Soft electroadhesive grippers with variable stiffness and deflection motion capabilities

Soft gripper robots provide superior safety, adaptability, and compliance compared to rigid robots. However, soft grippers must address inadequate stiffness and interference resistance. Soft pneumatic electroadhesion (EA) grippers with variable stiffness are potential options for addressing these difficulties. In this paper, we present a soft bionic gripper (SOBG) that resembles human finger movements, such as bending and deflection, employing pneumatic actuation, and whose stiffness is effectively decoupled from its position through a layer jamming-induced variable stiffness structure. By applying electroadhesive forces, the SOBG can perform complex motion tasks that would typically require a wrist joint, making them simpler to perform than with conventional flexible grippers. In addition, the SOBG can perform one-finger object manipulation to grasp flat, concave, and convex objects. To show the potential for more complex robotic applications, we evaluated each function independently by presenting a demonstration of cap-screwing, a material handling system, and an anti-interference research. The SOBG concept and solution proposed in this study may pave the way for the easy integration of EA into soft robotic systems and promote the wider use of EA technology.


Introduction
Soft robots place a greater focus on safe, flexible, and compliant working features than traditional rigid-body robots, Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
while the need for operations in human-centred applications and unstructured situations is rising [1][2][3][4][5].Compared to the rigid grippers, the soft grippers have the great flexibility and adaptability that are unique to soft robots, which is very useful for grabbing complex-shaped items.However, soft robots have the drawbacks of insufficient stiffness and inadequate interference resistance.Therefore, relevant researchers have introduced the concept of variable stiffness by integrating the variable stiffness structure with soft robots: when the work requires excellent adaptive performance and human-computer interaction safety performance, the structural stiffness can be reduced.When the task calls for increasing the ability to withstand external interference and the maximum load capacity, this may be accomplished by increasing the overall stiffness.Variable stiffness enables soft robots to retain their unique benefits while partly compensating for their shortcomings.Consequently, it is crucial to develop a soft robot with variable stiffness.Pneumatic-based soft bionic fingers have been and are still widely used because of the simplicity of their structure, the absence of contaminants, and the ease of maintenance.However, it is impossible to acquire a solid hold with the majority of soft pneumatic grippers now available without enveloping things and applying pressure.Furthermore, the performance of soft pneumatic grippers is severely constrained, particularly when it comes to taking up flat items [6].
In contrast to the traditional rigid actuator, as soft robots are able to bend, stretch, and twist, the actuators of soft robots must be comprised of a material that is soft and easily deformable and can adapt to the surrounding environment to a certain extent.There are now three prevalent soft robot actuation technologies, namely tendon driven [7,8], smart materials driven [9][10][11][12], and variable-pressure fluid driven [13][14][15].Fluid driven under variable-pressure is typically composed of soft materials, and actuator control can be achieved by varying the pressure of the fluid within a certain range.According to the medium utilized, fluid-driven soft robots can be divided into liquid-driven and gas-driven systems [13][14][15].Because the liquid's compression ratio is low and its reaction speed is quick, it offers excellent application potential in the area of soft grippers.Due of the impossibility of ignoring the liquid's bulk, modelling and motion control are strongly influenced by gravity.Soft robots were originally designed using pneumatic soft robotics [16].And because of their light weight, high efficiency, lack of pollution, strong environmental adaptability, etc, as well as the fact that they can be driven without ferromagnetic or electronic components, there are no activity components, they are soft, and they are reliable in harsh conditions such as intense radiation [17], electromagnetic interference [18], dust pollution, and impact [19].It has been extensively investigated and has a crucial role in the study of soft robotics.Since soft cavities of pneumatic soft robots are often composed of lighter soft materials, soft actuators have the inherent benefits of being soft, lightweight, and inexpensive, and may be compatible with pneumatic/hydraulic control systems already in place.This makes it appropriate for human-computer interaction applications in the domains of service robots [20], automated manufacturing [21], sports support devices [22,23], medical rehabilitation equipment [24], and aircraft [25], where human-machine interaction is increasingly prevalent [26].With the majority of soft pneumatic grippers available currently, it is impossible to obtain a secure grasp without enveloping objects and exerting pressure.Moreover, the performance of flexible pneumatic grippers is severely limited when it comes to picking up flat objects [6].
As the complexity of the objects being grabbed increases, scientists have combined electroadhesion (EA) with soft grippers [6].EA typically consists of two spaced electrodes implanted in an insulating material, with a high voltage (in the range of kV) applied to the electrodes such that the static charge accumulates on the object's surface, generating EA [27][28][29].Using the EA principle, researchers have built a soft gripper that is inexpensive, simple in structure, does not easily harm items, has strong electrical controllability, and can grasp flat objects that are difficult to grasp with conventional stiff grippers [30].In contrast to magnetic adsorption, negative pressure adsorption, and other adsorption techniques, EA relies on the polarization reaction of dielectric materials in the electric field, and the source of EA force is straightforward; EA is highly adjustable, and it can be regulated by adjusting the electric field voltage and current [31][32][33][34].
EA phenomena using soft electrodes have also been used to soft grippers [35].This kind of soft manipulator combines EA technology with other technologies and can adapt the EA by adjusting the applied voltage [36,37].Through the integration of dielectric elastomer actuators, a simple, quiet, and responsive soft gripper with a folding function that can alter the output EA force by altering the driving voltage was designed in [38].Although soft grippers integrated with EA technology have the advantages of low cost, simple structure, resistance to object damage, and good electrical controllability, as well as the ability to absorb thin objects, etc, the current robot gripper based on electrostatic adsorption has two major drawbacks: (1) it cannot accomplish excellent adsorption via conformal fit with complicated curved surface objects, consequently reducing the EA force; (2) during the gripping operation, it cannot retain its own deformation to maintain a good fit with the object's surface, making it difficult to establish reliable adsorption.How the EA gripper achieves conformal fit with irregular objects with complicated curved surfaces, as well as how to enhance its own stiffness and keep its own shape to maintain a good fit during adsorption, must thus be investigated in further detail.
Variable stiffness technology can increase the structural rigidity of soft robots so that they can function normally even in the presence of external forces [39], and can also adjust the corresponding stiffness according to different usage requirements; due to the degree of adhesion between the EA pad and the object's surface, EA systems can typically and effectively perform activities requiring the grabbing of objects; the closer the fit, the better the grasping ability.When employing EA technology to finish the adsorption of irregularly shaped things, the EA pad must have excellent flexibility so that it can complete the bonding of different portions as the object's shape changes.EA pads may conformal without warping or separating during deformation.EA pads must thus offer two different stiffnesses.As a consequence, the use of variable stiffness in conjunction with EA technology may improve EA performance [35].
Traditional rigid manipulator's components are often stiff, allowing for more torque and more accurate control [40], but the soft robot's preparation material is typically silica gel substance.Due to the poor rigidity of the silica gel substance, it is incapable of withstanding substantial forces.The load and torque is high, and it is easily deformed by external force.Therefore, the fundamental topic of soft robot research is how to increase the stiffness of the soft manipulator so that it can keep a reasonably stable form and produce a high torque.
Currently, the methods for adjusting the stiffness of soft robots fall into four categories: particle jamming, magnetic fluids, phase-change alloys with low melting temperatures, and layer jamming [41][42][43][44][45][46][47][48][49][50][51].Even if the form and structural design of layer jamming restricts its application range, it still has a high reaction speed, which is beneficial for applications demanding quicker response times [43].Layer jamming is the use of negative pressure to raise the forces (such as friction and extrusion force) of the layer in order to improve the overall stiffness.Based on layer jamming technology [46], designed a less invasive surgical manipulator.The manipulator features a thinwalled tubular mechanism with an adjustable stiffness that can serve as the primary body of some micro-snake manipulators used in surgery.When the robotic arm is just inserted into the human body, it can be in an extremely flexible condition, therefore preventing further harm to human body.And when the operational portion of the robotic arm is fully integrated into a human body, it will be changed into a more rigid condition to satisfy the demands of maintaining and operating the robotic arm.Layer jamming structures are exceptionally resistant to bending [52]; furthermore, they have a naturally thin and lightweight form factor, making them suitable for integration into soft robotic manipulators and wearable devices, which frequently require a high maximum bending stiffness and a low physical profile, both in relation to mass [53].
This research aims to build a soft bionic gripper (SOBG) that incorporates a pneumatic soft finger, a variable stiffness layer jamming structure, and a soft electroadhesive (EA) pad in order to overcome the limitations of soft gripper problems.The pneumatic soft finger is driven by air pressure, which allows SOBG to imitate the movement of human fingers, and realize the bending and deflection functions; the variable stiffness structure based on the principle of layer jamming is composed of a soft cavity and several layers of fillers, and the stiffness can be adjusted by applying different negative pressures; the soft EA pad is primarily comprised of conductive fabric electrodes and dielectric insulating layer, the electric field created by delivering a high voltage to the electrode can polarize the object's surface and generate electrostatic force, hence enabling the object's adsorption and gripping functions.
The contents are organized as follows: in section 2, the concept design and fabrication details of the SOBG are described; section 3 presents simulation and experiment analyses of SOBG bending, deflection, variable stiffness, and EA force evaluation characterization; in section 4, case studies of the demonstration of cap screwing, the object handling system, and the anti-interference experiment are detailed.Conclusions and future work are outlined in section 5.

SOBG concept and working principle
We intend to build and develop a SOBG that can mimic human finger movement and have bending and deflection motion, actively change its shape to adapt to different forms and adjust its stiffness while activated, and grasp objects of varying sizes and shapes.In order to do this, the SOBG consists of a pneumatic soft finger, a variable stiffness layer jamming structure, and a soft EA pad, as seen in figure 1(a), with the main structural dimension depicted in figure 1(b).The primary components of the pneumatic soft finger are a bending portion and a deflection segment with a folded multi-air chamber structure.There are nine bending air chambers and three deflection air chambers, and the tip of the bionic finger is a spherical structure that resembles human fingers.Utilizing a variable stiffness layer jamming structure generates variable stiffness capabilities.Figure 1(c) depicts the layer jamming mechanism's fundamental concept.When negative pressure is applied to layers that are filled and compressed inside the variable stiffness cavity, the layers become cohesive and increase stiffness [45].SOBG is thus not only capable of bending (figure 1(d)), but also deflecting (figure 1(e)), enabling it to mimic the movement posture of the human finger.Integrating a soft EA pad, SOBG is not only capable of handling concave and convex materials, but also flat things.

SOBG fabrication
The creation of the SOBG device entails the fabrication of a pneumatic soft finger, a variable stiffness layer jamming structure, and the integration of a soft EA pad: Moulding pneumatic soft finger.The pneumatic soft finger mould consists of the top and bottom moulds, which were 3D printed from acrylonitrile butadiene styrene.Dragon Skin 30 (Smooth-On Inc., USA) was degassed for 20 min, the mixture was placed into the top mould and the bottom mould separately, the moulds were placed in an oven to cure at 50 • C for 4 h.The upper and bottom of the pneumatic soft finger was removed from the moulds.The combination of Dragon Skin was then utilized to re-adhere the two parts of the pneumatic soft finger.Use scissors to cut out the form of the pneumatic soft finger's bottom from the thermoplastic polyurethane elastomer composite fibre fabric limiting layer.Align and fit the pneumatic soft finger to the limiting layer, and then use Dragon Skin mixture to attach the limiting layer to the pneumatic soft finger's bottom.In order for attaching the SOBG, the root of SOBG must contain an extension part for fixing, and 3D printed moulds for the fabrication of pneumatic soft fingers are shown in figure 2(a).
Moulding variable stiffness layer jamming structure.Dragon Skin mixture was poured again into a 3D printed variable stiffness layer cavity ABS mould (figure 2(a)), waited for the mixture to cure and remove it from the mould, then layer jamming filler (papers: 0.2 mm in thickness, 120 mm in length, and 36 mm in width) were placed in the cavity, and Dragon Skin mixture was used to seal the cavity.The variable stiffness chamber is 2 mm thick, 126 mm long, 47 mm wide, and 7 mm high.Cavity height is 3 mm.
Soft EA pad fabrication.The electrically conductive, soft, and stretchable EA pad (Conductive-knit made of 94% nylon and 6% elastomer and plated silver, Kitronik Ltd, UK) was  The limiting layer, the EA layer, the jamming layer, and the pneumatic soft finger are all coated with Dragon Skin 30, using the same silicone bonding compatibility to bind them together.

Bending and deflection characterisation
By adjusting its air pressure to achieve different levels of bending deformation, it can accommodate a variety of object sizes and shapes.To evaluate the bending characteristics of the developed SOBG, a customized bending performance test rig was constructed, as shown in figure 3(a), and a coordinate paper was used to measure the bending angles.By use of a pump, pneumatic triplexs (to filter the air), and solenoid valves, the SOBG was pressurized (to regulate the air pressure).An Arduino (Arduino Mega2560, Hesai Technology Inc., Shenzhen, China) were utilized to control PWM drivers (PWM driver, Shenzhen Yueyu Electronic Technology co., LTD, Shenzhen, China) and solenoid valves (Three position two way solenoid valve, RDPC Rongda Pneumatic Co., LTD, Ningbo, China) for adjusting the pressure of the SOBG.The internal pressure of the SOBG was recorded using a differential pressure gauge (Manometer, HT-1895, China).When the internal pressure of the SOBG was raised from 5 kPa to 60 kPa in 5 kPa increments.Figure 4 depicts only the bending  deformation of SOBG at 10 kPa to 60 kPa with a 10 kPa increase in pressure in order to more clearly illustrate the degree of angular variation.
Figure 3(b) demonstrates that the greatest bending angle reached 205 • when the internal pressure was 60 kPa.We also performed simulations using Abaqus/CAE (ABAQUS Inc., USA) utilizing the finite element method (FEM) to evaluate the SOBG's bending performance.Figures 3(b) and 4 depict the simulation and experimental results of the SOBG's bending performance.Simulation and experiment results reveal that the angle of bending rises with increasing inflation pressure in both the simulation and experiment, with a maximum deviation of 8.33 • at 60 kPa.
To analyse the deflection properties of the developed SOBG, a customized deflection performance test rig was constructed, as shown in the inset of figure 5, along with a coordinate paper for measuring the deflection angles.Air source control system was identical to bending performance test; however, the internal pressure of the SOBG deflection section increased by 10 kPa, from 20 to 60 kPa.As shown in figure 5, the greatest deflection angle of the SOBG at an internal pressure of 60 kPa was 15 • .Figures 5 and 6 depict the simulation and experimental results of the SOBG's deflection properties.The angle of deflection rises with increasing inflation pressure in both simulation and experiment, with a maximum difference of 0.93 • at 40 kPa.
All tests were conducted within an enclosed, clean box.According to a weather station, the temperature was 19.7 ± 0.1 • C, and the relative humidity was 61 ± 1%.The SOBG was measured at different inflation pressures; at each inflation pressure, five bending and deflection tests were conducted.

Variable stiffness verification
In order to test the variable stiffness performance and feasibility of the variable stiffness structure, a stiffness experiment was conducted; the experiment setup is shown in figure 7(a).Using a vacuum pump (ES-3910, Ganzhou, Jiangxi Province, Chaofei Ltd), the internal pressure of the variable stiffness cavity was reduced by 10 kPa, from 0 kPa to −50 kPa, after the variable stiffness structure was mounted on an aluminium profile frame.In the experiment, a coordinate paper was used to measure the displacement of the end of the variable stiffness structure, and the value was recorded on the dynamometer (SJ10N, SANLIANG, Japan) when the end of the variable stiffness structure reached the set position.The stiffness test results are displayed in figure 7(b).Compared to 0 kPa, −50 kPa exhibited a 329.3% increase in stiffness.
FEM was also performed on the variable stiffness structure, and figure 7(c) depicts the simulation results of a variable stiffness structure at 0 kPa and −50 kPa; figure 7(b) compares the experimental value to the simulation value; both simulation and experiment results confirm the variable stiffness feasibility of the variable stiffness structure.In both simulation and experiment, the stiffness increases with increasing negative pressure, with a maximum deviation of 5.5 N m −1 at −40 kPa.
In the actual test, when an external force is applied to the end of the variable stiffness layer to make it move, there will be friction between papers and between papers and silicone in the variable stiffness layer.This is the primary cause of the simulation error of variable stiffness.However, if this factor is incorporated into the simulation, it will become extremely complex.
Consequently, this factor is not accounted for in the simulation.This will make the measured stiffness greater than that acquired by simulation, and the friction will increase as the negative pressure increases, resulting in a greater difference between the simulation and experimental data as the negative pressure increases.The second reason why there will be shape differences between the paper and the fabrication is because there will be differences between the paper and the cutting.The second reason is that there will be differences between the papers, with more or less shape differences resulting from cutting.Future research will examine a more precise simulation model.The SOBG developed in this research is driven by air pressure, with conductive fabric serving as the electrode in the EA system.Due to the excellent elasticity of the conductive fabric, when the form of the electrode chamber of the dielectric insulating layer changes, the structural size of the conductive fabric material electrode also changes, resulting in a change in the EA force.This research simulates the electrode deformation induced by the bending of SOBG driven by air pressure and analyses the changes in electrode size during the bending deformation of the SOBG.In conjunction with previous study findings, the variations in soft bionic finger EA performance throughout the working process were investigated.In accordance with SOBG bending simulation and experimental results described previously (in 3.1 bending and deflection characterisation), the electrode was measured and evaluated under the maximum bending condition, and variations in electrode geometry and electrode spacing caused by the bent SOBG were obtained as shown in tables 1 and 2.
Figure 8 depicts the electrode numbers, electrode spacing numbers, and geometric dimensions of the electrodes without deformation.Since the electrodes were constructed from a soft, conductive fabric, nonlinear deformation will occur during loading.To do this, the difference in size between the electrodes before and after deformation was calculated.
According to the analysis of the data in tables 1 and 2, the maximum deformation of the SOBG caused the electrode width to decrease to varying degrees, with the electrode thickness in the middle of SOBG experiencing the greatest degree of reduction and the electrode width reduction degree at both ends of SOBG being the smallest.The results, whose degree of reduction reached up to 20.13%, was produced principally by the fact that both ends of the SOBG pressed the central electrode during the bending process.When the electrodes at both ends were squeezed from the centre and lack support on one side, this issue emerged.When the SOBG was bent, the distance between the electrodes is similarly decreased, with a maximum reduction of 69.3% of the original electrode spacing.
When the electrode spacing and the number of electrodes are held constant, increasing the width of the electrode will have a negative impact on the EA force; increasing the electrode spacing will also result in a significant decrease in EA force, but too small an electrode spacing will also increase the risk of breakdown [54]; the impact of electrode thickness on the EA force is very minimal [55].Therefore, to maintain an adequate and steady EA force per unit area, the width of the electrodes should be decreased and the density  of electrode distribution should be raised in an acceptable manner.
According to the data analysis, as the degree of bending of the SOBG created in this work increases, the following three conditions will occur: (1) The electrode width is decreased, with the centre electrode seeing a more pronounced reduction.As the spacing between electrodes increases, it becomes more difficult to break down the electrode.At this point, it is possible to increase the driving voltage appropriately.(2) The increase in thickness of the electrodes at both ends of the SOBG is more pronounced, although this has minimal impact on EA [55].However, the increase in thickness will cause the surface of the EA layer to expand at the electrode site, affecting the tightness of contact between the EA and substrates and reducing the EA force.If the bulging is too extreme, the EA force will be greatly decreased.(3) The electrode spacing is reduced, with the reduction at both ends of the SOBG being comparatively smaller.Since the central portion of SOBG curves with almost uniform curvature, the electrode spacing at the middle point is essentially same.Note that electrode spacing should not be too narrow, i.e. the degree of bending of SOBG should not be too great while functioning; otherwise, it will be prone to failure.

EA force measurement.
The normal and shear forces of an EA pad have been examined to identify the relationship between force and applied voltage.As shown in figure 9(a), a setup for measuring shear EA forces was developed.For the shear EA force test, papers were employed as the substrate and were absorbed on the EA pad after being charged for 30 s at different voltages using a high voltage amplifier source (HVA, EMCO E60, XP Power Ltd, USA).The HVA output was manually controlled through a variable power source.For the shear EA force experiment, a soft EA pad was positioned vertically and fastened to a flat acrylic plate.Ensure that the whole paper is next to the EA pad while measuring.EA pads were operated at various voltages, and the mass of the most paper that EA can adsorb can be used to represent the shear EA force by calculating the amount of paper, the total mass absorbed, and the shear EA force.The applied voltage ranges from 2 to 4.8 kV, with a 0.4 kV increment, and five tests for the EA pad were undertaken.Figure 9(b) depicts the test results for shear EA forces (the inset depicts the schematic diagram of the shear EA force test rig).It is evident that the shear EA forces, in particular, increase significantly as the driving voltage increases; the shear EA forces can reach a maximum of 1.24 ± 0.013 N under 4.8 kV, and the shear EA force increased by 38.29% between 2 kV and 4.8 kV.As seen in figures 9(c), a normal EA force measuring setup was developed to evaluate the normal EA force of the EA pads.The normal EA force setup was the same as the shear EA force configuration, with the EA pad placed face downward.As the substrate for the normal EA force test, papers were employed.In the figure 9(c) insert, the schematic design of the normal EA force test rig was shown.Figure 9(c) depicts the test results for the EA normal force; the EA pad was subjected to five tests.The average EA force increases by 34.3% from 2 kV to 4.8 kV.
In addition, all tests were conducted at 19.6 ± 0.1 • C, 65 ± 1% relative humidity, and 1020.5 ± 0.2 hPa ambient pressure using a weather station.

EA robustness test.
We evaluated the resistance of every stage of EA-treated fingers to external stimuli in order to discuss the robustness of EA in practical applications.The SOBG is used to grasp a piece of paper, a cylinder of 85 mm diameter (height 25 mm, weight 16 g), and a beaker of 50 mm diameter (height 70 mm, weight 11 g) under the conditions of 0 kPa, 35 kPa, and 45 kPa, respectively, such as the EA robustness test rig for grasping an cylinder shown in figure 10.
Record the value of the dynamometer by pushing the affixed object with the dynamometer (SH-100, SHSIWI, Shanghai) until the object detaches from the SOBG.The 1.5 N, 1.35 N, and 1.12 N are the forces exerted by SOBG on paper at 0 kPa, a cylinder with a diameter of 85 mm at 35 kPa, and a beaker with a diameter of 50 mm at 45 kPa, respectively.Therefore, it is known that EA has anti-interference properties in the gripping process.
In addition, all tests were conducted at 28.6 ± 0.1 • C, 66 ± 1% relative humidity, and 1019.5 ± 0.2 hPa ambient pressure using a weather station.

Cap screwing demonstration using a gripper with two SOBGs
SOBG is driven by air pressure, which enables it to emulate the movement posture of a human finger and achieve bending and deflection functionality, which possesses more freedom than typical pneumatic gripper and accomplish complex motion tasks (such as cap screwing) that are difficult for typical soft grippers [30].
Humans need at least two fingers to unscrew a bottle cap, and figure 11 depicts the force analysis performed on the bottle cap unscrewing operation.
F is the finger's torsion force; R is the cap's radius; M is the twisting torque; and m is the minimal torque necessary to twist the cap, where µ 1 and µ 1 are the coefficients of static and dynamic friction between the fingers and the cap, respectively, and F N is the normal pressure exerted by the fingers on the cap.
According to the force analysis, the following can be determined: (1) when F is greater than f 1 , the finger and bottle cap will glide relative to one another.Then, the torque required to screw the bottle cap is M = (F − f 2 ) × 2R.When M > m, the gripper can twist the bottle cap; otherwise, it cannot; nonetheless, the fingers and the bottle cap do not typically move relative to one another when the bottle cap is twisted by hand.
(2) When F ⩽ f 1 , the finger and bottle cap will not slide relative to each other; thus, the finger and bottle cap can be considered a unit, and the static friction f 1 is the internal force of the finger and bottle cap, so it can be disregarded.Currently, the torque required to twist the bottle cap is M = F × 2R.If M > m, the gripper can twist the cap; otherwise, it cannot.
To evaluate the SOBG's bending and deflection functionality, a cap-screw experimental testing rig was constructed.The system's schematic design is seen in figure 12. Arduinos (Arduino Mega2560, Hesai Technology Inc., Shenzhen, China) were utilized to control PWM drivers and solenoid valves for adjusting the pressure of the SOBGs.The air pump supplied both SOBGs with air (ZR370-02PM, air pump, Zhirong Vacuum Equipment Co., LTD, Shenzhen, China).The air outlet end of the solenoid valves of the driver system was connected to the air pipes of the SOBGs.Sending solenoid control instructions with an Arduino, and the instructions can be used to control the SOBGs' bending and deflection motions in order to drive gripper movement.
The bottle cap screwing experiment will include a twofinger gripper using two SOBGs.The distance between the two SOBGs' centres is 112 mm.Arduino was utilized to control the bending and swinging of the two SOBGs.Inflate the SOBGs to bend and clamp the bottle cap.Swing the fingers to twist the bottle cap.Swing to a specified position to release the bottle cap.Swing the SOBGs back to their original position in preparation for the next twist.All program codes were developed using Arduino IDE.Cap screwing demonstration can be seen in supplemental video 1.

Single SOBG material handling
We studied further the SOBG's capacity to grasp and transport flat things from one location to another.This demonstrates that only one single SOBG can effectively influence things.Figures 13(a) and (b) depicts the schematic design of the single SOBG material handling system that was developed.Place a 0.9 g rectangle paper (the length, width, and height of the paper are 55 mm, 40 mm, and 1 mm) in the given area on a table, fix a SOBG on a linear rail, move the SOBG close to the rectangular paper until it completely touches the paper, and then move the SOBG up.The paper can be adsorbed on the SOBG after applying 4.8 kV to the EA through a HVA (EMCO E60, XP Power Ltd, USA).After the SOBG had risen to the designated height, the deflection air cavity was inflated to an air pressure of 60 kPa, and the SOBG began to swing.The SOBG was then lowered until it was close to the table, and the EA was turned off to allow the paper to be released and dropped into the designated area.The SOBG was inflated using a pump, pneumatic triplexs (to filter the air), and solenoid valves (to regulate the air pressure).Arduino (Arduino Mega2560, Hesai Technologies Ltd, Shenzhen, China) was used to operate PWM drivers and solenoid valves for regulating the pressure of the SOBG.
For material handling tasks involving a single SOBG, the SOBG should be kept level so that the EA can effectively contact and grasp planar objects.Due to the influence of the gravitational field, however, SOBG will be in a state of sagging bending at low rigidity, making it difficult to grasp flat objects.In this experiment, a single SOBG finger with a soft state and a rigidity of 11.6 N m −1 (under 30 kPa of negative pressure) was used to hold planar objects.Low rigidity makes it difficult to hold flat objects, as demonstrated by the experiment results.Supplemental video 2 provides a demonstration of solitary SOBG materials handling.
SOBG, which integrates the soft pneumatic finger and the soft EA pad, is able to promote both EA and pneumatic technologies.Without activating the EA feature, SOBG may pressurize the typical soft pneumatic grasping function, which depends primarily on gripping force to grasp objects.It can also move big flat, thin, concave, and convex surfaces, something the typical soft pneumatic cannot accomplish due to its inability to wrap around the widest point.With the EA enhanced gripping mode, the SOBG gripper will address the deficit of classic soft pneumatic grippers while preserving the shape adaptable capabilities.Under 30 kPa, the SOBG cannot capture a convex object with a 10 1 m −1 curvature (120 mm straight side, 55 mm height, 5 g weight).When the EA is turned on and the item is under 4.8 kV, the convex object gripping illustrated in figure 13(c) can be seen in the supplemental movie 2. As a consequence, the gripper's dexterity and flexibility have improved dramatically.

Disturbances
SOBG incorporated a structure with variable stiffness based on the concept of layered jamming, and the stiffness can be adjusted by applying varying negative pressures.In the high stiffness mode, SOBG should be more resistant to disturbance loads than in the compliant mode.As illustrated in Placing the finger vertically and attaching it to a frame, adjust the variable stiffness structure using a vacuum pump (ES-3910, Ganzhou, Jiangxi Province, Chaofei Ltd) at 0 kPa, −10 kPa, −20 kPa, and −30 kPa, respectively.The inflation pressure of the SOBG was measured using a differential pressure gauge (Manometer, HT-1895, China).Then, using a hand to move the end of the finger horizontally for a defined distance (in this experiment, 50 mm was selected), release the hand, and the finger will return back and forth until it stops.The recording time period is 0-2 s (since the SOBGs with various negative pressures can cease swinging during this time), record the maximal displacement of the movement and the corresponding time, and the findings of the disturbance experiment are shown in figure 14(b).
Figure 14(b) shows that the SOBG at a high stiffness state greatly reduces the speed and amplitude of motion at which the SOBG is able to have bending motion compared to the SOBG at a low stiffness state, which exhibits good anti-disturbance characteristics.It took roughly 1.73 s at a pressure of 0 kPa to stabilize 0 mm of the SOBG, but only 1.18 s at a pressure of −30 kPa.The increase in the rate is 31.9%more rapid than the initial value.By adjusting the internal pressure of the signal supplied to the variable stiffness cavity, it is feasible to vary the speed at which the system reacts to the application of a disturbance force, as shown by the results.Supplemental video 3 provides a presentation of the SOBG disturbances experiment in various states of stiffness.
According to the proposed SOBG's construction, the variable stiffness structure only works on the bending motion, therefore the anti-interference cannot be improved for the deflection motion.Future work will focus on improving the anti-interference capacity by optimising the variable stiffness structure and introducing variable stiffness into the deflection motion.

Discussion
Using a pneumatic actuation system, the SOBG that simulates human finger function, including bending and deflection, and stiffness, can be successfully decoupled from its poison.EA enables one-finger object manipulation to grasp flat, concave, and convex objects.This enables the robot to execute complex motion tasks that would typically require a wrist joint, such as cap fastening.Due to the multiple functions of SOBG, however, miniaturization is challenging; therefore, future research will concentrate on miniaturization.Simultaneously, the preparation of the SOBG is insufficient, making it difficult to simulate or develop an accurate model, and consequently, to accomplish precise control.In the future, it will be necessary to rectify the deficiencies of the preparatory procedure.

Conclusions and future work
In this study, a SOBG that can reproduce human finger movement and perform bending and deflection activities, as well as having variable stiffness functionality and the ability to lift flat, concave, and convex items using EA, was constructed.The SOBG comprises of three major components (a soft pneumatic finger, a layer jamming induced variable stiffness structure, and an EA pad), which were created using a cost-effective and easy-to-implement technique.The soft pneumatic finger can simulate the movement of the human finger, and realize bending and deflection functions; the variable stiffness structure based on the layer jamming principles is composed of a soft cavity and several layers of fillers, and different negative pressures are applied and realize the stiffness adjustment; the soft EA pad is primarily composed of conductive fabric and dielectric insulation layer, which can realize the object's adsorptive function; the performance of the SOBG in terms of bending, deflection, and variable stiffness was investigated.The SOBG proposed that resembles human finger movements, such as bending and deflection, and whose stiffness is effectively decoupled from its position.Using EA forces, the SOBG can perform intricate motion tasks that would typically require a wrist joint, making them simpler to perform than with conventional flexible grippers.Furthermore, it can perform onefinger object manipulation to grasp flat, concave, and convex objects.
The main contributions of this research consist of: (1) the development of a SOBG that can mimic human finger action and execute bending and deflection activities, actively change its shape to adapt to diverse shapes and vary its stiffness while activated, and lift flat, concave, and convex objects, (2) simulation and methods for elucidating the effect mechanism on SOBG bending and deflection, as well as variable stiffness performance, (3) analysis of electrode deformation for revealing EA performance, and (4) the development of a control system for the cap-screw, a material handling system, and a disturbance test system.Future study will include modelling and improving the SOBG design for improved performance, such as lifting bigger loads.Additionally, SOBG design scalability will be explored.Future research will also focus on developing a more accurate simulation model and increasing interference resistance by optimising the variable stiffness structure and incorporating variable stiffness into the deflection motion.

Figure 1 .
Figure 1.SOBG concept design.(a) A 3D diagram of the SOBG (The figure up top is an exploded depiction, and the assembly schematic is at the bottom.).(b) Main structural dimension.(c) Principle functioning of the variable stiffness structure.(d) A demonstration of bending motion and its corresponding actuation part.(e) A demonstration of deflection motion and its corresponding actuation part.

Figure 2 .
Figure 2. The fabrication of SOBG.(a) 3D printed moulds for the fabrication of pneumatic soft fingers.(b) 3D printed moulds for variable stiffness layer jamming structure.(c) Schematic diagram of the SOBG in 3D.(d) The SOBG prototype.(e) Schematic diagram of the EA pad.

Figure 3 .
Figure 3.The bending performance test of the SOBG.(a) A customized bending performance test rig.(b) Results from simulation and experiment regarding the SOBG's bending performance test data curve.

Figure 4 .
Figure 4.The simulation and experimental results of the SOBG's bending performance.(a) The simulation's results.(b) The results of the experiment.The black scale bars denote 10 mm.

Figure 5 .
Figure 5. SOBG deflection performance test data curve generated from simulation and experimental results.Inset displays a deflection performance test setup.

Figure 6 .
Figure 6.The simulation and experimental results of the SOBG's deflection performance.(a) The simulation's results.(b) The results of the experiment.The black scale bars denote 10 mm.

Figure 7 .
Figure 7.The performance and feasibility evaluation of the variable stiffness structure.(a) A variable stiffness performance evaluation test rig.(b) Comparison of experiment value with simulation value diagrams.(c) The 0 kPa and −50 kPa FEM simulation results for the variable stiffness structure.

Figure 8 .
Figure 8.The EA pad's electrode number, electrode spacing number, and geometric dimensions.

Figure 9 .
Figure 9.The driving voltage ranges from 2 to 4.8 kV, with 0.4 kV increments, for electroadhesive pad shear and normal force testing.(a) A shear EA force measurement setup.(b) The results of EA shear force testing (inset shows the schematic diagram of the shear EA force test rig).(c) A normal EA force measurement results (inset shows the schematic diagram of the normal EA force test rig).The error bars represent the standard deviation of each test's five results.

Figure 11 .
Figure 11.Two fingers for a task force analysis on cap screwing.(a) Front view.(b) Top view.

Figure 12 .
Figure 12.The cap-screw experimental testing rig.(a) Schematic diagram, where 1 is the bottle with a cap that needs to be unscrewed, 2 is the gripper with two SOBGs, 3 is the frame, 4 is the solenoid valves, 5 is the solenoid valve driver, 6 is the Arduino mega 2560 and (b) the system prototype.

Figure 13 .
Figure 13.Single SOBG material handling.(a) System diagram.(b) Physical setup.(c) Shape adaptive grasping of a convex object.The black scale bar denote 10 mm.

Figure 14 .
Figure 14.Disturbances experiment of SOBG in various states of stiffness.(a) Disturbances experiment test rig.(b) Disturbance experiment results.

Table 1 .
Variations in electrode geometry caused by the SOBG's maximal state of bending.

Table 2 .
Variations in electrode spacing caused by the bent SOBG.