u003cp\u003eGrippers typically need to reach large bending angles. To assess if we can increase the maximum bending angle, we vary the thickness of our actuator printed with a 90% infill density while keeping the thickness of the DNGH-S constant at 1 mm. Indeed, the bending angle exponentially increases if the thickness of the actuator is increased beyond 0.6 mm, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE. We associate this increase to a larger actuation force generated by thicker DNGH-A layers. A 1.8 mm thick actuator exhibits a maximum bending angle of 180\u0026deg;, which is desirable for our soft robotic gripper application. A further increase of the thickness to 2 mm does not yield a significant increase in the bending angle, most likely because the entire material becomes too stiff.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Somatosensory Gripper and Closed-loop Control\u003c/h2\u003e\u003cp\u003eSomatosensory feedback of our soft robot gripper requires a reliable piezoresistive signal response. To verify the reliability of the DNGH-S sensor, we subject it to 20 cycles where the strain is varied between 0\u0026ndash;30%. Within these cycles, we do not measure any signal drift, as shown in Fig. S11. Even if tested under quasi-static conditions by introducing a hold time of 30 s at maximum and minimum strain, the DNGH-S exhibits minimal signal drifts: In this case, the signal relaxes by 5% at maximum strain and not at all at minimum strain, as shown in Fig. S11. These results demonstrate the high reliability of the piezoresistive response of our sensor. The detection limit of the sensor is 0.5% strain, which is equivalent to 0.1\u0026deg; of bending.\u003c/p\u003e\u003cp\u003eSoft grippers must be able to autonomously open and close. In addition, it is often advantageous if they can raise and lower their positions after gripping or releasing objects. To introduce this functionality into our gripper, we establish a closed-loop control between our gripper and a robot arm that relies on the somatosensory feedback from the DNGH-S to recognise the time point when the desired object has been gripped, as schematically shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA. We do not apply any voltage during the descent of the arm. When the robot arm is well positioned, voltage is supplied to the DNGH-H to activate the DNGH-A that then grips the object. The sensor signal from the DNGH-S records this motion, signalling the robot arm to lift the gripper with the object when the gripper reached its final gripping position. When the gripper reaches its release position, the controller withholds the voltage from the DNGH-H such that the DNGH-A slowly releases the object. During this action, the arm returns to its initial position where it releases the gripped object, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eSomatosensory feedback can be exploited to classify gripped objects. It should be noted that a 35 g DNGH gripper can lift up to three-fold its own weight, up to 110 g. To evaluate the possibility to recognize objects, we aim to classify strawberries among a selection of strawberries, oranges and grapes. When a 3 cm strawberry is gripped, the change in relative resistance, caused by an increased strain of the DNGH-S, is 0.18, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC. Lifting larger fruits results in smaller changes in strain of the DNGH-S and hence, smaller variations in the relative resistance. Indeed, the relative resistance changes only by 0.08 if a 4.5 cm diameter cluster of grapes is gripped and by 0.05 if a 6 cm diameter orange is gripped. Based on those values, we program the controller to deactivate the heating element if the relative resistance is lower than 0.18 such that the gripper releases the object. By contrast, if the strawberry is gripped, the robot arm lifts, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD and Movie 2. Such object recognition features enable sorting and packaging applications.\u003c/p\u003e\u003cp\u003eSoft robotic grippers with closed loop control and integrated somatosensory sensing have been previously reported. However, these grippers are traditionally based on tendon and pneumatic-based actuation such that they cannot be 3D printed. Moreover, these grippers typically exhibit Shore hardnesses around 40A. Our DNGH gripper for the first time combines closed loop control, softness and 3D printability, as summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eE. Furthermore, our DNGH gripper requires only 5 V to controllably grip objects, such that it can be operated with small batteries. This voltage is lower than that required to actuate conventional actuators, like pneumatic systems that rely on pumps, which typically require 24 V. In addition, the power we apply to the Joule heater to activate the gripper is only 2 W, which is significantly lower that the 12 W required for powering pneumatic based grippers. We foresee the energy-efficient actuation combined with the closed-loop control and processing flexibility to open up new opportunities in the design of intelligent soft grippers and robots.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWe introduce soft grippers that can lift up to three times their own weight from double network granular hydrogels. These materials display Shore hardnesses below 10A while reaching a bending angle of 180\u0026deg; withing 90 s. Importantly, this bending is reached without significantly changing the surrounding temperature using a voltage as low as 5 V and a power as low as 2 W. We actuate the gripper by exploiting Joule heating, caused by a conductive double network granular hydrogel-based material (DNGH-H) that transforms electricity into localized heat. The localized heat causes PNIPAM to partially collapse, thereby inducing a bending of the gripper without excessively heating the surroundings. This feature clearly sets apart our DNGH-based actuator from more conventional PNIPAM-based counterparts that require heating of the bulk surrounding for actuation, rendering them less energy efficient and slower. Furthermore, integrated soft Joule heaters attribute electronic programmability to the gripper, a feature necessary for closed-loop control in soft robotics. The granular nature of the ink enables its direct ink writing into intricate multi-material 3D structures such as a 15 cm sized gripper capable of lifting objects as heavy as 75 g if immersed in a water bath. Leveraging the piezoresistivity of the passive layer, we design a gripper that operates with a closed-loop control, which enables picking up a pre-defined element, such as a strawberry from a fruit mixture. We foresee the combination of multi-functionality, programmability, softness and processibility of our DNGH to open up new applications in energy-efficient autonomous soft robots.\u003c/p\u003e"},{"header":"Experimental Section","content":"\u003cp\u003e\u003cem\u003eMaterials\u003c/em\u003e\u003c/p\u003e\u003cp\u003eN-isopropylacrylamide (NIPAM) (Carl Roth, Germany), Acrylamido-2-methylpropane sulfonic acid (AMPS) (Sigma-Aldrich, USA), an aqueous solution containing 30 wt % and 40% acrylamide (Fisher Scientific, USA), N,N-methylene bisacrylamide (MBA) (Carl Roth, Germany), 2-hydroxy-2-methylpropriophenone (Sigma-Aldrich, USA), zinc chloride (Sigma-Aldrich, USA), PEDOT:PSS (Sigma-Aldrich, USA) were used as received.\u003c/p\u003e\u003cp\u003e\u003cem\u003eMicrogel and DNGH Preparation\u003c/em\u003e\u003c/p\u003e\u003cp\u003eBulk PAMPS samples were cast from an aqueous precursor solution containing 25 wt% AMPS, 5 wt% MBA relative to the total AMPS concentration and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, 2-Hydroxy-2-methylpropiophenone that we use as a photoinitiator. To initiate the free radical polymerization, we expose the precursor to 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e of 364 nm UV light for 5 min. PAMPS microgels were produced by fragmenting bulk hydrogels at room temperature using a grinder equipped with steel blades. To produce the double network granular hydrogel sensor (DNGH-S), PAMPS microgels were incubated overnight in an aqueous solution containing 30 wt% acrylamide, 2% MBA relative to the total polyacrylamide, 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI, 0.027 wt% PEDOT:PSS. To produce the double network granular hydrogel joule heating elements (DNGH-Hs), PAMPS microgels were incubated overnight in an aqueous solution containing 30 wt% acrylamide, 2% MBA relative to the total polyacrylamide, 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI, 0.027 wt% PEDOT:PSS and 40 wt% ZnCl\u003csub\u003e2\u003c/sub\u003e.\u003c/p\u003e\u003cp\u003eBulk PNIPAM samples were cast from an aqueous precursor solution containing 20 wt% NIPAM, 2 wt% MBA relative to the total NIPAM concentration and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e 2-Hydroxy-2-methylpropiophenone PI. The cast samples were solidified by exposing them to 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e of 364 nm UV light for 5 min to initiate the free radical polymerization of the precursors. PNIPAM microgels were produced by fragmentation at room temperature using a grinder equipped with steel blades. To produce the double network granular actuator (DNGH-A), PAMPS microgels were incubated overnight in an aqueous solution containing 20 wt% NIPAM, 2% MBA relative to the total NIPAM concentration and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI. Microgels were jammed through centrifugation at 4500 g for 10 min. The supernatant was discarded. The resulting DNGH ink was cast or 3D printed before the samples were exposed to UV 364 nm, 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 5 minutes to initiate the free radical polymerization of the precursors for the 2nd network.\u003c/p\u003e\u003cp\u003e\u003cem\u003eDouble Network Bulk Preparation\u003c/em\u003e\u003c/p\u003e\u003cp\u003eDouble network bulk samples for the sensor and heating element were prepared by casting an aqueous solution containing 25 wt% AMPS, 5 wt% MBA relative to the total AMPS concentration and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI. The films were polymerized by exposure to UV 366 nm, 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 10 minutes. For the sensor, the films were swollen in an aqueous solution containing 25 wt% AAm monomer, 0.3% mol MBA, 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI in relation to the AAm and 0.027 wt% PEDOT:PSS. For the heating element, samples were swollen in an aqueous solution containing AAm and 40 wt% ZnCl\u003csub\u003e2\u003c/sub\u003e. After overnight swelling of the microgels, the films were exposed to UV 366 nm, 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 10 minutes to initiate the free radical polymerization.\u003c/p\u003e\u003cp\u003eDouble network bulk actuators were prepared by casting an aqueous solution containing 20 wt% NIPAM, 2 wt% MBA and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI. The films were polymerized by exposure to UV 366 nm, 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 10 minutes. The films were incubated in an aqueous solution containing 20 wt% NIPAM, 2 wt% MBA and 0.5 \u0026micro;l ml\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e PI. After overnight swelling, the films were exposed to UV 366 nm, 1 mW cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 10 minutes to initiate the free radical polymerization of the 2nd network.\u003c/p\u003e\u003cp\u003e\u003cem\u003eJoule Heating Effect\u003c/em\u003e\u003c/p\u003e\u003cp\u003eTo investigate the joule heating effect, a power source from PeakTech (PeakTech, Germany) was used to apply a voltage of 5V. The temperature was measured with a DHT11(Adafruit Industries, New York, USA) and an Arduino microcontroller. The temperature of the surrounding water was measured with an infrared camera.\u003c/p\u003e\u003cp\u003e\u003cem\u003eImpedance Spectroscopy\u003c/em\u003e\u003c/p\u003e\u003cp\u003eTwo-probes electrical impedance spectroscopy was performed with circular samples of 1 cm diameter and 2 mm thickness, using a Gamry potentiostat (Gamry Instruments, USA). Two gold-plated glass slides were used as electrodes. The frequency of the alternating current (AC) was varied from 0.1 Hz to 0.1 MHz using 5 measurement points per decade. The amplitude of the signal is 10 mV. The equivalent circuit was fitted to the experimental data using the Gamry software (EChem Analyst 2, Gamry). The samples were tested at swelling equilibrium in deionized water. Three samples were measured per sample type and plots represent the average of those three measurements.\u003c/p\u003e\u003cp\u003e\u003cem\u003eRheological Characterization\u003c/em\u003e\u003c/p\u003e\u003cp\u003eTo characterize the rheology of the ink, shear rate sweeps, strain sweeps, and shear recovery tests were performed with a DHR-3TA Discovery instrument (TA Instruments, USA). An 8 mm parallel plate geometry was used with a gap of 0.8 mm. The TRIOS software from TA Instruments was used for recording the data. Amplitude sweep measurements were performed at 1.0 rad s\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and subjected to strains over a range of 0.01 to 300%. Shear thinning measurements were performed in the rotational mode with a strain rate between 1-100 s\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e. Shear recovery measurements were performed at 1.0 rad s\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and strains of 1 and 30% were applied alternatively for 60 s. The cycle was repeated two times.\u003c/p\u003e\u003cp\u003e\u003cem\u003eTensile testing and piezoresistive sensor response characterization\u003c/em\u003e\u003c/p\u003e\u003cp\u003eTo fabricate samples for tensile testing, the DNGH-S hydrogel was cast into dog-bone shaped Teflon molds with a cross-section of 4.6 mm\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Tensile tests were performed with a Zwick Roell Z005 universal testing machine (Zwick Roell, Ulm, Germany) with a strain rate of 200 mm/min. To characterize the piezoresistive response, ten cycles of 0\u0026ndash;30% strain and 0-100% strain were performed while simultaneously recording the electrical resistance with an Arduino microcontroller equipped with an analogue to digital converter (ADC) unit. Quasi-static measurements were performed by applying five cycles of 0\u0026ndash;30% strain with a dwell time of 30 s applied at maximum and minimum strains. The drift was calculated as the percentage difference of the relative resistance value at maximum strain between the second and last cycle. The relaxation was calculated as the percentage difference of the electrical resistance at the beginning and end of the dwell time, during the quasi-static measurements.\u003c/p\u003e\n\u003ch3\u003e3D Printing\u003c/h3\u003e\n\u003cp\u003eA BIO X bioprinter (CellInk, USA) equipped with three extruders was used to fabricate actuator strips and the soft robotic gripper in a one-step process. The ink was extruded with a 20G conical nozzle (0.603 mm diameter) through a pressure driven piston with a speed of 10 mm s\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cem\u003eUV-VIS spectroscopy\u003c/em\u003e\u003c/p\u003e\u003cp\u003eUV\u0026ndash;Vis spectroscopy of the supernatant of jammed microgels was performed between 200\u0026ndash;700 nm with Perkin Elmer Lambda 365 spectrophotometer (Perkin Elmer,USA).\u003c/p\u003e\u003cp\u003e\u003cem\u003eTests with Gripper and Robot Arm\u003c/em\u003e\u003c/p\u003e\u003cp\u003eA Universal Robots UR5 robotic arm was used for lowering and lifting the gripper with a speed of 10 mm/s. The closed loop control was established linking the sensor response of the gripper with the lifting and lowering function of the robot arm at control frequency 200 Hz.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eThis project was financially supported by the NCCR Bioinspired Materials (51NF40-205603) through the fellowship program Women in Science.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003cu\u003eAuthor Contributions\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAntonia Georgopoulou\u003c/strong\u003e: Conceptualization, Validation, Formal Analysis, Investigation, Resources, Data curation, Writing \u0026ndash; Original Draft, Visualization, Supervision. \u003cstrong\u003eMalena Aguiriano Calvo\u003c/strong\u003e: Formal Analysis, Investigation, Data curation, Visualization Writing \u0026ndash; Review and Editing. \u003cstrong\u003eLorenzo Lucherini\u003c/strong\u003e: Validation, Investigation, Writing \u0026ndash; Review and Editing. \u0026nbsp;\u003cstrong\u003eSudong Lee\u003c/strong\u003e: Investigation, Data Curation, Writing \u0026ndash; Review and Editing. \u003cstrong\u003eJosie Hughes\u003c/strong\u003e: Validation, Writing \u0026ndash; Review and Editing, Supervision. \u003cstrong\u003eEsther Amstad\u003c/strong\u003e: Conceptualization, Methodology, Validation, Formal Analysis, Writing \u0026ndash; Review and Editing, Supervision, Project administration. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003cu\u003eData Availability Statement\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData for this article, including electromechanical characterization data, rheological characterization data, impedance spectra, bending angles during actuation and sensor signals are available at Zenodo: https://doi.org/10.5281/zenodo.15922612\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eHughes, J., Abdulali, A., Hashem, R. \u0026amp; Iida, F. Embodied Artificial Intelligence: Enabling the Next Intelligence Revolution. \u003cem\u003eIOP Conf. Ser.: Mater. Sci. Eng.\u003c/em\u003e \u003cstrong\u003e1261\u003c/strong\u003e, 012001 (2022).\u003c/li\u003e\n\u003cli\u003eMengaldo, G. \u003cem\u003eet al.\u003c/em\u003e A concise guide to modelling the physics of embodied intelligence in soft robotics. \u003cem\u003eNat Rev Phys\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 595\u0026ndash;610 (2022).\u003c/li\u003e\n\u003cli\u003eLaschi, C. The multifaceted approach to embodied intelligence in robotics. \u003cem\u003eScience Robotics\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, eadx2731 (2025).\u003c/li\u003e\n\u003cli\u003eZhao, Z. \u003cem\u003eet al.\u003c/em\u003e Exploring Embodied Intelligence in Soft Robotics: A Review. \u003cem\u003eBiomimetics\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 248 (2024).\u003c/li\u003e\n\u003cli\u003eLoutfi, A. \u0026amp; Coradeschi, S. Smell, think and act: A cognitive robot discriminating odours. \u003cem\u003eAuton Robot\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 239\u0026ndash;249 (2006).\u003c/li\u003e\n\u003cli\u003eMa, B., Xu, C., Cui, L., Zhao, C. \u0026amp; Liu, H. Magnetic Printing of Liquid Metal for Perceptive Soft Actuators with Embodied Intelligence. \u003cem\u003eACS Appl. Mater. Interfaces\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 5574\u0026ndash;5582 (2021).\u003c/li\u003e\n\u003cli\u003eGeorgopoulou, A. \u003cem\u003eet al.\u003c/em\u003e 3D printing of self-healing longevous multi-sensory e-skin. \u003cem\u003eCommun Mater\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 121 (2025).\u003c/li\u003e\n\u003cli\u003eDong, Y. \u003cem\u003eet al.\u003c/em\u003e Multi-stimuli-responsive programmable biomimetic actuator. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 4087 (2019).\u003c/li\u003e\n\u003cli\u003eZhang, X. \u003cem\u003eet al.\u003c/em\u003e The Pathway to Intelligence: Using Stimuli-Responsive Materials as Building Blocks for Constructing Smart and Functional Systems. \u003cem\u003eAdvanced Materials\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1804540 (2019).\u003c/li\u003e\n\u003cli\u003eLiu, Z., Wang, W., Xie, R., Ju, X.-J. \u0026amp; Chu, L.-Y. Stimuli-responsive smart gating membranes. \u003cem\u003eChemical Society Reviews\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 460\u0026ndash;475 (2016).\u003c/li\u003e\n\u003cli\u003eKim, Y. S. \u003cem\u003eet al.\u003c/em\u003e Thermoresponsive actuation enabled by permittivity switching in an electrostatically anisotropic hydrogel. \u003cem\u003eNature Mater\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1002\u0026ndash;1007 (2015).\u003c/li\u003e\n\u003cli\u003eLee, H., Choi, H., Lee, M. \u0026amp; Park, S. Preliminary study on alginate/NIPAM hydrogel-based soft microrobot for controlled drug delivery using electromagnetic actuation and near-infrared stimulus. \u003cem\u003eBiomed Microdevices\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 103 (2018).\u003c/li\u003e\n\u003cli\u003eLi, X., Cai, X., Gao, Y. \u0026amp; J. Serpe, M. Reversible bidirectional bending of hydrogel-based bilayer actuators. \u003cem\u003eJournal of Materials Chemistry B\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 2804\u0026ndash;2812 (2017).\u003c/li\u003e\n\u003cli\u003eTan, Y. \u003cem\u003eet al.\u003c/em\u003e A Fast, Reversible, and Robust Gradient Nanocomposite Hydrogel Actuator with Water-Promoted Thermal Response. \u003cem\u003eMacromolecular Rapid Communications\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 1700863 (2018).\u003c/li\u003e\n\u003cli\u003eHaraguchi, K., Kimura, Y. \u0026amp; Shimizu, S. Reversible generation of large retractive tensile forces in isometric chemo-mechanical actuators composed of nanocomposite hydrogels and aqueous NaCl solutions. \u003cem\u003eSoft Matter\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 927\u0026ndash;933 (2018).\u003c/li\u003e\n\u003cli\u003eLiu, J. \u003cem\u003eet al.\u003c/em\u003e Gradient porous PNIPAM-based hydrogel actuators with rapid response and flexibly controllable deformation. \u003cem\u003eJournal of Materials Chemistry C\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 12092\u0026ndash;12099 (2020).\u003c/li\u003e\n\u003cli\u003eSpratte, T. \u003cem\u003eet al.\u003c/em\u003e Increasing the Efficiency of Thermoresponsive Actuation at the Microscale by Direct Laser Writing of pNIPAM. \u003cem\u003eAdvanced Materials Technologies\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 2200714 (2023).\u003c/li\u003e\n\u003cli\u003eYao, C. \u003cem\u003eet al.\u003c/em\u003e Poly(N-isopropylacrylamide)-Clay Nanocomposite Hydrogels with Responsive Bending Property as Temperature-Controlled Manipulators. \u003cem\u003eAdvanced Functional Materials\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 2980\u0026ndash;2991 (2015).\u003c/li\u003e\n\u003cli\u003eA. Baulin, V. \u003cem\u003eet al.\u003c/em\u003e Intelligent soft matter: towards embodied intelligence. \u003cem\u003eSoft Matter\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 4129\u0026ndash;4145 (2025).\u003c/li\u003e\n\u003cli\u003ePfeifer, R., Iida, F. \u0026amp; Lungarella, M. Cognition from the bottom up: on biological inspiration, body morphology, and soft materials. \u003cem\u003eTrends in cognitive sciences\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 404\u0026ndash;413 (2014).\u003c/li\u003e\n\u003cli\u003eLiu, J., Jiang, L., He, S., Zhang, J. \u0026amp; Shao, W. Recent progress in PNIPAM-based multi-responsive actuators: A mini-review. \u003cem\u003eChemical Engineering Journal\u003c/em\u003e \u003cstrong\u003e433\u003c/strong\u003e, 133496 (2022).\u003c/li\u003e\n\u003cli\u003eStuart, M. A. C. \u003cem\u003eet al.\u003c/em\u003e Emerging applications of stimuli-responsive polymer materials. \u003cem\u003eNature Mater\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 101\u0026ndash;113 (2010).\u003c/li\u003e\n\u003cli\u003eShen, Z., Chen, F., Zhu, X., Yong, K.-T. \u0026amp; Gu, G. Stimuli-responsive functional materials for soft robotics. \u003cem\u003eJournal of Materials Chemistry B\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 8972\u0026ndash;8991 (2020).\u003c/li\u003e\n\u003cli\u003eWang, H.-X. \u003cem\u003eet al.\u003c/em\u003e Thermal-Responsive Hydrogel Actuators with Photo-Programmable Shapes and Actuating Trajectories. \u003cem\u003eACS Appl. Mater. Interfaces\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 51244\u0026ndash;51252 (2022).\u003c/li\u003e\n\u003cli\u003eFeng, X.-F. \u003cem\u003eet al.\u003c/em\u003e A multi-stimuli-responsive actuator for efficient thermal management and various biomimetic locomotion. \u003cem\u003eCell Reports Physical Science\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 101588 (2023).\u003c/li\u003e\n\u003cli\u003eKotikian, A. \u003cem\u003eet al.\u003c/em\u003e Innervated, Self-Sensing Liquid Crystal Elastomer Actuators with Closed Loop Control. \u003cem\u003eAdvanced Materials\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 2101814 (2021).\u003c/li\u003e\n\u003cli\u003eTang, X. \u003cem\u003eet al.\u003c/em\u003e Temperature self-sensing and closed-loop position control of twisted and coiled actuator. \u003cem\u003eSensors and Actuators A: Physical\u003c/em\u003e \u003cstrong\u003e285\u003c/strong\u003e, 319\u0026ndash;328 (2019).\u003c/li\u003e\n\u003cli\u003evan der Weijde, J., Vallery, H. \u0026amp; Babu\u0026scaron;ka, R. Closed-Loop Control Through Self-Sensing of a Joule-Heated Twisted and Coiled Polymer Muscle. \u003cem\u003eSoft Robotics\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 621\u0026ndash;630 (2019).\u003c/li\u003e\n\u003cli\u003eWeerathunga, H. \u003cem\u003eet al.\u003c/em\u003e Washable and Flexible All Carbon Electrothermal Joule Heater for Electric Vehicles. \u003cem\u003eAdvanced Materials Technologies\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 2201538 (2023).\u003c/li\u003e\n\u003cli\u003eTembei, S. A. N., Hessein, A., Fath El-Bab, A. M. R. \u0026amp; El-Moneim, A. A. A low voltage, flexible, graphene-based electrothermal heater for wearable electronics and localized heating applications. \u003cem\u003eMaterials Today: Proceedings\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 1840\u0026ndash;1844 (2020).\u003c/li\u003e\n\u003cli\u003eChan, Y. H. T. \u003cem\u003eet al.\u003c/em\u003e Portable Flexible Spherical Origami Joule-Heaters with Aerogel. \u003cem\u003eAdvanced Materials Interfaces\u003c/em\u003e \u003cstrong\u003en/a\u003c/strong\u003e, 2400544.\u003c/li\u003e\n\u003cli\u003eGeorgopoulou, A., Diethelm, P., Wagner, M., Spolenak, R. \u0026amp; Clemens, F. Soft Self-Regulating Heating Elements for Thermoplastic Elastomer-Based Electronic Skin Applications. \u003cem\u003e3D Printing and Additive Manufacturing\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, e828\u0026ndash;e838 (2024).\u003c/li\u003e\n\u003cli\u003eHuang, J. \u003cem\u003eet al.\u003c/em\u003e Stretchable and Heat‐Resistant Protein‐Based Electronic Skin for Human Thermoregulation. \u003cem\u003eAdvanced Functional Materials\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 1910547 (2020).\u003c/li\u003e\n\u003cli\u003eGeorgopoulou, A. \u003cem\u003eet al.\u003c/em\u003e Mechanoreceptive soft robotic molluscoids made of granular hydrogel-based organoelectronics. \u003cem\u003eMaterials \u0026amp; Design\u003c/em\u003e 114297 (2025).\u003c/li\u003e\n\u003cli\u003eCheng, H., Yue, S., Le, Q., Qian, Q. \u0026amp; Ouyang, J. A mixed ion-electron conducting carbon nanotube ionogel to efficiently harvest heat from both a temperature gradient and temperature fluctuation. \u003cem\u003eJ. Mater. Chem. A\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 13588\u0026ndash;13596 (2021).\u003c/li\u003e\n\u003cli\u003eAmstad, E. \u003cem\u003eet al.\u003c/em\u003e Mechanoreceptive soft robotic molluscoids made of granular hydrogel-based organoelectronics. Preprint at https://doi.org/10.21203/rs.3.rs-5196882/v1 (2024).\u003c/li\u003e\n\u003cli\u003eYuk, H. \u003cem\u003eet al.\u003c/em\u003e 3D printing of conducting polymers. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1604 (2020).\u003c/li\u003e\n\u003cli\u003eYoo, D. \u003cem\u003eet al.\u003c/em\u003e Gradual thickness-dependent enhancement of the thermoelectric properties of PEDOT:PSS nanofilms. \u003cem\u003eRSC Advances\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 58924\u0026ndash;58929 (2014).\u003c/li\u003e\n\u003cli\u003eKhan, Z. \u0026amp; Tanoli, M. A. Synthesis, Characterization, and In-Vitro Anti-Microbial and Anti-Oxidant Activities of Co+2, Ni+2, Cu+2 and Zn+2 Complexes of 5-chloro-2-hydroxybenzaldehyde-N-(2-oxo-1,2-dihydro-3H-indol-3-ylidene)hydrazone. \u003cem\u003ePakistan Journal of Chemistry\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 143\u0026ndash;149 (2015).\u003c/li\u003e\n\u003cli\u003eMasaro, L. \u0026amp; Zhu, X. X. Physical models of diffusion for polymer solutions, gels and solids. \u003cem\u003eProgress in Polymer Science\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 731\u0026ndash;775 (1999).\u003c/li\u003e\n\u003cli\u003eYilmaz, R. B., Chaabane, Y. \u0026amp; Mansard, V. Development of a Soft Actuator from Fast Swelling Macroporous PNIPAM Gels for Smart Braille Device Applications in Haptic Technology. \u003cem\u003eACS Appl. Mater. Interfaces\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 7340\u0026ndash;7352 (2023).\u003c/li\u003e\n\u003cli\u003eStrachota, B., Oleksyuk, K., Strachota, A. \u0026amp; \u0026Scaron;louf, M. Porous hybrid poly(N-isopropylacrylamide) hydrogels with very fast volume response to temperature and pH. \u003cem\u003eEuropean Polymer Journal\u003c/em\u003e \u003cstrong\u003e120\u003c/strong\u003e, 109213 (2019).\u003c/li\u003e\n\u003cli\u003eKim, S. Y. \u003cem\u003eet al.\u003c/em\u003e Sustainable manufacturing of sensors onto soft systems using self-coagulating conductive Pickering emulsions. \u003cem\u003eScience Robotics\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, eaay3604 (2020).\u003c/li\u003e\n\u003cli\u003eKoivikko, A. \u003cem\u003eet al.\u003c/em\u003e Integrated stretchable pneumatic strain gauges for electronics-free soft robots. \u003cem\u003eCommun Eng\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 1\u0026ndash;10 (2022).\u003c/li\u003e\n\u003cli\u003eHam, J., Han, A. K., Cutkosky, M. R. \u0026amp; Bao, Z. UV-laser-machined stretchable multi-modal sensor network for soft robot interaction. \u003cem\u003enpj Flex Electron\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 1\u0026ndash;9 (2022).\u003c/li\u003e\n\u003cli\u003eHoang, T. T. \u003cem\u003eet al.\u003c/em\u003e Soft robotic fabric gripper with gecko adhesion and variable stiffness. \u003cem\u003eSensors and Actuators A: Physical\u003c/em\u003e \u003cstrong\u003e323\u003c/strong\u003e, 112673 (2021).\u003c/li\u003e\n\u003cli\u003eGeorgopoulou, A., Vanderborght, B. \u0026amp; Clemens, F. Fabrication of a soft robotic gripper with integrated strain sensing elements using multi-material additive manufacturing. \u003cem\u003eFrontiers in Robotics and AI\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 615991 (2021).\u003c/li\u003e\n\u003cli\u003eTruby, R. L. \u003cem\u003eet al.\u003c/em\u003e Soft Somatosensitive Actuators via Embedded 3D Printing. \u003cem\u003eAdvanced Materials\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 1706383 (2018).\u003c/li\u003e\n\u003cli\u003eSpina, F., Pouryazdan, A., Costa, J. C., Cuspinera, L. P. \u0026amp; M\u0026uuml;nzenrieder, N. Directly 3D-printed monolithic soft robotic gripper with liquid metal microchannels for tactile sensing. \u003cem\u003eFlex. Print. Electron.\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 035001 (2019).\u003c/li\u003e\n\u003cli\u003eShih, B. \u003cem\u003eet al.\u003c/em\u003e Design Considerations for 3D Printed, Soft, Multimaterial Resistive Sensors for Soft Robotics. \u003cem\u003eFrontiers in Robotics and AI\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, (2019).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-flexible-electronics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjflexelectron","sideBox":"Learn more about [npj Flexible Electronics](http://www.nature.com/npjflexelectron/)","snPcode":"41528","submissionUrl":"https://submission.springernature.com/new-submission/41528/3","title":"npj Flexible Electronics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"smart materials, stimuli-responsive hydrogels, 3D printing, soft robotics","lastPublishedDoi":"10.21203/rs.3.rs-7149271/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7149271/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRobotic intelligence has advanced greatly in the past decade. Nevertheless, integrating embodied intelligent and responsive behaviour into soft robotic systems remains challenging because it typically requires bulky hardware for environmental feedback and decision-making. While soft materials like poly(N-isopropylacrylamide) (PNIPAM) offer potential for simplified material-based actuation through temperature-responsive motion, their slow response and high energy demands limit their use in closed-loop control systems. To overcome this limitation, we present soft PNIPAM-based actuators with integrated hydrogel-based Joule heating, enabling localized actuation without significantly altering the temperature within 1 cm of the actuator. The potential of the material is demonstrated by processing it in into a soft gripper that can lift up to three-fold its own weight with integrated capability to adjust its actuation in response to the gripped object. This design is well-suited for energy-efficient manipulation and sorting of delicate items, such as those found in automated packaging systems.\u003c/p\u003e","manuscriptTitle":"Programmable somatosensory soft robots","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-25 17:47:05","doi":"10.21203/rs.3.rs-7149271/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-12-03T05:03:11+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-11T04:27:54+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-09T15:53:26+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-09T12:08:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"323666492815745181823995623264515274818","date":"2025-10-28T05:59:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"46030055208750344051409197640153949148","date":"2025-10-28T05:21:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"64440448691123915667576803450030182976","date":"2025-10-27T06:06:59+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-18T00:39:19+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-21T13:57:30+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-21T11:55:21+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Flexible Electronics","date":"2025-07-17T12:42:44+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-flexible-electronics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjflexelectron","sideBox":"Learn more about [npj Flexible Electronics](http://www.nature.com/npjflexelectron/)","snPcode":"41528","submissionUrl":"https://submission.springernature.com/new-submission/41528/3","title":"npj Flexible Electronics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a3e3bbdf-052e-45dc-b56b-9a55d63dbbae","owner":[],"postedDate":"August 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":53418422,"name":"Physical sciences/Engineering"},{"id":53418423,"name":"Physical sciences/Materials science"}],"tags":[],"updatedAt":"2026-03-09T16:00:45+00:00","versionOfRecord":{"articleIdentity":"rs-7149271","link":"https://doi.org/10.1038/s41528-026-00558-0","journal":{"identity":"npj-flexible-electronics","isVorOnly":false,"title":"npj Flexible Electronics"},"publishedOn":"2026-03-07 15:57:29","publishedOnDateReadable":"March 7th, 2026"},"versionCreatedAt":"2025-08-25 17:47:05","video":"","vorDoi":"10.1038/s41528-026-00558-0","vorDoiUrl":"https://doi.org/10.1038/s41528-026-00558-0","workflowStages":[]},"version":"v1","identity":"rs-7149271","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7149271","identity":"rs-7149271","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}