Design, Control, and Applications of Granular Jamming Grippers in Soft Robotics

Abstract

Granular jamming grippers have emerged as a versatile solution in soft robotics due to their ability to manipulate objects of various shapes and sizes, earning them the label of “universal grippers”. They are composed of granular material confined within an elastic membrane that conforms to the object like a fluid and solidifies upon vacuum application, enabling a firm grip through friction and grain interlocking. This work provides a systematic review of the state of the art, addressing their physical principles, the influence of grain and membrane properties, performance characterization methods, and applications across diverse fields. Additionally, the main control variables of these grippers closely related to state variables used in control systems are discussed, along with the current knowledge gaps. Finally, five potential directions for future research are proposed.

1. Introduction

In the robotic field, grip and manipulation of objects are fundamental functions. A simplified definition of grasping is the ability to grasp and hold an object despite external disturbances. As for manipulation, it refers to the ability to exert forces on the object, achieving a rotation or translation in a frame of reference of the manipulator. In robotics, the end effector is the device located at the end of the robotic arm. It must be specifically designed to interact with the environment and perform specific tasks [1]. These devices are mainly classified into two categories: grasping elements usually called “grippers” and tools designed to perform specific tasks [2].

Most traditional robotic grippers are built using rigid components, mainly joints and linkages [3]. They can vary from simple two-fingered designs to more complex anthropomorphic hands, which try to replicate the structure and movement of the human hand, including fingers and sometimes a palm [4]. The idea behind choosing anthropomorphic grippers usually comes from the advantages that human hands offer versatility, precision, and adaptability. These features are important when robots are meant to operate in human-oriented environments, or when using teleoperation systems like smart gloves, or simply when handling a wide variety of objects. Despite these advantages, anthropomorphic grippers are quite complex in terms of mechanics, they also tend to struggle when dealing with soft or irregularly shaped items and often require highly precise control to even come close to matching the dexterity and strength of a human hand [5].

Simpson [6], in 1971, was probably the first author to suggest adding cavities of granular material to the gripping surfaces of traditional grippers. He found that adding these surfaces significantly improved grip stability because the surface had been deformed to the shape of the object. This increased the frictional forces between the powder particles, which consequently blocked them. A few years later, in 1978 [7], Schmidt presented a gripping jaw filled with some type of granules contained by a membrane. The exact material was not clearly specified. These granules adapted to the surface of the object, deforming slightly, and a pneumatic system evacuated the fluid to facilitate gripping.

Thanks to the advancement in the knowledge of soft and smart materials, a subfield of robotics called “soft robotics” has emerged. The article presented by Jun Shintake [8] condenses the state of the art related to soft robotics grippers. Grippers are classified by: controlled actuation, controlled stiffness and controlled adhesion. This article addresses the section of controlled stiffness “granular jamming”, where stiffness adjustment results from taking advantage of the transition from a “fluid” state to a “stuck” state. There are three distinct types of jamming investigated in soft robotics: granular, layer, and fiber. All three types of interference require a membrane [9,10].

Granular jamming grippers are based on reversible phase transition in which a material transitions from a fluid state to a disordered solid [11,12,13,14,15,16]. This shift, especially in macroscopic granular systems, is brought about by external forces such as vacuum pressure or volume changes. The mechanical integrity of the jammed state depends entirely on the properties of its intricate contact network and the forces transmitted through it [17,18,19]. Force networks represent how stress is transmitted by forming “force chains” (see Figure 1a) [16,20,21,22] in anisotropic “jamming” states, providing stiffness and the ability to resist stress. The Liu-Nagel [14,15,23] jamming diagram (see Figure 1b provides a theoretical framework for understanding these transitions, unifying disparate systems such as granular materials, foams, and glasses [12] within a parameter space defined by packing fraction, shear stress, and temperature. For granular materials, this diagram highlights how parameters such as density and applied stress control the transition, and how there is a special point (J) at zero temperature and stress, which marks the onset of sticking in frictionless particles.

This article addresses granular jamming grippers with bag structure, as shown in Figure 2. The gripping routine is generally summarized in four steps: first, the gripper approaches the object in a soft state; second, the gripper takes the shape of the object; third, the blockage is generated within the membrane evacuating the fluid (usually air); and fourth, the object can be manipulated.

Granular jamming grippers with bag structure share the basic sequence for object manipulation and additionally the same variables that affect the gripper performance. Figure 3 presents the input variables that affect the gripper. These variables must be considered in the gripper design and performance parameters typically used to evaluate the various technologies and variations that researchers apply to granular jamming grippers.

Figure 3 illustrates the conceptual framework linking input variables, gripper configuration, and performance indicators in granular jamming grippers. The input variables define the operating conditions that trigger the jamming transition, while the configuration parameters of the gripper determine how these inputs translate into macroscopic grasping behavior.

Specifically, negative pressure is the primary driver of the jamming transition. Higher vacuum levels increase particle interlocking, which directly enhances the holding force and success rate of grasps, but at the same time may increase the required activation force depending on the object geometry. Additionally, the magnitude of negative pressure governs the response time, since stronger vacuum systems can shorten the stiffening process. In parallel, the Z-shift, defined as the penetration depth of the gripper into the target object, controls the degree of geometric interlocking. Larger Z-shift values promote stronger confinement and thus higher holding forces and grasp reliability, although at the cost of higher activation forces and potential risk of object damage.

The intrinsic design of the granular matter strongly influences adaptability and load capacity. The material and morphology (particle shape, size distribution) regulate the internal friction and packing density, which impact both the maximum attainable holding force and the gripper’s ability to conform to irregular objects. Mixtures of grains and optimized fill percentage balance rigidity in the jammed state with fluidity in the unjammed state, affecting both success rate and response time.

The membrane constitutes the interface between the granular core and the external environment. Its material and thickness affect elasticity and sealing, which directly influence the activation force and the capacity to achieve airtight jamming. Moreover, the morphology (shape and size of the membrane) determines the range of objects that can be enveloped, thereby influencing grasp success rate and adaptability.

Finally, the reset method defines how the gripper transitions back to the unjammed state. Manual resets, positive pressure, or fluidization differ in efficiency and repeatability. These mechanisms have a direct impact on response time and long-term operational reliability, since fast and consistent resets are crucial for applications involving repeated grasping cycles. Collectively, these relationships clarify how input variables propagate through the gripper’s material and structural configuration to define the measurable indicators of performance: success rate, response time, activation force, and holding force.

Throughout this document, each of the variables presented in Figure 3 will be expanded, along with comparative tables of the different performances of the grippers, the applications that have been made for different fields, the support of simulation tools in the design of grippers, and a discussion of the existing knowledge gaps and possible future directions.

Amend et al. [25] demonstrated that the gripping force of a locking gripper is maximized when the target object diameter is approximately 50% of the gripper diameter, while the grip is most reliable when the object is between 30% and 70% of this range. These results highlight that size design is not merely a geometric consideration but a key factor directly affecting gripping adaptability and success. A gripper that is too large relative to the object can reduce contact efficiency and increase air consumption, while one that is too small may lack sufficient contact area to generate reliable gripping forces. Therefore, careful specification of gripper size is essential to ensure robust performance on a variety of objects, making it a central parameter in both experimental evaluations and commercial implementations of granular locking technology.

2. Fundamental Principles of Granular Jamming

Understanding granular jamming is a concept that facilitates the design of grippers with adjustable stiffness and adaptive capacity. This process occurs when a granular material changes from a fluid to a solid state reversibly upon applying a pressure differential [24].

2.1. Phase Transition Mechanism

Granular jamming is a phenomenon where a granular material, contained within a flexible non-porous membrane, is able to have a reversible transition between a low-stiffness fluid state to a high-stiffness solid state [26]; this transition is mainly induced by a negative pressure difference within the membrane (vacuum) [27].

In the unjammed state, there is a lack of mechanical coupling between the particles, so they move freely relative to each other due to low intergranular friction. When vacuum pressure is applied inside the membrane, the external pressure (atmospheric or hydrostatic) compresses the membrane and the granular material, increasing the density of the system and the friction between the grains, and it is considered that in the unjammed state there is an absence of an elastic limit [12].

After applying vacuum pressure and reaching the jammed state, the grains present a mechanical coupling that causes a resistance to sliding or rearrangement [14,19,21,28,29]; in this state, the mass of granular matter is able to distribute the force giving rise to a rigid material capable of supporting loads. This process is reversible, and when the internal pressure is restored, the grains are released and the material returns to its deformable state [30].

2.2. Grip Forces

Granular jamming grippers generate gripping force through a combination of three mechanisms, which are as follows:

• Static Friction: Static friction forces are due to the tangential tension in contact of the hardened gripper membrane with the surface of the object (see Figure 4A); the use of soft particles can intensify this friction by generating a “squeezing” effect on the object due to the reduction in volume when applying a vacuum inside the membrane [31]. When the gripper does not lock the object and only exerts force through static friction, the maximum holding force does not depend on the granular material used [32].
• Geometric interlocking: When the gripper in its flexible state deforms around the target body and then stiffens, the resulting shape can create geometric constraints that prevent the object from slipping out of the gripper [33] (see Figure 4B). As in the static friction mechanism, the use of white particles can improve the ability to wrap around the object by enhancing the interlock [34]. The addition of programmable protrusions or deformations in the membrane can facilitate interlocking, especially for smaller objects [35].
• Suction force: If the gripper membrane succeeds in hermetically sealing a small part of the object surface, by applying vacuum inside the membrane and reaching the granular interference phase, an additional suction force is produced that contributes to the overall gripping force [36] (see Figure 4C). This mechanism has been little studied, possibly because it requires the object surface to be smooth or wet to achieve effective sealing [33].

Figure 4. Gripping forces of a granular jamming gripper, the black contours correspond with the gripper and green contour correspond with target object. (A) Static friction force [37]; (B) geometric interlocking force [37]; (C) vacuum suction force from an airtight seal [37].

Figure 4. Gripping forces of a granular jamming gripper, the black contours correspond with the gripper and green contour correspond with target object. (A) Static friction force [37]; (B) geometric interlocking force [37]; (C) vacuum suction force from an airtight seal [37].

Figure 4 illustrates the three mechanisms that combined generate the gripping force of the granular jamming grippers; on the left, the green object represents the static friction force, the center mechanism refers to the geometric interlocking, and as for the suction mechanism given by the blue object, it is given by the air cavity between the top of the object and the gripper membrane.

2.3. Stiffness Control

As mentioned in the review article by J. Shintake [8], the grippers in granular jamming are classified within the soft robotics as “Gripping by Controlled Stiffness”, implying that the main advantage and mechanism of operation of these grippers is the stiffness control. This stiffness control is performed by applying vacuum inside the membrane, and in the literature, the tests usually use between −60 kPa to −90 kPa [18,25,31,34,37,38,39,40,41,42,43,44,45].

Response time or state transition time is an important factor for granular jamming gripper applications. Fortunately, these grippers have a fast response time for most applications. According to Howard [46], this response time is approximately 1 s. Another study by Joseph et al. [47] indicates that the transition time from minimum stiffness to maximum stiffness is between 2 and 3 s. In the review by J. Shintake [8], the response time is between 0.1 and 1.1 s. This time generally depends on the pressure difference, the flow rate provided by the pump, and the volume inside the membrane.

2.4. Influence of Grain Properties

Granular matter is one of the elements that most influences the versatility and gripping force of granular jamming grippers. Therefore, in [25,48], the authors measured how grain size, material, and morphology affect the interaction during granular interference, varying the gripping and activation forces. The materials used in the research can be classified into two categories: natural (coffee powder, sawdust, rice, salt) and artificial (glass spheres, 3D printing, rubber). The main difference is that natural grains present uncontrolled variations in shape and size, while artificial grains allow for precise control of the morphology and size of their components.

2.4.1. Grain Size

Santarossa et al. [33] showed that particle size affects the gripper suction mechanism, one of the three mechanisms that when combined exert the gripping force of the gripper.

Figure 5 shows that the particles found in the equator of the target body compress the object from the sides, resulting in the gripping force product of the static friction mechanism, in addition to small cavities between the target body and the membrane generating the suction force by hermetic sealing.

The authors of [33] conclude that for particle diameters of 120 µm, the membrane can form the hermetic seals between the gripper and the object, activating the suction mechanism, while for larger particles (4 mm diameter), the gripper membrane does not fully fit the object, preventing the creation of hermetic seals and in turn the suction mechanism. However, Howard et al. [49] performed a characterization of 35 types of granular matter, and within their findings, it is found that the largest grains (5–7 mm) are capable of generating the greatest gripping force due to the geometric interlocking mechanism.

Nishida et al. [50] suggested that small particles (0.1–1.5 mm) are desirable, in addition to a 50% membrane fill ratio. In another study, Mort [28] observed that the occurrence of jamming onset is affected by the grain size distribution.

De Rodrigo et al. [37] observed that, for a thin latex membrane, glass beads of approximately 1 mm in diameter obtained a better performance in gripping objects, suggesting that the size of the granular material may influence the external friction when the membrane is thin.

2.4.2. Materials of the Grains

In ref. [51], the authors conclude that one of the most important factors in maximizing gripping force is the water content of the grains, with a reduction of up to 60% in gripping force when the granular material is not completely dry. Götz et al. [31] demonstrated that the use of soft particles, such as expanded polystyrene beads, can lead to significantly higher gripping forces compared to rigid particles. This is due to a compression effect between the gripper and the object, which increases the static friction mechanism.

The stiffness of the granular material according to Howard et al. [49] is a key differentiating factor because the softer grains performed the best in gripping all target bodies in the study, and the author concludes that this is likely due to their ability to easily conform to each other and the test object [19].

A large-scale application developed by Miettinen et al. [52] using materials such as wood pellets, sawdust, plastic spheres and sand found that material compressibility is related to grip performance, with sand being the least compressible material, which performed the worst in the tests.

Authors Santarossa and Pöschel [34] investigated the influence of particle stiffness, finding that soft particles improve object envelopment, while stiff particles improve geometric interlocking. In addition, they introduced the mixture of hard and soft particles, which showed the best performance. Expanded polystyrene (EPS) were the soft particles that the authors chose to perform the experiments, as evidenced in

Table 1. Frequency, gripping force range, stiffness range, and response time of granular matters.

Type of Grain	Frequency	Gripping Force Range	Stiffness Range	Response Time	References
Coffee powder	19	5–100 N	up to ~100 kPa effective modulus	0.2–1 s	[24,35,39,48,49,50,53,54,55,56,57,58,59,60,61,62,63,64]
Glass beads	12	10–60 N	50–150 kPa (depending on packing)	0.5–1.2 s	[31,32,33,34,37,49,50,65,66,67,68,69]
3D Prints	3	>40 N	Tunable depending on geometry	0.3–0.8 s	[46,49,70]
EPS	3	5–20 N	low stiffness, <50 kPa	0.2–0.5 s	[31,34,71]
Sand	2	20–50 N	~100–200 kPa	0.5–1.5 s	[32,51]

, and expanded polystyrene is used in some studies on granular jamming grippers.

It is important to clarify that the data presented in Table 1 are average values and are subject to variations due to the configuration used in each gripper and the nonlinearity of the granular jamming gripper systems. As can be seen in Table 1, since the first studies concerning granular jamming grippers, the use of ground coffee has become popular. The main reasons why this material is the most widely used may be its easy access in many parts of the world, that the particle size is useful for a wide variety of objects with different shapes and sizes to be gripped, and that it is easy to filter to prevent granular matter from entering the pneumatic circuit.

Recent advances have also explored the use of 3D-printed particles to overcome the limitations of naturally available fillers such as coffee or glass beads. Matsumura et al. [72] demonstrated that additive manufacturing enables the design of particles with controlled shapes and surface morphologies, which directly affect packing density, interlocking, and the overall jamming transition. By tailoring these geometric parameters, the study showed that gripping performance can be optimized in a systematic and reproducible way, highlighting 3D printing as a promising tool for engineering particle–membrane interactions in granular jamming grippers.

2.4.3. Grain Shape

While natural grains have amorphous shapes and are not well defined, artificial grains can be controlled in both shape and size. The most common shapes are ellipsoids, spheres, cubes and superellipsiods. Using DEM (Discrete Element Method) simulations [45], it has been shown that these shapes maximize the diversity of behavior of the granular material in the compacted state [73,74]. The grains are parameterized as follows:

$${\left({\displaystyle \frac{x}{a}}\right)}^m+{\left({\displaystyle \frac{y}{b}}\right)}^m+{\left({\displaystyle \frac{z}{c}}\right)}^m=1$$

(1)

The coefficients for the sphere, ellipsoid, superellipsoids, and cube shapes are presented in

Table 2. Coefficients of the grain shapes [49].

Shape	M	A	B	C
Sphere	2	1	1	1
Ellipsoid	2	1	0.65	0.65
Superellipsoids	3	1	0.75	0.6
Cube	5	1	1	1

.

Some authors such as Fitzgerald et al. [75] establish aspect ratios $\alpha ={\displaystyle \raisebox{1ex}{$b$}\!\left/ \!\raisebox{-1ex}{$a$}\right.} , \beta ={\displaystyle \raisebox{1ex}{$c$}\!\left/ \!\raisebox{-1ex}{$a$}\right.}$ and shape factor m. To better understand the influence of each form factor, the change in the shape of the grains is visually presented in Figure 6, varying α, β and m.

The orange figures in Figure 6 represent the evolution of a sphere to a cube with rounded edges; only varying factor m, larger values of m generate increasingly cubic shapes. When the ratios α and β change in the same proportion, an elongated grain is obtained, while if only one of the two ratios varies, a geometry like a coin is obtained. It is important to clarify that in the case of purple figures, if the values of α and β intersect, the same geometry is obtained but represented in a perpendicular orientation with respect to the current plane.

In the context of granular jamming, the packing factor can be defined as the fraction of the total volume of the granular medium that is effectively occupied by the grains, relative to the void space between them. This parameter directly influences the stiffness modulation, contact distribution, holding force, and overall performance of jamming-based grippers, as it governs how tightly particles can rearrange and interlock under external loading or vacuum conditions.

Delaney and Cleary [73], through computational simulations and Discrete Element Method (DEM), studied how the shape of a particular grain affects its tendency to form an ordered or disordered packing, and the tendency of the packing factor of the grains against the variations in m and α. Their results showed that particle anisotropy and surface curvature are the main determinants of packing behavior. At high m values, grains with large flat regions promoted face-to-face contacts, which favored strong orientational alignment and the formation of dense, ordered structures. Conversely, increasing anisotropy by deviating α from unity reduced both ordering and density, as the symmetry necessary for crystalline-like arrangements was lost. Interestingly, at lower m values, where curvature dominates, moderate anisotropy could enhance the packing fraction, even though the global structures remained largely disordered. These findings highlight the dual role of anisotropy and curvature in governing whether granular assemblies evolve toward random packings or near-crystalline configurations.

3D-printed grains [40,46,49,70] with unconventional shapes show improved performance in some specific applications. For grain shapes such as cubes, the gripping force increases significantly for objects that can be geometrically jammed. However, for objects where geometric interlocking is not possible, the performance of cubes decreases. Spheres are the most consistent shape in tests, regardless of the target body. Combining these grains, depending on the specific case, can improve gripper performance. Nevertheless, the selection of the grain shape is highly application-dependent, which represents significant potential for optimizing gripper performance.

Wissman et al. [76] suggested that polyhedral grains, compared to perfectly spherical grains, exhibit the greatest change in gripper stiffness in the locked state. Particle shape influences their distribution, intergranular friction, and locking with the object.

3. Adaptive Finger-like Grippers with Granular Jamming

Another approach to using granular jamming is as a complement to a finger system, either to improve grip on the target body [77] or to achieve finger flexion by changing the stiffness of the grains in their internal structure [16,36,78,79,80,81,82].

Hou et al. [77] proposed a two-fingered gripper with variable-stiffness soft fingertips based on the granular jamming effect, where negative pressure is applied to modulate fingertip rigidity. Their design demonstrated that, compared with rigid-fingertip counterparts, jamming-enabled fingertips provide enhanced grasping stability due to the increased contact area and force distribution, while also reducing motor power consumption by up to 40% under equivalent conditions. Moreover, the prototype achieved improved load-bearing capacity and relatively higher placement precision (±0.62 mm) compared to most previous soft grippers, although some residual resistance and deformation were noted to affect repetitive accuracy. These findings highlight the potential of adaptive finger-like grippers with granular jamming to combine stability, efficiency, and versatility, making them suitable for both industrial manipulation and emerging applications in fields such as medical rehabilitation.

Li et al. [80] developed a soft pneumatic actuator with integrated granular jamming to modulate stiffness while maintaining compliance. The actuator demonstrated a bending range of 12.8–45.6° and showed improved resistance to buckling compared with hollow-chamber designs. Although the inclusion of granular media narrowed the working pressure range, it enhanced stiffness particularly along the bending direction highlighting both the potential of this approach for stability and the need for structural refinements to support multidirectional loading.

Liu et al. [79] reported a hollow variable-stiffness bellows structure based on granular jamming, manufactured with Eco-flex 00 30 silicone via wax-lost casting and filled with glass beads of varying diameters. Using a finite element model with the Ogden formulation, they showed that stiffness could be increased by up to 380% under negative pressure, while maintaining local surface softness. The study also analyzed how structural parameters, such as chamber angles, thickness, and particle size, affect stiffness variation, demonstrating the potential of bellows-type designs to achieve both global rigidity control and localized adaptability.

Jiang et al. [81] introduced a differential-drive particle jamming mechanism for soft actuators, featuring a dual-chamber structure in which one chamber is filled with particles. This design enables the independent control of bending angle and stiffness: simultaneous chamber inflation produces bending without stiffening, while differential pressure activates particle jamming to significantly increase rigidity. Both theoretical modeling and experimental validation confirmed the accuracy of the proposed approach, demonstrating an extended range of stiffness modulation and bending control compared with conventional vacuum-based jamming methods.

Chung and Chow [83] developed a honeycomb-based jamming gripper as an alternative to particle filled systems, aiming for lightweight and slender fingers with tunable stiffness. The honeycomb sandwich structure allowed for directional deformation and stiffness control through air pressure modulation, enabling transitions between flexible and rigid states. Bending load tests revealed that Nomex honeycomb provided gradual stiffness changes without permanent deformation, while aluminum honeycomb displayed high stiffness but was prone to plastic yielding under higher loads. Reinforcements such as Velcro increased stiffness in the Nomex configuration, making it suitable for grasping harder objects. Pressure-dependent tests indicated that a vacuum of around 20 kPa was sufficient to reach stable force displacement behavior while avoiding structural damage. The prototype successfully grasped a wide range of objects, including fragile items like eggs and fruits, irregular shapes such as garlic bulbs, and heavier tools, demonstrating both compliance and load capacity. These results highlight the adaptability of honeycomb jamming structures and their potential for tasks requiring direction-dependent stiffness, such as bolt alignment and assembly.

Jiang et al. [84] further proposed a chain-like structure (CLS) with granular jamming capable of achieving up to 50.7 fold stiffness variation without the need for vacuum systems. The design, consisting of rectangular granules forming a chain-like arrangement, was integrated into hybrid actuators and tested in an anthropomorphic hand. Experiments demonstrated the ability to handle a wide range of objects from delicate items to loads exceeding 3 kg by switching between soft and rigid states. Although challenges remain in automating stiffness regulation and optimizing hybrid structures, this approach highlights a promising direction toward versatile, high-performance soft–rigid robotic systems. Collectively, studies on adaptive finger-like grippers with granular jamming demonstrate the potential of these mechanisms to combine compliance with enhanced load capacity and precision, paving the way for applications in industrial manipulation, service robotics, and human–robot interaction.

4. Influence of the Membrane

The membrane is responsible for containing granular matter, establishing the interface between the object and the grains, so the performance of the gripper largely depends on it, as the literature indicates the membrane must be flexible, adapt to the objects, and withstand cycles of use without failing due to material fatigue [33,34,53,74].

4.1. Membrane Material and Properties

In studies that have made their own membrane, usually made from silicone and a mold, it has been shown that for finger-shaped actuators, a greater thickness of the membrane generally decreases the flexural stiffness [10]. The elasticity of the membrane can have an adverse effect on the ability of the gripper to adapt to objects [65,85]; if it is not adequate, the granular material will not adhere to the object, reducing the gripping force.

The durability of the membrane is crucial for the long-term operation of the gripper, as it must withstand sufficient phase transition cycles for an optimal lifetime. Therefore, the development by Amend et al. [25], which focused on manufacturing the membrane by dip molding [86], conflicts with two characteristics of the gripper durability vs. gripping force, since to increase its durability, it is necessary to periodically increase the stiffness, which generates a decrease in the gripping force due to less adaptation to the object.

The ideal membrane must be extremely resistant to wear, abrasion, and perforation; very thin and flexible to adapt as much as possible to the target body; and resistant to chemical degradation. Figure 7 exemplifies the importance of a thin and flexible membrane, since this directly affects the conformability of the membrane to the target body and the generation of gripping force through the granular jamming gripper mechanisms. The most effective membrane found by Amend et al. [25] is a polychloroprene-based formulation that has an average of 50,000 grips until failure, although this number depends largely on the target bodies. The failures that occur in the membranes are stretching (plastic deformation), abrasion, and perforation. For round and smooth target objects, this membrane is estimated to reach up to 90,000 grips, while for sharp or abrasive objects, an estimated 5000 cycles of use are expected.

As shown in

Table 3. Frequency of membrane material.

Type of Grain	Frequency	References
Latex	24	[24,31,32,35,37,46,48,49,50,51,53,55,56,58,60,61,62,63,64,65,66,69,87]
Silicone	6	[25,37,39,43,57,71]
3D Printing	2	[54,70]
Fiber	1	[52]

, latex membranes have been used most frequently in research. This may be due to the ease of purchasing party balloons, their wall thickness of approximately 1 mm, and their resistance to multiple cycles of use, making this option the most convenient for research purposes.

4.2. Membrane Morphology

The shape and structure of the membrane, also known as “morphology”, are key factors in gripper performance and design. Howard et al. [54] used evolutionary algorithms to optimize membrane morphology by modeling the membrane profile (Figure 8a), showing that the common bag-like shape is far from optimal to take full advantage of the three force mechanisms of the gripper; the genetic algorithm evolves to maximize the force mechanisms on a selected sphere as a target body, and a potential problem is the over-specialization of the membrane profile to the target body, which is why they tested different target bodies to evaluate the transferability of the designs and the development of the fundamental force mechanisms to generate grip force.

Thanks to 3D printing, the optimized models generated by genetic algorithms can be implemented in a practical and economical way, something that would be more expensive and time-consuming with the method of generating membranes through molds. Figure 8b exemplifies the large number of tests the authors performed [54].

Kapadia and Yim [35] introduced ‘nubs’, which correspond to protuberances that are programmable membrane deformations. However, these “nubs” only improve the gripper’s performance for target bodies that fit the membrane, being a disadvantage to the gripper’s adaptability. Howard et al. [70] broadens the spectrum of programmed deformations, using 3D printing experiments carried out where the thickness and pattern of the membrane were mainly varied. The presented design is a twisted membrane approach which introduces a new alternative in programmable membrane deformations for granular jamming grippers.

5. Reset Methods

One of the characteristics of granular jamming grippers is their ability to revert between the stuck and fluid states; this action is known as resetting. To ensure optimal gripper performance, it must be reset to the fluid state before starting a new gripping cycle; otherwise, its ability to grasp objects deteriorates rapidly.

In this process, air must be reintroduced into the membrane to equalize the internal pressure with atmospheric pressure, reducing the forces that maintain the rigidity of the gripper, reducing intergranular friction (friction between grains), recovering the relative mobility of the grains and reducing geometric jamming because the contact network relaxes.

5.1. Manual Reset

The manual reset method for granular jamming grippers involves restoring the granular material to its original state after a gripping operation. Once the internal pressure of the membrane equals atmospheric pressure, the granular material within the membrane can freely rearrange itself. At this stage, manual techniques such as shaking, pressing, or molding the gripper on a flat or reference surface are used. This process eliminates any residual interlocked structures or compacted areas that could compromise formability or repeatability of the next gripping cycle. Although it is a simple and effective method, its manual nature limits the automation of the system and can introduce variability in performance.

5.2. Positive Pressure

An efficient alternative to manual resetting of granular interference grippers is the use of positive pressure. Instead of relying on external physical actions to redistribute the particles, this approach uses short bursts of pressurized air injection into the membrane to rapidly decompress the granular medium. As observed by Amend et al. [53], this procedure not only accelerates the reset time but also provides additional advantages in terms of repeatability and system control. The burst of air generates a momentary expansion inside the gripper, which disrupts the interlocking of the particles and facilitates their reorganization. This automated mechanism has been adopted by most of the experiments related to the investigation of granular jamming grippers. Wang et al. [88] proposed that the gripper molding phase be performed with an additional filling of air (130% additional) and subsequently obtained that with this configuration the gripping force was increased and the activation force was decreased, which generates an increase in the “Ratio of Pull-off Force to Applied Force”.

5.3. Active Fluidization

In the pharmaceutical, agricultural, and food industries it is common to find the transport of granular matter; to allow a smooth flow, hopper vibratory systems are commonly used to unclog the granular matter [89,90]. Vibrated granular materials show different internal energies depending on the input frequencies, and even if the vibrational energy is kept constant the resonant modes of the granular material will have a different energy [91]. Lemrich et al. [92], by discrete element simulation (DEM) applying vibration, showed the presence of linear vibrational modes at low amplitudes with a softening behavior of Young’s modulus, which was attributed to the re-composition of the internal forces of the contact chain during vibration.

Building on previous studies on granular matter vibration, Mishra et al. [55] adapted a small vibrating element with a 3D-printed holder in direct contact with the membrane. This vibration induced in the granular material can improve the gripping force, reduce the activation force and the force exerted by the gripper on the target body. Furthermore, the authors show how different waveforms generate different clamping forces, paving the way for optimizing the gripper for a specific application.

Continuing with the fluidization system, Coombe et al. [56] developed an active fluidization system that details the procedure for gripping objects, characterizing their gripping forces, and turning audio signals on and off. A sweep was performed for time-varying and time-invariant signals, amplitude (volume) changes, and a frequency sweep. Among the results found is that, for time-invariant waves, fluidization results in a better contact angle, greater object envelopment, a greater effect of the static friction force and, therefore, better gripper performance. For time-varying waves, the gradual decrease from high to low vibration allows the granular material to settle into a more compact configuration; this effect arises with frequency and volume ramps.

6. Performance

The purpose of grippers is to securely grip and manipulate the various items required. For safe operation, the gripper must guarantee adequate holding force, adaptability to grip the widest range of objects, and optimal response time. The grip is tested to quantify the gripper’s ability to grasp objects of different shapes/sizes and the force exerted on an object during the grip (if the object is fragile, this force must be controlled); these quantitative measures are useful as design parameters, and depending on the application the gripper’s characteristics can be improved.

6.1. Success Rate

The success rate is a key performance indicator of a granular interference gripper, as it assesses its ability to successfully complete the intended gripping tasks. This parameter is defined as the percentage of attempts in which the gripper successfully grasps, manipulates, and releases the object in a controlled manner, without unwanted slippage or drops. To determine this, repeated tests are performed with objects of different geometries, dimensions, weights, and materials, recording the number of successful operations compared to the total number of attempts. In the study of D’Avella et al. [61], the results of these models can be comparing thanks to the success rate.

6.2. Response Time

Response time refers to the interval between the system’s activation signal and achieving a stable grip. In granular jamming grippers, this time is influenced by factors such as the evacuation air rate [28] (in pneumatic systems), the deformability of the membrane, the internal particle distribution, the internal volume of gripper, and the control system used. A reduced response time increases operating speed and system performance, which is advantageous in high-speed applications. However, it is important to optimize this parameter without compromising grip quality, as excessively rapid actuation could lead to uneven internal particle distribution, negatively affecting contact strength and stability. As mentioned in Section 2.3 stiffness control, the approximate response time of granular interference gripper applications ranges from 0.1 s to 3 s [47]. Although not mentioned in the study of Miettinen et al. [52] due to the large size of the gripper, the response time had to be longer.

6.3. Activation Force

This parameter is the force required by a gripper to adapt its initial shape to the target object. This value is obtained during the compression phase, when the gripper is unstuck. This value is relevant for applications where the object to be gripped is fragile or highly deformable [37]. Fujita et al. [57] proposes a variable internal volume mechanism where the pushing force exerted by the gripper on the target body is significantly changed.

Underwater applications such as those of Licht et al. [66,67,69] seek to minimize the force that the gripper exerts on the object, as this can cause deformations or the object to sink deep in the water. The main strategy they used was to not completely fill the gripper, and to vary its filling percentage.

6.4. Holding Force

This refers to the maximum force that the gripper can exert on the object without it slipping or falling. In the specific case of granular jamming grippers, Brown et al. [24] obtained the gripping force by pulling the target body out of the gripper and recording the pulling force as a function of the vertical extension. This method for obtaining grip strength is a version of the tensile test using a universal testing machine.

Because there are three mechanisms by which the gripper exerts force on the object, (detailed in Section 2.2 Grip forces) several studies use hemispheres to characterize mainly the static friction force, which depends on the size and shape of the target body, the contact angle, the surface conditions, and the applied force or activation force at the time of gripping [24,53].

The data presented in

Table 4. Maximum gripping forces of different granular jamming grippers configurations.

N	Article	Target Bodie	Granular Matter	Holding Force [N]	Activation Force [N]	Membrane Diameter [mm]
1	[35]	Cylinder (6 mm)	Coffee powder	31.88	15	65
		Sphere (6 mm)		35.85	15
		Parallelepiped (6 mm)		29.14	10
2	[58]	Sphere (8 mm)	Coffee powder	0.1	-	100
		Cube (10 mm)		1.4	-
		Square pipe (10 cm)		0.3	-
3	[59]	Cylinder	Coffee powder	104.17	20	50
4	[49]	Sphere (20 mm3)	3D printing materials, ground coffee, glass spheres	10	-	125
		Cube (20 mm3)		13.5	-
		Coin (20 mm3)		1.7	-
		Star (20 mm3)		14.3	-
5	[39]	Hemispheres (12 mm)	Coffee powder	6.06	1.2	36
		Hemispheres (27 mm)		11.8	1.6
		Hemispheres (32 mm)		11.06	2.35
6	[56]	Sphere (30 mm)	Coffee powder	16	250	45
7	[50]	Ballpoint pen	Ground coffee	27	2	523 cm3
			Aromatic beads	20	2
			Adzuki beans	13	2
8	[33]	Sphere (20 mm)	Glass beads (4 mm)	4.38	37	73
			Glass beads (120 µm)	6.73	37
9	[32]	Glass sphere coated in rubber (17 mm)	Polymer	25.3	26.8	60 cm3
			Sand	19.8	24.1
			Ceramic beads	25.3	24
			Amaranto	27.4	26
			Glass beads (2 mm)	12.4	24.1
			Glass beads (200 µm)	21.27	18.7
10	[34]	3D-printed cylinder	Glass beads	5.3	-	70
			Expanded polystyrene	18.2	-
			Combination of 10% glass beads 90% expanded polystyrene	31.7	-
11	[51]	-	Coffee powder	77.6	-	43
			Sawdust	66.1	-
12	[54]	Star (20 mm3)	Coffee powder	6.6	-	25–40
		Coin (20 mm3)		16	-
		Cube (20 mm3)		22.7	-
		Sphere (20 mm3)		24.7	-
13	[60]	Surgical instruments	Coffee powder	13.3	-	40
14	[65]	Sphere (21.5 mm)	-	11.5	9.1	43
15	[70]	Cube (20 mm3)	3D-printed grains	7.862	-	-
		Sphere (20 mm3)		7.369	-
		Coin (20 mm3)		2.891	-
		Star (20 mm3)		9.563	-
16	[46]	Cube (20 mm3)	3D-printed grains	6.9	-	125
		Sphere (20 mm3)		4.74	-
		Coin (20 mm3)		2.59	-
		Star (20 mm3)		16.03	-
17	[31]	Smooth steel sphere (20 mm)	Glass beads (4 mm)	0.78	-	75
			Expanded polystyrene pearls (4.2 mm)	14.83	-
18	[71]	3D-printed cylinders	Expanded polystyrene pearls	15.78	3.2	80
19	[93]	3D-printed cylinders	Ground coffee	38.6	31.2	46

were obtained by analyzing the graphs in each of the cited articles. The maximum gripping forces reported are drawn from studies employing diverse experimental setups, materials, and measurement protocols. These include variations in gripper size, membrane thickness, particle type and filling ratio, actuation method, and even environmental conditions (e.g., underwater vs. atmospheric pressure). Due to the lack of standardized testing procedures in the field, direct numerical comparison of the values across studies should be approached with caution. Rather than providing absolute benchmarks, the data illustrate relative trends in performance and highlight how design choices and operating conditions influence the capabilities of granular jamming grippers.

Activation force is a parameter to consider when the objects to be manipulated are soft or delicate, as excessive force can damage or break the object. The maximum gripping force parameter is useful to obtain an idea of the number of objects the gripper is capable of securely holding. Maximizing this value without creating an oversized gripper is a valid design parameter for the implementation of granular jamming grippers, and Licht et al. [66], with their partially filled gripper, managed to reduce the activation force needed for a secure grip.

The exceptionally high gripping force (≈100 N) reported by Fujita et al. [59] can be attributed to a combination of design and experimental conditions. First, the use of a three-layer membrane structure with inner and outer rubber sheets enclosing a layer of coffee grounds increased the effective packing density while preventing buckling and enhancing rigidity in the jammed state. Second, the application of a high vacuum level (−89.9 kPa) between the membranes promoted a robust jamming transition, thereby maximizing resistance to object slippage. Third, the selection of granular mass and filling amount was critical: intermediate loads provided the best trade-off between low insertion force and high holding capability, ensuring both adaptability and stiffness. Finally, the testing conditions, which involved a smooth-surfaced MC nylon cylinder with a diameter set to 80% of the gripper’s internal diameter, ensured a large contact area and a stable interlocking state. Collectively, these factors enabled the gripper to achieve one of the highest gripping forces reported in granular jamming literature.

7. Applications

While most studies have been conducted in controlled environments and focus on expanding existing knowledge about granular jamming grippers and the effect of variables such as grain material and morphology, membrane material and morphology, and auxiliary systems such as positive pressure or fluidization, authors have also taken granular interference grippers to practical applications and use in uncontrolled scenarios. This section describes the applications and different integrations that granular jamming grippers have.

Jacob and Secco [87] presented a low-cost granular interference gripper; using granular matter of ground coffee and flour, a plastic funnel, and a vacuum pump, they managed to achieve a maximum payload of 200 g and a high success rate on conventional office objects, carrying out pick and place tasks.

7.1. Computational Applications and Modeling

Jiang et al. [94] presented a learning algorithm that improves the gripper’s success rate regardless of the gripper configuration. Using this model, it is not necessary to have a mathematical model of the gripper to use it with objects other than those used in the algorithm training. Continuing with algorithms, Fajardo et al. [63] using an artificial neural network that calculates the optimal negative pressure difference (vacuum) based on its weight and pixel size, and this optimization is useful to minimize the energy consumption of the vacuum pump. D’ Avella et al. [61] developed a perception algorithm that identifies the object to be grabbed using a Kinect system and suggests the optimal grip point for the object in front of the camera.

A fundamental pillar in the field of granular interference gripper simulation is the Discrete Element Method (DEM) [31,75], which allows for modeling interactions at the particle level, crucial for understanding the emergent behavior of granular systems. Complementing the DEM, the use of evolutionary algorithms (EAs), such as NSGA-III or NSDE [95], marks a milestone, enabling the exploration and multi-objective optimization of complex design spaces for granular materials. Free software such as MercuryDPM and LIGGGHTS are libraries that allow the numerical simulation of particle-to-particle interactions and 3D visualization of granular matter interactions. Dierks et al. [45] simulated grain interaction as performed using the software LIGGGHTS, and a pseudo membrane of bound particles was used to model the membrane deformation in the molding process.

Fitzgerald et al. [75] combined NSGA-III evolutionary algorithms with DEM to perform multi-objective grain morphology optimization for use in locking grippers for a variety of target object sizes. By scanning different grain morphologies, sizes, and target bodies, they found a table of optimal grain shapes for different target objects and the importance of exploring the macroscopic properties of the interaction of the grains with the target body is emphasized.

The computational approaches described here are generally validated under simplified or idealized conditions. For instance, assumptions such as homogeneous grain distribution, symmetric membrane deformation, and quasi-static loading do not fully reflect real operational scenarios. In practical use, granular media exhibit heterogeneous packing, localized force chains, and dynamic rearrangements that can strongly affect gripping reliability

7.2. Underwater Environments

Granular jamming occurs when a negative pressure difference is generated between the environment and the interior of the membrane containing the granular material. Typically, air is the fluid used for this application. However, water can also be used as a fluid medium to control granular interference [67,69,96]. This characteristic makes granular jamming grippers useful for applications involving the collection of a wide variety of objects underwater. This is why S. Licht et al. [69] developed an extension of the gripper using liquid water as the internal fluid within the submerged membrane and pressure-tolerant glass beads. The authors show an example of how to grip different objects such as glasses or a hairbrush at 200 m underwater.

The motivation behind using granular jamming grippers for underwater applications lies in marine archeology and the manipulation of objects that are weakly suspended in marine sediments. Using concepts such as those of Amend et al. [53] and Wang et al. [88] about the positive pressure reducing the activation forces required for the membrane to adopt the shape of the target body, and by playing with the membrane filling proportions, the required activation force can be varied and, therefore, the force that the gripper exerts on the target body can be reduced.

Following underwater investigations, S. Licht et al. in [66] showed that the use of partially filled granular grippers is the key to successfully grasping objects resting freely on soft, pliable sediments, and they also show that the main determinant of the relationship between upward and downward forces on the target body is the combined volume of fluid and particles during the approach.

However, it is important to highlight that experimental validation was primarily conducted in controlled laboratory environments with simplified objectives, and that the system relied on manual fluid evacuation, limiting its immediate applicability for autonomous subsea operations. Furthermore, while partially filled grippers proved advantages for wedging objects in soft sediments, conventional fully filled grippers still failed in certain geometries, highlighting the challenge of achieving truly universal gripping in real-life marine environments.

7.3. Tactile Perception

For robots to interact safely and accurately with the real world, it is necessary to use sensors that can provide feedback in some way on whether the robot is successfully performing the programmed operation, or if, on the contrary, corrections must be made to accomplish the task. This is why detection makes sense. In the case of granular jamming grippers, while its simplicity in terms of components and operation facilitates operation and it is a gripper known for its great versatility, this simplicity also presents a challenge for incorporating tactile sensors on its surface to detect objects grasped by the gripper. In [97], J. Hughes and F. Lida present a granular interference gripper with a polymer-based strain sensor, which can detect surface deformation. However, the arrangement of these sensors located on the membrane degrades the gripper’s performance.

In [98], the authors use transparent particles with filling liquid as granular material, and incorporating a camera inside the membrane allows for the optical detection of the membrane deformation; although high-resolution images were not obtained, Li et al. [67] later presented their detection, called “TaTa” (Touch and take), which incorporates improvements in the grains used, the membrane, and the LED lighting, among other improvements that obtain images of detection of the internal deformation of the membrane. This technology allows for the detection of the target body. In this research, the images obtained are subsequently processed to determine whether the grip was carried out or not.

7.4. Medical Applications

Recently, the authors Halouani et al. [62] presented a granular jamming gripper focused on ankle rehabilitation that is fixed to the plantar surface of the foot and molded to its geometry; when the foot needs to be released, air is injected back into the membrane. This is important due to the little involvement that assistive robotics has within ankle rehabilitation, while rehabilitation of other parts of the body already involves specialized assistive robotics devices. Figure 9 shows the configuration used for ankle rehabilitation, where the patient’s foot is held in place by taking advantage of granular jamming.

Another application that has been given to granular interference grippers is the ability to grip a set of forceps for wisdom teeth extraction. Badilla-Solórzano et al. [60] present a granular jamming gripper configuration that increases the gripping force by geometric interlocking, and therefore the performance of the gripper. Figure 10a shows that the “Hybgrip” gripper presses the granular jamming gripper to lift the object and enhance the effect of the geometric interlock on the gripping force. Additionally, the authors explore the use of a mesh beneath the target body in order to allow the hybgrip mechanism to achieve object lift and better geometric interlocking; the results of standard granular jamming gripper, hybgrip, and hybgrip whit grid are shown in Figure 10b.

Medical applications of granular jamming grippers still face several challenges and limitations. The reliance on external vacuum pumps results in high power consumption, noise, and limited portability, which restricts their integration into compact rehabilitation systems. The membrane, while functional, is prone to wear and tear, requiring frequent replacement to ensure consistent performance. Furthermore, the choice of granular material is not universally optimal; for example, sugar demonstrated high grip strength but was sensitive to temperature and humidity, compromising reliability. From a usability perspective, difficulties inserting the foot into the membrane remain critical, especially for patients with reduced mobility, and the pressure exerted on the limb has not yet been systematically measured, limiting safety assessment.

7.5. Collaborative Robotics

Collaborative robotics develops robots that can interact with humans safely and without physical barriers. D’Avella et al. [61] incorporates a granular interference gripper into a Baxter robot as a collaborative platform. Additionally, the study focuses on an object perception algorithm in a cluttered environment. The system provides the optimal grip point, and the Baxter platform generates movement. The advantage of the gripper in these scenarios is its high adaptability and the fact that the object does not need to be at the center of the gripper. Harada et al. [64] uses a granular jamming gripper coupled to a HIRO dual-arm 6-degree-of-freedom robotic manipulator, which has a part centering mechanism. The gripper was primarily used for 2-part assembly, where the first part is centered by the mechanism, and then gripped and directed to a second part and assembled by pressure; this assembly module allows the pieces to be oriented and then efficiently gripped through granular jamming and then assembled.

Another application explored for granular interference grippers is aerial manipulation, as demonstrated by Kremer et al. [71,99]. The authors present “TRIGGER”, a lightweight universal granular jamming gripper designed to be mounted on a multicopter, enabling omnidirectional grasping and serving as landing gear. The system achieves up to 15 N of holding force on objects without geometric interlocking, relying on friction and suction, and operates with a low activation force suitable for small to medium UAVs. Figure 11 shows the gripper performing a pick-and-place task in a laboratory environment, successfully grasping various synthetic shapes under aerial conditions. This capability expands the use of granular jamming technology to scenarios where lightweight, compliant, and adaptable grasping is required in unstructured environments.

In aerial manipulation, reliable grasping is highly sensitive to object geometry, with spherical or pointed shapes often leading to slippage or failed engagements. The dependence on human operators for gripper activation and the need for precise positioning further limit autonomy, while the small membrane size constrains tolerance to positional errors. Moreover, the use of motion-capture systems in controlled laboratory settings raises concerns about scalability to unstructured outdoor environments, where disturbances such as wind, sensor inaccuracies, and uneven surfaces introduce additional challenges.

7.6. Semi-Active Drive Method Using Granular Jamming

Beyond standalone granular jamming grippers and finger-like systems, recent studies have explored hybrid designs that integrate granular jamming with other actuation methods to overcome the intrinsic limitations of soft robotics. Combining jamming with pneumatic or tendon-driven actuation provides a strategy to simultaneously achieve compliance, tunable stiffness, and reliable grasping performance. For instance, Li et al. [80] demonstrated that embedding granular media within a soft pneumatic actuator significantly improved resistance to buckling and enhanced directional stiffness while preserving bending compliance. Similarly, Jiang et al. [81] introduced a differential-drive particle jamming mechanism in a dual-chamber actuator, enabling independent modulation of bending and stiffness, an approach validated experimentally to extend the functional workspace compared with conventional vacuum-driven systems. These hybrid strategies highlight how stiffness gradients induced by jamming can effectively compensate for the low structural rigidity of soft manipulators, addressing one of their long-standing challenges. More broadly, the integration of jamming into soft actuation schemes not only improves grasp stability and load capacity but also points toward a new class of adaptive, multifunctional manipulators. Such designs illustrate the dual role of granular jamming as both an application and a complementary technology, underscoring its potential as a research direction in the development of versatile robotic grippers and manipulators.

Recently, Bartkowski et al. [100] investigated the integration of granular jamming into pneumatically driven finger-like actuators (see Figure 12), incorporating the jamming mechanism on the contact surface to enhance load-bearing capacity. The study systematically analyzed the response of such structures under cyclic loading, with empirical tests in compression, tension, and bending. The results demonstrated a strengthening effect with increasing cycles (up to 2000), primarily caused by the sticking of particles that increased resistance to displacement and led to macroscopic hardening. To capture this behavior, a constitutive equation and modeling methodology were proposed, achieving strong agreement with experimental results, although with limitations such as the inability to describe unloading. The concept was validated through the development of a soft robotic gripper, where the hybrid actuation showed improved grasp stability and long-term functionality, emphasizing the potential of combining pneumatic actuation with granular jamming to achieve adaptability and durability in soft robotic manipulators.

Cheng et al. [82] proposed a semi-active drive method for hyper-redundant manipulators that integrates local tunable stiffness via granular jamming with cable-driven actuation. Instead of relying on distributed actuators at each joint, the manipulator achieves articulation through tension cables actuated by off-board spooler motors, while granular media enclosed in flexible membranes provide reversible stiffening under vacuum. Experimental evaluation of a prototype highlighted rapid jamming and unjamming transitions (Between 0.1 s and 0.2 s), as well as a substantial increase in payload capacity when tension cables were coupled with jamming, enabling the system to support more than 200% of its own weight. Moreover, the manipulator demonstrated high dexterity and the ability to conform passively to obstacles, achieving a nearly spherical workspace and maintaining articulated configurations under jammed conditions. This semi-active approach underscores the potential of granular jamming to enhance strength, adaptability, and safety in robotic manipulators, particularly in applications requiring lightweight and highly flexible architectures.

The addition of granular chambers increases structural complexity, fabrication time, and system weight compared to purely pneumatic soft grippers. Furthermore, the long-term stability of particle–particle adhesion mechanisms under extended cyclic loading remains poorly understood, with potential risks of fatigue or clogging effects that could compromise repeatability.

8. Discussion and Future Directions

This research addressed the physical principles of granular jamming grippers, the studies supporting these theories, the influence of grains, the influence of membranes, reset systems such as positive pressure or fluidization, and applications in various fields. Considering the studies published to date, five potential research topics in these devices are presented below.

8.1. Grain Databases and Exploration of 3D-Printed Grains

Progress in gripper design using granular jamming requires a more exhaustive characterization of the granular materials used. Currently, information on the mechanical, tribological, and deformability properties of grains is limited to a small set of materials, which limits the ability to predict and optimize gripper behavior. Systematic data sets, such as those presented by Howard et al. in [49], need to be established, which include variables such as shape, size, material, and relative hardness. In particular, the use of grains manufactured by 3D printing opens a new field of possibilities, allowing precise control of the internal and external geometry of the granular elements, as well as the incorporation of internal structures designed to modify their mechanical response in a programmable way.

8.2. Computational Modeling

Computational gripper modeling primarily uses the discrete element method (DEM) to simulate intergranular interactions. This method is useful for understanding contact dynamics and frictional forces, but not for simulating the effects of membrane boundaries with grains. This grain–membrane interaction is highly nonlinear and difficult to model with approximations or simplifications. In these cases, it is appropriate to develop membrane simulations using finite element analysis (FEA) and evaluate possible integration with the discrete element method (DEM) for the interior of the granular material. Such models would allow for exploring complex configurations, anticipating gripper performance, and optimizing, for example, membranes with programable deformations and 3D-printed grains with a specific morphology.

Beyond physical modeling, future research could leverage generic optimization algorithms, such as particle swarm optimization or Bayesian optimization, to systematically explore the design space of grain morphologies, membrane geometries, and material properties. These algorithms can be integrated with DEM-FEA simulations or surrogate models to rapidly evaluate thousands of configurations, enabling the data-driven identification of optimal designs for specific gripping tasks. Combining these optimization frameworks with advanced manufacturing methods, such as additive manufacturing for customized grain shapes or membranes with programmable deformations, could significantly accelerate the development of high-performance granular jamming grippers.

8.3. Exploration in the Mixture of Grains

Santarossa and Pöschel [34], utilizing a heterogeneous mixture of hard and soft grains, determined that the gripping force depends largely on the composition of the granular matter; this strategy is not explored in the literature and could provide synergistic properties that improve the adaptability of the gripper and its ability to generate geometric interlacing or suction force by achieving a hermetic seal with the surface of the object. Another option is the combination of contrasting geometries (sphere, cubes, superellipsoids, etc.) or different grain sizes. Future studies could investigate the effect of particle density gradients, surface textures, or even coated grains with adhesive or low-friction layers to further tailor the internal rearrangement during jamming. Such combinations could enable granular jamming grippers to achieve optimized performance for specific applications, such as handling highly irregular, porous, or fragile objects.

8.4. Control Systems

Currently, two feedback systems have been developed for granular jamming grippers, strain sensors [50] and tactile feedback [68]; these developments are especially useful for moving towards incorporating real-time control systems. This control could dynamically adjust negative pressure, make corrections to trajectories, or trigger automated reset cycles, especially useful for handling fragile or unstable objects. Kremer et al. [71] suggests that the gripper can be described as two parallel springs (See Figure 13). The first spring $k_{air}$ corresponds to the air-filled membrane, which is a variable spring. The other spring $k_{lmp}$ corresponds to the other variables involved in the gripper’s configuration.

Building on this integration, future research could explore closed-loop controllers that use tactile or force feedback to adapt vacuum levels, membrane deformation, and approach trajectories in real time. This would enable granular jamming grippers to autonomously adapt to diverse object geometries and mechanical properties, even in dynamic or uncertain environments. For instance, the model could serve as the predictive core of a model-based controller, where sensor feedback corrects deviations between expected and measured activation force or holding force, triggering adjustments in β or automated reset cycles when loss of grip is detected. Such advances would not only improve reliability for delicate or unstable objects but also expand the operational domain of granular jamming grippers into aerial, underwater, and collaborative robotics applications.

Table 5. Resume of control systems in granular jamming grippers.

N	Article	Algorithm Type	Input Signals	Output Actuators	Parameter Dimensions	Reported Accuracy or Metric	Limitations
1	[53]	Open-loop Pressure Control/State Modulation	Pressure commands (vacuum/positive pressure)	Vacuum pump, positive pressure port, solenoid valves	Reliability, fault tolerance, positioning accuracy, launch capability	Reliability increases up to 85%, error tolerance increases up to 25%, positioning accuracy ±60 mm (95% conf), gripping rate 16.2 picks/min.	Precision too coarse for high-precision manufacturing tasks. Difficult to compare with other grippers due to lack of standard benchmarks.
2	[50]	Force feedback (threshold-based)	Force measured by strain gauges	Electromagnetic valve (to activate vacuum pump)	Optimal filling volume, adequate pressing force	The effectiveness of the force feedback system was evaluated.	N/A.
3	[63]	Artificial Neuronal Network (ANN)	Pixel area of one side of the object, pixel area of the other side, weight of the object	Negative pressure control system (vacuum pump)	Pixel size, weight, optimal negative pressure	A 99.131% accuracy in determining optimal negative pressure.	The gripper and vacuum cleaner are not industry standard. Future work includes using commercial grippers and a higher-voltage motor.
4	[61]	Heuristic perception algorithm (based on depth features)	Single camera depth image	Baxter Platform, Granular jamming gripper with coffee powder	Point of grasp (x,y,z, orientation)	Approximate success rate of 75%. Competitive with DNN solutions in computation time and capture success.	Difficulty with porous and bigger objects than gripper size. It is not robust to external forces/torques for some objects. The algorithm is specific to the proposed gripper.
5	[35]	Fluidization control strategy	Pumping frequency, duty cycle	Bidirectional pump	Pumping frequency, duty cycle, average positive pressure	Significantly higher clamping forces (typically 60%), wider range of object geometries.	Optimization of protrusion geometry is an area of further study.
6	[99]	Model Predictive Control (MPC) with force control	UAV status (position, speed), grip strength, mission requirements	UAV movement commands, gripper force commands	UAV trajectory, grip force, safety restrictions	Automated gripping with greater robustness and versatility, greater operational reliability	Simulation study, not real-time embedded performance. The previous open-loop setup required a human operator.
7	[39]	Mode switching logic (GJ, EA, combination)	Object properties (shape, surface, stiffness)	Vacuum pump (for GJ), high voltage source (for EA)	Grip strength, object sizes, object types	GJ lifts 38 times its weight. EA handles flat/fragile objects. Combined mode: 35% more grip strength.	GJ struggles with flat/fragile/delicate objects or objects larger than the bag. EA struggles with oily/wet surfaces.
8	[56]	Active vibration fluidization	Frequency and amplitude of wave	Computer-controlled audio exciter	Frequency, amplitude, temporal properties of the waveform	Improves grip strength. Grip strength favors low frequencies and high volumes thanks to the reorganization of grains and increased contact area.	Little work has been performed exploring other effects of granular physics. There is a need for a better understanding of the effects of time-varying vibration.
9	[98]	Pose estimation (point-to-feature ICP)	Visual detection of the jig membrane with integrated cameras	Hydraulic drive system (for oil quantity)	Object pose (orientation, position, angle)	RMS error <4° for orientation. Repeated object fixation <0.5 mm (position), <1.1° (angle).	It is a jig, not a gripper. The ICP can be computationally intensive.
10	[101]	Deep Learning Convolutional Neural Network (CNN) (strength prediction)	Images (video simulation) of gripper deformations	Strength prediction/stress maps	Contact forces, stress maps	Rigorous evaluation of prediction performance under variations in contact point, object material/shape, viewing angle, and occlusion.	FEA is computationally intensive for real-time feedback. Sensing in soft robotics is challenging. It is prediction, not direct control.
11	[75]	Multi-objective Evolutionary Algorithm (NSGA-III) for design optimization	Grain morphology parameters, target object sizes/shapes	Grain shapes optimized for 3D printing	Grain morphology, target object shape/size	Optimization for “optimal grip performance”.	Design optimization, not real-time control. Granular materials are complex to design. DEM modeling typically uses spherical grains for greater accuracy.

GJ—Granular jamming mode in study [39]. EA—Electro Adhesion mode in study [39]. RMS—Root Mean Square. ICP—Iterative Closest Point. FEA—Finite Element Analysis. DEM—Discrete Element Method. NSGA-III—Nondominated Sorting Genetic Algorithm III.

summarizes representative control strategies that have been applied to granular interference grippers and related soft robotic systems over the past decade. Approaches range from simple open-loop pressure modulation and threshold-based force feedback to more sophisticated methods such as model predictive control (MPC), convolutional neural networks (CNNs), and multi-objective evolutionary algorithms. The diversity of input signals (pressure, force, visual or tactile signals) and output actuators (vacuum pumps, solenoid valves, or hybrid drives) reflects the multidisciplinary nature of this field, where mechanical flexibility is leveraged to reduce the computational burden.

Recent advances in adaptive vacuum control algorithms have demonstrated the potential to significantly improve the responsiveness and robustness of jamming-based grippers. For example, Sun et al. [102] introduced an adaptive integral sliding mode control (AISMC) strategy in a vacuum system, which achieved faster pressure tracking and higher stability compared to traditional PID-based approaches. The key advantage of AISMC lies in its ability to dynamically adjust the control law according to the magnitude of the pressure error: large deviations are corrected rapidly, while small deviations are corrected smoothly to mitigate chattering. Such an approach is directly relevant to granular jamming grippers, where rapid yet stable modulation of negative pressure is critical to balance adaptability and stiffness. These adaptive strategies highlight a promising direction for embedding robust controllers into soft robotic systems, ensuring precise vacuum regulation under nonlinearities, leakage disturbances, or unpredictable environmental variations.

8.5. Relationship Between Object Size, Fluidization Frequency, and Gripper Dimensions

Finally, an underexplored but potentially significant aspect is the relationship between the size of the target body, the oscillation frequencies used in fluidization techniques (audio exciter), and the overall dimensions of the gripper. The frequency and amplitude of these stimuli can significantly affect the internal redistribution of grains, facilitating their reorganization to improve coupling to the object. Understanding how to scale these variables based on object and gripper size would allow for the design of more efficient systems capable of performing controlled transitions between fluid and rigid states optimally for different manipulation scenarios.