A Sensorized Mechanically Self-Guided Suction Cup for Improved Adhesion in Complex Environments

Abstract

Octopuses can effectively interact with environments using their agile suction cups, in which abundant neuroreceptors are embodied inside. Inspired by this, we proposed an electronics-integrated self-guided adhesive suction cup (E-SGAS) capable of environmental sensing and adaptively adhesion on diverse surfaces. E-SGAS features an inflatable adhesive membrane and an under-actuated design, enabling it to adapt to various angles and surface roughness under low preloads. A theoretical model is presented to predict self-guided adhesion outcomes. The integrated multilayer stretchable liquid metal sensory circuit (with a maximum deformation rate of 186%) in the adhesive membrane allows for detecting expansion, contact, suction, leakage, and surface roughness. E-SGAS can also process the sensory information to guide intelligent gripping in various complex environments. Experimental results demonstrate the ability of E-SGAS to autonomously grip under a preload force of 0.11 N, a maximum adhesion force of 57.9N, and a detachment force of only 0.34 N. It can adhere to surfaces up to 60-grit roughness and accommodate a surface with a relative angle of 90°. We also show that E-SGAS can capture flying objects or work in a confined space. The proposed adhesion and sensing strategies aim to enhance the performance and expand the application range of suction cup-like grippers. E-SGAS’s results can provide design insights into creating stretchable electronics-integrated bioinspired adhesive systems that can interact with unconstructed environments.

Introduction

Suction grippers are widely used in industrial applications for picking and placing,^1–3 and securing objects to surfaces.⁴ Recently, there have been growing requirements for gripping and manipulation in unstructured environments, such as autonomous warehouse picking,⁵ unmanned aerial vehicle perching,⁶ and aerial grasping.⁷ To achieve a successful attachment, traditional suction devices rely on the perfect seal between the suction cup’s lip and the object’s surface.⁸ However, the current industrial suction cups are limited in attaching to objects with notable orientation misalignment and lack of environmental sensing capacity. These limitations hinder their effective operation in unstructured environments.⁹

Looking into nature for inspiration, as depicted in Figure 1a, the biological octopus suckers are connected to the arm by short muscle bundles and demonstrate high flexibility.^10–12 The octopus sucker is also embodied with ganglia and abundant neuroreceptors.¹³ The neuroreceptors’ inputs project to motoneurons (directly or indirectly via interneurons) to trigger motor responses of the sucker.¹⁴ The short muscle bundles enable the sucker to rotate or extend for attachment on various complex surfaces with misalignments (Fig. 1b). In addition, the soft tissues on the suction cup’s surface ensure an adaptable seal on uneven surfaces.¹⁵ These morphological features enable the octopus sucker’s adaptability during environmental interaction.

FIG. 1.

Design and operating principle of E-SGAS. (a) Schematic illustration of octopus suction cup. (b) After sensing a target, the octopus suction cup can bend to conform to the surface. (c) The overall structure of E-SGAS: the system mainly comprises a pneumatic acetabulum chamber with an adhesion membrane with electronics and a flexible ligament tube connecting the chamber and the fluid port. (d) The adhesion membrane with electronics consists of multi-layer liquid metal circuits and an outer sticky layer. (e) The operating principle of E-SGAS to achieve self-guided adhesion and deadhesion. E-SGAS, electronics-integrated self-guided adhesive suction cup.

In previous studies, mechanical intelligence has been implied, for example, actively extending and granular stalk jamming suction cups can adapt angular and depth errors.^6,16,17 In addition, the surface topography adaptation approaches, including multi-chamber, soft multilayer, adhesive gel lip rings, and membrane-based suction cups, can adapt to objects with varied surface curvature and roughness.^18–22 Dual-layer suction cups enhance anchoring by adjusting to contact impacts,⁷ while combinations of multiple adhesion points with base compliance ensure secure wrap-around attachment.^3,23,24 Despite progress in contact adaptation, the effectiveness of individual suction cups for various adaptive requirements needs to be further improved.

Regarding the sensing and perception capacity of the suction cups, integrated strain sensing in suction cups enables the detection of various suction and object characteristics.^24–27 In addition, optical sensing systems and integrated chips enable sensitive and spatial sensing.^28,29 Further developments, including capacitive proximity sensing and fluid flow monitoring, have been employed to improve pickup posture and optimize positioning.^30–33 Recent advancements have enhanced the environmental sensing of suction cups. However, their over-reliance on perception to adapt to misalignment surfaces presents substantial challenges.

Integrating both mechanical and electronic intelligence into a single suction cup for operation in unconstructed environments remains a major challenge. In this work, we took inspiration from an octopus sucker to combine a sensing/processing network and the adaptive anchoring motion to create an electronics-integrated self-guide adhesion suction cup (E-SGAS) capable of sensing and a versatile, adaptive object adhesive gripping. The paper is organized as follows. Supplementary Section 2 introduces the design and fabrication method of E-SGAS. Supplementary Section 3 presents the theoretical modeling of mechanically self-guided adhesion. Supplementary Section 4 shows the experimental results of the analysis, adhesive performance tests, and sensor’s experiments. In section discussion, we discuss the performance of E-SGAS, including in unstructured settings, where the proposed suction cup was validated by automatically capturing fast-flying balls with challenging contact angles and impact forces and picking up targets in confined spaces. Its sensing and adaptive capabilities in unstructured environments underscore its efficacy in varied contact situations. Finally, we conclude this article. This work introduces a new approach for designing efficient suction cups applicable in unstructured environments, which might shed new light on complex fields such as automated random picking, aerial perching and grasping, and exploring unknown areas.

Design and Manipulate Principle

As shown in Figure 1c, E-SGAS comprises a spiral connecting tube and an acetabular chamber with an adhesion membrane with electronics. Key features include the “adhesion-guided suction” strategy. Pneumatically actuated by one fluid port, the adhesive membrane can swell or retract, which adheres to the target and guides the suction cup to adhere and generate negative pressure. This allows E-SGAS to adapt to a wide range of preloads, contact angles, and surface roughness and produce reliable adhesion force under various conditions. To prevent overstretching of the spring tube, a load fiber is incorporated between the base and the acetabulum. Another featured design is the multilayer liquid metal sensory network embedded in the adhesive membrane of the suction cup (Fig. 1d). This circuit, comprising inner and outer sensing layers and an intermediate circuit layer, mimics the functionality of an octopus sucker nerve, which can perceive and process contact, suction, and leakage sensory information, as well as identify surface roughness. The manufacturing process of E-SGAS is shown in Supplementary Figure S1.

Specifically, the central circuit layer houses essential electronic components such as the microcontroller unit and light-emitting diode (LED) and is flanked by inner and outer spiral liquid metal sensing layers. The advantage of this design is that the deformation of the silicone membrane, whether through expansion, contraction, or contact with a target object, generates different variations in sensing in the inner and outer layers. Meanwhile, the central liquid metal circuit maintains its signal acquisition and processing functionality during deformation. Consequently, the suction cup can autonomously assess its contact, adhesion, or leakage status and convey this information through its built-in indicator light.

Notably, the outermost layer of the silicone membrane possesses adhesive properties, enabling the realization of an adhesion-guided suction strategy, as depicted in Figure 1e. This feature facilitates adaptive attachment and detachment. Specifically, before attempting to adhere to a target, the acetabulum chamber inflates the adhesive membrane. The resulting expanded adhesive membrane adopts a hemispherical shape with low stiffness, facilitating adhesion in a wide range of contact forces, even on rough surfaces (Fig. 1e, Panel I). After the initial adhesion, the adhesive membrane retracts quickly by switching the air pressure to negative pressure, causing the acetabulum section to adjust the contact angle to match the surface (Fig. 1e, Panel II). As the adhesive membrane indents due to negative pressure, it effectively transmits the negative pressure to the cavity formed between the membrane and the target surface. The soft adhesive membrane edge adheres and seals strongly to the target surface, ensuring secure anchoring (Fig. 1e, Panel III). When detachment is necessary, the introduction of air causes the membrane to expand into a hemispherical shape, reducing the contact and adhesion area for effortless detachment (Fig. 1e, Panel IV).

Design analysis

The adhesion process of the E-SGAS can be summarized as the following steps: (1) freely expanding the membrane, (2) contacting with the object, (3) attaching and lifting the object, and iv) detaching. The material properties, thickness, and radius of the membrane structure play a vital role in this adhesion process. Considering that the thickness-to-width ratio of the membrane is less than 0.05, we assume it to be a shell when E-SGAS inflated. To quantitatively understand the deformation of suction cup devices with membrane structures, we established an analytical model of E-SGAS based on finite strain membrane theory and incompressible neo-Hookean material.

Membrane deformation

For hyperelastic membranes, the stress resultants or line tensions $T_{i}$ in the principal directions are defined as,^34,35 $\begin{array}{r} T_{1} = 2 t_{m} λ_{3} (λ_{1}^{2} - λ_{3}^{2}) (U_{1} + λ_{2}^{2} U_{2}) \\ T_{2} = 2 t_{m} λ_{3} (λ_{2}^{2} - λ_{3}^{2}) (U_{1} + λ_{1}^{2} U_{2}) \end{array}$ (1)where $U_{i}$ denotes $\partial U / \partial I_{i}$ , $t_{m}$ is the thickness of the membrane, $λ_{1}, λ_{2}$ and $λ_{3}$ are the principal stretches of the hyperelastic membranes. The equilibrium equation for a membrane in terms of curvature $κ_{i}$ and line tensions are defined as $κ_{1} T_{1} + κ_{2} T_{2} = p$ (2)where $p$ is the lateral pneumatic pressure. For a circle membrane, the change in dimensions along the principal directions is assumed to be the same (Fig. 2a). Therefore, the principal stretches $λ_{1}$ and $λ_{2}$ can be expressed as $λ_{1} = λ_{2} = λ$ (3)

FIG. 2.

Analytical and experimental overview of E-SGAS. (a) Circular membrane, undeformed. (b) Membrane under positive pressure, hemispherical expansion. (c) Hemispherical membrane contacts with a flat surface. (d) Acetabular force analysis when self-guided adhesion under negative pressure. (e, f) Self-guided adhesion: successful (e) and failed (f) cases. (g, h) Air pressure and extrusion depth impacts: (g) on membrane expansion during inflation, (h) on contact area during contact stage. (i) FTIR optical image of inflated membrane contact. (j, k) Adhesive membrane guidance force experiments: (j) varying deflation pressures and (k) varying adhesion areas. (l, m) Extrusion depth and connection stiffness impacts on self-guided adhesion: (l) analytical, (m) experimental results. E-SGAS, electronics-integrated self-guided adhesive suction cup.

The membrane material is assumed to be isotropic and incompressible. For an incompressible material, the principal stretch $λ_{3}$ can then be written as $λ_{3} = {(λ_{1} λ_{2})}^{- 1} = \frac{1}{λ^{2}}$ (4)

The strain energy density function $U$ for a neo-Hookean material,³⁶ is defined by $U = \frac{μ}{2} (I_{1} - 3)$ (5)

After substituting (5) into (1), the line tensions can be expressed as $T_{1} = T_{2} = μ t_{m} (1 - \frac{1}{λ^{6}})$ (6)

Now, the only undefined parameter in (2) is curvature $κ_{i}$ . According to the membrane’s different deformation conditions, we define the curvature for each of the three states of the actuator respectively.

Free-space deformation

Firstly, during the freely expanding process, the membrane was assumed to be inflated into a circular curve with curvature $κ_{1} = κ_{2} = 1 / R$ (see Fig. 2b). The radius of the curve $R$ can be expressed as $R = \frac{a}{\sin θ_{m}}$ (7)the swept angle $θ_{m}$ and cord length $a$ is given in Figure 2b.

The circumferential strain $λ_{1}$ can be written as $λ_{1} = λ = \frac{θ_{m}}{\sin θ_{m}}$ (8)

The maximum inflation $δ$ of the membrane is derived from the deformed geometry as $δ = atan (\frac{θ_{m}}{2})$ (9)

After substituting the curvatures into (2), we have $\frac{2}{R} T_{1} = p$ (10)

Equation (11) is the final governing equation for the free-space deformation of the membrane. After substituting (3), (4), (6), (7), and (8) into (10), we have $\frac{2 \sin θ_{m}}{a} μ t_{m} (1 - \frac{{\sin θ_{m}}^{6}}{{θ_{m}}^{6}}) = p$ (11)which is reduced to a nonlinear algebraic equation.

With Equation (11), the actuator’s expanding radius $R$ can be derived from the provided pneumatic pressure.

Membrane in contact with substrate

When the membrane is inflated and in contact with a plane, from the deformed geometry (see Fig. 2c), we have $R = \frac{d}{1 - \cos θ_{m}}$ (12) $c_{a} = a - d \cot (\frac{θ_{m}}{2})$ (13)and $λ_{1} = 1 + \frac{d}{a} [\frac{θ_{m}}{2} \csc^{2} (\frac{θ_{m}}{2}) - \cot (\frac{θ_{m}}{2})]$ (14)

After substituting (3), (4), (6), (12), and (14) into Equation (10), we can calculate the actuator’s expanding radius R with given pneumatic pressure p. The contact region exhibits a circle shape, as validated experimentally in Figure 2i. The contact area $A_{c}$ for a circle region can then be calculated as $A_{c} = π c_{a}^{2}$ (15)

The contact force $F_{c}$ can be calculated as $F_{c} = p A_{c}$ (16)

From the above conclusion, it is evident that the contact area is determined by both the air pressure $p$ and the extrusion is determined by both the air pressure $p$ and the extrusion displacement $d$ .

Adhesion-guided process

After the contact and initial adhere state mentioned above, we switched the air pressure p to negative pressure $p_{n}$ for the sucker’s further anchoring on the surface (Fig. 2d). To ensure successful sealing between the viscous membrane and the target surface (Fig. 2e), the initial adhesion must be robust enough to withstand the pull exerted by the spring (with a stiffness of $K$ ) connected to the rear of the suction cup. Otherwise, the membrane may detach from the surface, resulting in a failure (Fig. 2f).

The adhesion—delamination process of a prestressed circular film can be described as two parts: stretching and delamination process.³⁷ In our membrane spring system, the delamination is unstable due to the large air pressure changes in the membrane cavity. When delamination occurs, adhesion fails. Thus, the maximum adhesion of the film before delamination is critical to this analysis.

We consider the maximum force $F_{M}$ that the initial adhesion between the membrane and the target surface function of the contacting area $A_{c}$ , as $F_{M} = f (A_{c})$ (17)

The spring tension $F_{s}$ increases during the adhesion-guiding process, and finally reaches its peak when the suction cup is pulled to the surface (Fig. 2e), at this time, $F_{s} = K d$ (18)

To achieve a successful adhesion-guiding process, $F_{M} + F_{p}$ > $F_{s}$ should be satisfied.

We later experimentally measured the values of $F_{M} + F_{p}$ in various states, and established the failure boundary conditions of the suction cup under different extrusion displacements.

Experiment Results

Analysis

The theoretical and experimental mid-plane inflation of the membrane is presented in Figure 2g. The theoretical and experimental deformed profiles for the actuators in contact with a rigid flat substrate are presented in Figure 2h. The theoretical curves presented are derived from Equations 11 and 14. The specific parameters used for calculating the analytical model results are as follows: cord length ( $a$ ) is 20 mm, membrane thickness ( $t_{m}$ ) is 3 mm, and the neo-Hookean model parameter ( $μ$ ) is $3.24 \times 10^{4}$ . Experimental data were collected under a constant air pressure of 10 kPa. The experimental results align well with the theoretical predictions within the pressure range of 6–12 kPa. Two factors may account for discrepancies in the analytical model: the assumption of a thin membrane in finite strain membrane theory and the nonlinear behavior of materials during large deformations. The image of the contact area is shown in Figure 2i, and more images are shown in Supplementary Figure S5 and Supplementary Movie S5.

According to the experimental results, it is observed that the adhesive guiding force of the membrane exhibits a positive correlation with the deflation pressure (Fig. 2j) and the contact area (Fig. 2k). After keeping the deflation pressure constant, we established the relationship between adhesive guiding force and contact area through numerical fitting (Fig. 2k). Based on the above fitting results and the previous analysis model, the theoretical prediction failure boundary curve of adhesion-guided suction is presented in Figure 2l. The experimental results were explored by adjusting the spring stiffness and extrusion displacement, as shown in Figure 2m. The predicted failure boundary curve has the same trend as the experimental result curve. The results indicate that the low-stiffness connection between the suction cup and the base facilitates successful self-guided adhesion with minimal contact displacement, corresponding to lower preload conditions. The experimental success and failure cases of self-guided adhesion were recorded in Supplementary Movie S6.

Contact force adaptation

Effective preload adaptability enhances the suction cup’s capability to manage diverse contact scenarios. It can be easily observed from the experimental results shown in Figure 2m that the required extrusion displacement for successful self-guided adhesion varies under different connection stiffness. When the connection stiffness is 0.18 N/mm, the suction cup only needs an extrusion displacement of 1.5 mm to complete self-guided adhesion. Considering the relationship between extrusion displacement and contact force, it can be found that reducing the connection stiffness can effectively reduce the required preload, with a minimum of 0.11 N as shown in subsequent mechanical experiments.

Contact angle error adaptation

As Figure 3a shows, the suction cup moves vertically toward the 45° inclined surface. Results indicate that at a 45° contact angle error, the bellows suction cup partially adapts due to its compliance during pressing (see Fig. 3b). However, only part of the lip ring adheres to the surface, hindering the suction cup from forming a complete seal.

FIG. 3.

The contact angle adaptation principle of E-SGAS. (a) Experiment design schematic: Both the commercial bellows suction cup and E-SGAS approach the FTIR plate at a 45° angle. (b, c) Optical views including side view and bottom view during testing: commercial bellows (b) and E-SGAS (c). The commercial bellows fail to adapt to the angle, whereas E-SGAS adjusts via two-stage rotation, achieving successful attachment. E-SGAS, electronics-integrated self-guided adhesive suction cup; FTIR, frustrated total internal reflection.

In contrast, E-SGAS adapts to a broader range of contact angle errors through passive deformation and active pulling, facilitating effective sealing and suction (see Fig. 3c). As the suction cup moves downward, its expanded adhesive membrane first contacts the target surface, forming an oval-shaped adhesion area. This adhesion prevents relative sliding between the suction cup and the target surface. During further descent, the membrane stays adhered, and the spring tube enables the suction cup’s rotation, facilitating the first phase of angular adaptation. Subsequently, the negative pressure causes the adhesive membrane to retract, pulling the suction cup towards the target surface, completing the second stage of angular adaptation, and forming a robust annular seal. In addition, our suction cup’s adhesive membrane produces a brighter light spot on the frustrated total internal reflection (FTIR) device compared to bellows suction cups, indicating more effective contact. This echoes the suction cup’s excellent sealing and adaptability to minor preload. This experiment was recorded in Supplementary Movie S3.

Mechanical evaluation of surfaces with different roughness

The initial experiment was conducted with the suction cup at 0 kPa air pressure to ensure adhesive membrane positioning. The robotic arm was programmed to descend the suction cup at 20 mm/s, halting upon contact with the acrylic surface. Subsequent air pressure adjustment within −20 kPa and a consistent lift-off speed of 20 mm/s allowed for the measurement of the force required to detach the suction cup from the surface, as illustrated in Figure 4a. The findings reveal a negligible pull-off force (approaching 0 N) in the absence of negative pressure, likely due to insufficient contact and adhesion formation. Similarly, at −2 kPa, the adhesion remained ineffective. However, a significant change occurred at −4 kPa, where the suction cup’s adhesion force noticeably surpassed the force generated by negative pressure alone (as indicated by the gray dashed line data). As the negative pressure further increases, the adhesion force also increases following the slope of the theoretical vacuum suction curve. Notably, the combination of adhesion and vacuum forces creates a substantial anchoring force, with the most pronounced enhancement observed at lower negative pressures.

FIG. 4.

Mechanical performance of E-SGAS. (a) Negative driving pressure’s impact on adhesion force: the blue line for adhesion force and the gray dashed line for theoretical negative pressure force. (b) Positive driving pressure’s impact on release force. (c) E-SGAS adhesion force on varied roughness surfaces. (d) Comprehensive comparison of preload, adhesion force, and release force in the working process of E-SGAS.E-SGAS, electronics-integrated self-guided adhesive suction cup.

In the subsequent experiment, the suction cup, initially adhered to the acrylic surface at −20 kPa, was subjected to increasing positive air pressures, starting from 0 kPa. The detachment force required to separate the suction cup from the surface was recorded while lifting it at 20 mm/s, as shown in Figure 4b. At 0 kPa, the suction cup exhibited a substantial separation force, exceeding 7 N. A notable decrease in release force was observed as the driving air pressure increased to 4 kPa. When the pressure reached 8 kPa, the required force to detach the suction cup was reduced to less than 1 N. These results stem from the membrane’s middle part expansion under positive pressure, leading to the lip ring’s detachment from the surface. The greater the adhesive membrane’s bulge, the more pronounced the lip ring separation. Driving air pressure of 8 kPa effectively reduces the separation force, thereby facilitating easier detachment of the suction cup.

In the third experiment, we evaluated the suction cup’s performance on surfaces of varying roughness. Initially, a 10 kPa air pressure was applied to expand the membrane. The robotic arm then maneuvered the suction cup to vertically approach each surface at 20 mm/s, stopping as soon as the membrane made contact. This approach ensured a consistent, minimal preload force. After setting the air pressure to −20 kPa for self-guided adhesion, the suction cup was retracted at 20 mm/s, and the force necessary for interface separation was recorded (see Fig. 4c). The results indicate that the suction cup effectively adheres to surfaces ranging from smooth acrylic to the roughest 60-grit sandpaper. Notably, as for the sandpaper, the adhesion force slightly decreases with increasing surface roughness, likely due to increased leakage opportunities on rougher surfaces diminishing adhesion efficiency. Interestingly, the adhesion force on sandpaper exceeded that on a smooth surface, possibly because the sandpaper’s higher tangential friction impedes the lip ring’s contraction during detachment, thereby enhancing adhesion force. This test was recorded in Supplementary Movie S4.

By comparing the preload, adhesion, and release forces of the suction cup (Fig. 4d), it is observed that the suction cup achieves self-guided adhesion with a preload under 0.11 N, generating an adhesion force of 57.9 N (with an actuation pressure of −60 Kpa). The required separation force during detachment is less than 0.34 N. These results highlight the efficiency of the adhesion-guided negative pressure suction strategy, achieving a balance between ease of adhesion and strong anchoring, demonstrated by a preload to adhesion force conversion of 526 times and an adhesion to release force transition of 170 times. The detachment operation was recorded in Supplementary Movie S7. The mechanical evaluation setup is shown in Supplementary Figure S6.

Adhesion adaptability demonstration

The grasping capabilities of our suction cup are depicted. Figure 5a–d illustrates the process of the suction cup, attached to a robotic arm, successfully picking up a right-angled triangular prism (with acute angles of 20° and 70°) from a horizontal table under various placements, corresponding to contact angle errors of 0°, 20°, 70°, and 90°. The successful acquisition of the prism in all instances demonstrates the suction cup’s ability to handle objects irrespective of their orientation, as further exemplified in Supplementary Movie S3.

FIG. 5.

Diverse E-SGAS performance scenarios. (a–d) E-SGAS adeptly handles a triangular prism in multiple poses, showcasing an adaptive picking up with the contact angle from 0° to 90°. (e–h) The versatile grasping capabilities of E-SGAS are displayed through successfully handling a variety of objects, such as delicate fruit, pliable snack bags, small keys, and coarse stones. E-SGAS, electronics-integrated self-guided adhesive suction cup.

Figure 5e–h expands on this capability, showing the suction cup grasping a range of objects including fragile fruit, a soft snack bag, a small key, and a rough stone (weight 1826 g). These demonstrations underscore the suction cup’s proficiency in securely anchoring a wide range of objects. The adhesion strategy employed allows for the successful pickup of objects with challenging shapes and sizes that might otherwise be difficult to grip through sealed suction. Its adaptability to rough surfaces enhances its performance in grasping both rough and heavy objects. Overall, these results indicate that the suction cup is capable of reliably grasping a diverse array of items, regardless of size, hardness, roughness, or positioning, with minimal operational complexity.

Multi-state sensorized suction cups

As shown in Figure 6a, the membrane is capable of sensing and processing during undergoing both circumferential and radial stretching, with the area expanding up to 186% compared to its initial state. To assess the sensing capabilities of the suction cup across three key phases—expansion, contact, and suction, we conducted signal calibration in each phase.

FIG. 6.

Performance of the multilayer liquid metal network. (a) The circuit membrane with multi-layer liquid metal sensing circuit, MCU, and LEDs was inflated to expand to 186% of its original surface area. (b, c) Electronic membrane inflation experiment: (b) schematic diagram and (c) resistance variations in inner and outer layer sensors relative to inflation pressure. (d, e) Electronic membrane inflation contact experiment: (d) schematic view and (e) sensor resistance response to contact force. (f, g) Leakage sensing experiment: (f) schematic representation and (g) sensor resistance alterations in response to gas leakage volume. (h, i) Electronic membrane for multi-state sensing: monitoring the E-SGAS’s state, guiding the grasping process, and varying colors of indicator lights. E-SGAS, electronics-integrated self-guided adhesive suction cup; MCU, microcontroller unit; LEDs, light-emitting diodes.

In the membrane inflation stage, we assessed the performance of the sensing signals at various inflation air pressures. As illustrated in Figure 6b, the membrane inflation pressure was increased from 0 kPa to 12 kPa at intervals of 1 kPa. As depicted in Figure 6c, when the input air pressure was less than 12 kPa, the change in resistance of the inner sensor remained smaller than that of the outer sensor due to the various degrees of deformation in the two sensors. The resistance changes gradually approached as the pressure increased because the membrane thickness decreased with increasing air pressure, leading to the approaching of the deformation (curvature) of the inner and outer sensing layers.

Figure 6d illustrates a schematic diagram of the suction cup contacting a plane. Initially, the membrane expanded and the robot arm placed the suction cup at the position where the membrane was tangent to the plane. Subsequently, the robotic arm drove the suction cup downward in increments of 0.5 mm each, resulting in a total displacement of 5 mm. The expanded membrane deformed due to the contact force (Fc), measured by the force sensor. This entire process was repeated at five different constant air pressures (7–11 kPa). From Figure 6e, it was evident that the change in resistance of the outer sensor increased substantially due to squeezing, while the inner sensing layer remained almost unaffected. Two sensing layers configuration effectively decoupled the act of touching the object from the inflation of the suction cup. Observing curves of different color depths, it was found that when the inflation air pressure was less than 9 kPa, the sensitivity of the outer sensing layer increased with higher inflation air pressure. In other words, under the same contact force, the resistance change of the outer sensor was larger when the pressure was higher. Conversely, when the inflation air pressure exceeded 9 kPa, the curves representing the outer sensor overlapped, indicating there was no longer a large variability of sensitivity along with the increase in inflation air pressure.

To assess the leakage signal detected by the suction cup, an artificial leakage was created between the suction cup and the plane, as depicted in Figure 6f. The center of the plane was connected to a syringe which regulated the volume of gas leakage (V). The inner cavity of the suction cup maintained a negative pressure (P) of −20 kPa. As illustrated in Figure 6g, as the leakage volume increased, the signal of the inner sensor exhibited continuous growth due to the inward membrane deformation. In contrast, the outer layer sensing resistance displayed a tendency to decrease initially and then increase. The suction cup was maintained in a suction state until the leakage volume reached 3 mL. In the case of complete leakage, the inner sensor and the inner surface of the acetabulum became fully affixed. The liquid metal channel for the inner sensor was squeezed, leading to a sudden spike in the resistance change. This result indicated the suction cup’s capability to detect suction status or complete leakage.

Figure 6H demonstrated that, by processing the sensing signal, the suction cup could function autonomously, with its status indicated by the LED light’s color. Initially, the membrane expanded, and the LED displayed a green light, signifying its readiness to contact the target (Fig. 6h, Panel I). Subsequently, as the suction cup moved downward and recognized the contact, the LED changed to yellow (Fig. 6h, Panel II). Following this, the air pressure inside the suction cup was transformed into negative pressure, guiding the suction cup to perform vacuum suction. If the suction was successful, the LED turned blue (Fig. 6h, Panel III); conversely, if the suction failed, the LED turned red (Fig. 6h, Panel IV). Figure 6i illustrates the variations in sensing signals throughout the experiment. Signals from the inner and outer layers collectively indicated the suction cup’s status. The flexible electronic membrane processed these signals in real-time, thereby facilitating illumination control and directing the suction cup toward its subsequent action. Further demonstration of membrane deformation and multi-state sensing of E-SGAS was recorded in Supplementary Movie S2 and Supplementary Figure S2. For more sensing principles, see Supplementary Figure S3 and Supplementary Figure S4.

Discussion

Demonstration of potential use case

To evaluate the efficacy of E-SGAS in managing non-standard contacts, we performed experiments in unstructured environments, including capturing flying balls and grasping in constrained settings.

In the flying ball capturing test, we mounted E-SGAS on a stationary robotic arm, initially inflating it to 10 kPa. The system was programmed to automatically reduce the air pressure to −20 kPa when contact was detected. Then a small ball was propelled toward the suction cup at a speed of 14 m/s from the direction of a large contact angle error (Fig. 7a). As detailed in Figure 7b, upon collision, the ball engaged with the suction cup’s hemispherical adhesive membrane, which absorbed a portion of the ball’s kinetic energy, averting rebound. In response to the impact, the suction cup transitioned to negative pressure suction, effectively securing the ball. The sensing signals for each stage are illustrated in Figure 7c. Here, the background color, matching the light color of E-SGAS, denotes the respective suction stages.

FIG. 7.

Demonstration of the E-SGAS’s performance in unstructured settings. (a, b) E-SGAS automatically detects and captures flying balls traveling at 14 m/s from a lateral direction. (c) Changes in sensing signals during ball capture. (d) Diagram illustrating the non-cooperative target in a narrow cave. (e) With an extended rod, E-SGAS automatically targets acquisition within a cave. (f) Sensory feedback during the grasping operation inside the cave. E-SGAS, electronics-integrated self-guided adhesive suction cup.

In the confined environment experiment, we constructed a vertically elongated and narrow model to mimic caves and fissures commonly observed in nature. Positioned at the base of this model was a sleek, triangular pyramid, serving as the non-cooperative target for grasping, illustrated in Figure 7d. Operators delivered the E-SGAS to the bottom of the cavity using an extended rod. When the E-SGAS contacted the pyramid, it executed an automated self-guiding adhesion process, showcased in Figure 7e. Subsequently, the anchored pyramid was successfully extracted from the cavity. The sensor data showed the phases of expansion, contact, and adhesion, as presented in Figure 7f. The sensing and adaptive capturing capabilities of the suction cup in dynamic and unstructured situations are further illustrated in Supplementary Movie S1.

The example above illustrates that E-SGAS exhibits multi-type adaptability and multi-state sensing capabilities, offering significant advantages for interacting with complex objects in various environments. In the future, these suction cups could enhance industrial efficiency in applications such as multi-product production line sorting, construction material handling, robotic operations, and aerial grasping.

Comparison of existing work

Inspired by the morphological structure and sensory capabilities of octopus suckers, we developed a mechanically self-adaptive suction cup (E-SGAS). The highly deformable adhesion membrane with electronics of E-SGAS facilitates the self-guided adhesion strategy and incorporates a highly stretchable liquid-metal-based multimodal sensing capacity. To a certain extent, this prototype combines physical intelligence and computational intelligence for environmental interaction. E-SGAS exhibits adaptability across various situations (different contact angles, surface roughness, and preload), outperforming previously developed adaptive and sensor-equipped suction cups (see Table 1). In addition, E-SGAS effectively blends three essential characteristics: easy adhesion with small preloads, robust attachment with high adhesive force, and controllable detachment.

Table 1.

Comparison of Adaptable or Sensorized Suction Cups

Reference	Contact angle	Roughness	Preload requirements	Sensors per suction cup	Perception	Onboard data processing
¹⁶	85°	120 grid	<0.5 N	0	—	No
¹⁷	60°	—	—	0	—	No
²²	30°	180 grid	∼5 N	0	—	No
²⁹	30°	—	0.5 N (on the 5° substrate)	1	Proximity, contact	No
²⁸	—	—	—	1	Proximity, contact	No
²⁵	—	—	—	1	Contact	No
²⁴	—	—	—	2	Suction, temperature, stiffness	No
⁴	—	—	—	2	Suction, leakage	No
³¹	—	—	—	3	Proximity, contact	No
²⁶	—	—	—	4	Contact, direction, inclination, stiffness	No
³³	30°	—	—	4	Roughness, curvature	No
²⁷	—	—	∼0.5 N	4	Contact, direction, inclination, weight	No
This work	90°	60 grid	0.11 N	2	Contact, suction, leakage, roughness	Yes

As a result, E-SGAS can perform high-dynamic adhesive gripping, such as capturing flying balls, which other suction cups have yet to achieve. In addition, the inflatable membrane with multi-layer sensing enables E-SGAS to sense various adhesive states (contact, leak, and attach to surfaces with different roughness). Based on onboard data processing, the E-SGAS utilizes its deformable electronic membrane to generate distinct signals in different suction states. It displays a responsive reaction to environmental interactions, enhancing the soft system’s intelligence and flexibility. The integrated circuit (IC) design also mitigates interface and power disturbances from external wires, reducing the need for external equipment and improving integration and usability. Regarding current limitations, E-SGAS necessitates an external pneumatic supply, partially limiting higher-level system integration,³⁸ and flexible wireless configuration.³⁹ For a future study, we suggest applying more advanced technologies such as soft pumps,⁴⁰ and stretchable soft battery,⁴¹ to realize a fully wireless, untethered, and closed-loop suction cup module capable of sensing, decision-making, and actuating. This would allow for more versatile applications in unstructured environments.

Modeling the self-guided adhesion

Our analytical model can be applied to directing the suction cup joint’s stiffness design and pinpointing the minimum preload for self-guided adhesion. Based on hyperelastic membrane theory, we derived an approximate solution for our prototype’s deformed elastic thin membrane. However, there are some limitations regarding this model. The thickness of the device membrane exceeds the boundary conditions of the thin membrane theory to a certain extent. The model’s accuracy improves as the elastic membrane inflates and the thickness-width ratio approaches that of an ideal hyperelastic membrane; conversely, smaller expansions result in lower analytical accuracies (see Fig. 2g). When the thickness-width ratio of the elastic membrane deviates significantly from a hyperelastic membrane, these assumptions may require reconsideration. In addition, the predicted failure boundary curve does not correspond well to the experimental result curve, when the extrusion displacement is over 3 mm (Fig. 2m). Accurately describing the non-quasi-static membrane detachment during the self-guided adhesion process via the analytical model is still challenging. In the future, we will perform mode detailed investigations concerning systematic experiments and analyses on the adhesive membrane adhesion’s highly dynamic separation process.

Adaptable adhesion

During self-guided adhesion, the spiral structure offers redundancy in degrees of freedom, and the adhesive membrane guides the suction cup to the target surface (refer to Fig. 1e). This design facilitates adaptive anchoring. However, it is important to consider that the adhesive layer may gradually accumulate dust, lint, and other particles with repeated use, leading to diminished adhesion. This challenge could be mitigated by engineering the adhesive layer to be easily cleanable or replaceable. Beyond adaptive adhesion, the biological octopus’ sucker is also capable of executing functions like rotation, twist, and tuning stiffness. Future prototypes may consider related approaches (like preprogrammed structures) integrating with an easily integrable actuation,⁴² enhancing post-adhesion stiffness, and expanding operational capabilities. Furthermore, octopuses utilize multiple suckers for crawling and in-arm manipulation in unstructured environments.¹⁰ In practical applications, an adhesion-guided suction design powered by a single fluid port is expected to combine ease of control with behavioral flexibility, particularly when extended to multi-suction cup systems. Consequently, a bioinspired soft robot equipped with multiple E-SGAS is anticipated to perform a diverse range of operational tasks in unstructured settings.

Conclusion

This study introduces E-SGAS, a new suction cup designed for autonomous and adaptive suction. With the adhesion-guided suction strategy, E-SGAS exhibits strong adhesion capabilities (maximum adhesion force 57.9 N, equivalent to 275 times its weight), with a minimal preload (minimum 0.11 N) and release force (minimum 0.34 N) requirements, and can handle targets with up to 90° contact angle error and 60-grit roughness. E-SGAS enables reliable anchoring and easy release, all powered by a single fluidic port, making it user-friendly and easily deployable. The embedded multi-layer liquid metal sensing circuits and IC chips facilitate the perception and processing of information related to expansion, contact, adhesion, leakage, and attachments on surfaces with different roughness, equipping E-SGAS with the ability for automatic adhesion. Demonstrations with E-SGAS, including capturing flying objects and retrieving items from constrained spaces, showcase its proficiency in unstructured settings. E-SGAS shows promise in enhancing robotic efficiency and functionality in diverse unstructured environments, such as automated random picking, aerial perching and grasping, and exploring unknown territories.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

Authors’ Contributions

F.Y., L.T., and H.X. conducted the design, fabrication, and experiment. F.Y. and H.X. and conducted the design analysis. L.T. built the FEA environment and conducted the simulation. F.Z., W.W., and Z.X. assisted the design and fabrication of the prototype. B.Y., T.W., X.D., and L.W. initiated and directed the study. F.Y. wrote the original article, and L.W. reviewed and edited the final article.

Funding Information

This work was supported in part by the National Key R&D Program of China under Grants 2024YFB4707300, 2022YFB4701800, and in part by the National Science Foundation support projects, China under Grant T2121003, Grant 92048302, and Grant 61822303.

Supplementary Material

References

Mueller

, Behrens

, Elkmann

. A new multi-modal approach towards reliable bin-picking application. In: Proceedings of ISR 2016: 47st International Symposium on Robotics VDE; 2016; pp. 1–7.

Schwarz

, Milan

, Lenz

, et al. Nimbro Picking: Versatile part handling for warehouse automation. In: 2017 IEEE International Conference on Robotics and Automation (ICRA) IEEE; 2017; pp. 3032–3039.

Xie

, Domel

, An

, et al. Octopus arm-inspired tapered soft actuators with suckers for improved grasping. Soft Robot, 2020; 7(5):639–648; doi: 10.1089/soro.2019.0082

, Liu

, Tian

, et al. An Aerial–aquatic hitchhiking robot with remora‐inspired tactile sensors and thrust vectoring units. Adv Intell Syst, 2023:2300381.

Hao

, Biswas

, Hawkes

, et al. A multimodal, enveloping soft gripper: Shape conformation, bioinspired adhesion, and expansion-driven suction. IEEE Trans Robot, 2021; 37(2):350–362; doi: 10.1109/TRO.2020.3021427

, Wang

, Zhang

, et al. Aerial-aquatic robots capable of crossing the air-water boundary and hitchhiking on surfaces. Sci Robot, 2022; 7(66):eabm6695.

Liu

, Dong

, Ma

, et al. Adaptive aerial grasping and perching with dual elasticity combined suction cup. IEEE Robot Autom Lett, 2020; 5(3):4766–4773; doi: 10.1109/LRA.2020.3003879

Monkman

, Hesse

, Steinmann

, et al. Robot Grippers. John Wiley & Sons; 2007.

Xiang

, Wang

, Li

, et al. Classification-design-optimization integrated picking robots: A review. J Intell Manuf, 2024; 35(7):2979–3002.

10.

Kier

, Smith

. The structure and adhesive mechanism of octopus suckers. Integr Comp Biol, 2002; 42(6):1146–1153; doi: 10.1093/icb/42.6.1146

11.

Grasso

. Octopus sucker-arm coordination in grasping and manipulation*. Am Malacol Bull, 2008; 24(1):13–23; doi: 10.4003/0740-2783-24.1.13

12.

Kier

, Stella

. The arrangement and function of octopus arm musculature and connective tissue. J Morphol, 2007; 268(10):831–843; doi: 10.1002/jmor.10548

13.

Graziadei

PPC

, Gagne

. Sensory innervation in the rim of the octopus sucker. J Morphol, 1976; 150(3):639–679; doi: 10.1002/jmor.1051500304

14.

Ramdya

, Ijspeert

. The neuromechanics of animal locomotion: From biology to robotics and back. Sci Robot, 2023; 8(78):eadg0279; doi: 10.1126/scirobotics.adg0279

15.

Tramacere

, Kovalev

, Kleinteich

, et al. Structure and mechanical properties of Octopus vulgaris suckers. J R Soc Interface, 2014; 11(91):20130816; doi: 10.1098/rsif.2013.0816

16.

Goshtasbi

, Sadeghi

. A bioinspired stiffness tunable sucker for passive adaptation and firm attachment to angular substrates. Front Robot AI, 2023; 10:1080015.

17.

Yoo

, Eom

, Park

, et al. Compliant suction gripper with seamless deployment and retraction for robust picking against depth and tilt errors. IEEE Robot Autom Lett, 2023; 8(3):1311–1318; doi: 10.1109/LRA.2023.3238903

18.

, Song

, Ma

, et al. A bioinspired adhesive sucker with both suction and adhesion mechanisms for three-dimensional surfaces. J Bionic Eng, 2022; 19(6):1671–1683; doi: 10.1007/s42235-022-00238-5

19.

Tsukagoshi

, Osada

. Soft hybrid suction cup capable of sticking to various objects and environments. Actuators, 2021; 10(3):50; doi: 10.3390/act10030050

20.

Song

, Drotlef

, Son

, et al. Adaptive self‐sealing suction‐based soft robotic gripper. Adv Sci, 2021; 8(17):2100641; doi: 10.1002/advs.202100641

21.

Yue

, Si

, Partridge

, et al. A Contact-triggered adaptive soft suction cup. IEEE Robot Autom Lett, 2022; 7(2):3600–3607; doi: 10.1109/LRA.2022.3147245

22.

Yue

, Bloomfield-Gadêlha

, Rossiter

. Shape-Conformable suction cups with controllable adaptive suction on complex surfaces. IEEE Robot Autom Lett, 2023; 8(11):7735–7742; doi: 10.1109/LRA.2023.3322071

23.

Mazzolai

, Mondini

, Tramacere

, et al. Octopus‐inspired soft arm with suction cups for enhanced grasping tasks in confined environments. Adv Intell Syst, 2019; 1(6):1900041; doi: 10.1002/aisy.201900041

24.

Xie

, Yuan

, Liu

, et al. Octopus-inspired sensorized soft arm for environmental interaction. Sci Robot, 2023; 8(84):eadh7852; doi: 10.1126/scirobotics.adh7852

25.

Aoyagi

, Suzuki

, Morita

, et al. Bellows suction cup equipped with force sensing ability by direct coating thin-film resistor for vacuum type robotic hand. IEEE/ASME Trans Mechatron, 2020; 25(5):2501–2512; doi: 10.1109/TMECH.2020.2982240

26.

Shahabi

, Visentin

, Mondini

, et al. Octopus‐inspired suction cups with embedded strain sensors for object recognition. Adv Intell Syst, 2023; 2200201; doi: 10.1002/aisy.202200201

27.

Lee

, Baik

, Hwang

, et al. An electronically perceptive bioinspired soft wet-adhesion actuator with carbon nanotube-based strain sensors. ACS Nano, 2021; 15(9):14137–14148; doi: 10.1021/acsnano.1c05130

28.

Sareh

, Althoefer

, Li

, et al. Anchoring like octopus: Biologically inspired soft artificial sucker. J R Soc Interface, 2017; 14(135):20170395; doi: 10.1098/rsif.2017.0395

29.

Frey

, Haque

ABMT

, Tutika

, et al. Octopus-inspired adhesive skins for intelligent and rapidly switchable underwater adhesion. Sci Adv, 2022; 8(28):eabq1905; doi: 10.1126/sciadv.abq1905

30.

Huh

, Sanders

, Danielczuk

, et al. A Multi-chamber smart suction cup for adaptive gripping and haptic exploration. In: 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). 2021; pp. 1786–1793; doi: 10.1109/IROS51168.2021.9635852

31.

Doi

, Koga

, Seki

, et al. Novel proximity sensor for realizing tactile sense in suction cups. In: 2020 IEEE International Conference on Robotics and Automation (ICRA). 2020; pp. 638–643; doi: 10.1109/ICRA40945.2020.9196726

32.

Nadeau

, Abbott

, Melville

, et al. Tactile sensing based on fingertip suction flow for submerged dexterous manipulation. In: 2020 IEEE International Conference on Robotics and Automation (ICRA). 2020; pp. 3701–3707; doi: 10.1109/ICRA40945.2020.9197582

33.

Lee

, Lee

, Huh

, et al. Haptic search with the smart suction cup on adversarial objects. IEEE Trans Robot, 2024; 40:226–239; doi: 10.1109/TRO.2023.3331063

34.

Libai

. The Nonlinear Theory of Elastic Shells: One Spatial Dimension. Elsevier; 2012.

35.

Kydoniefs

. Finite axisymmetric deformations of an initially cylindrical elastic membrane enclosing a rigid body. Q J Mech Appl Math, 1969; 22(3):319–331.

36.

Ogden

. Non-Linear Elastic Deformations. Courier Corporation; 1997.

37.

Anonymous. Full Article: Wong M, Duan G, Wan K. Adhesion–delamination mechanics of a prestressed circular film adhered onto a rigid substrate. The Journal of Adhesion 2007;83(1):67–83. n.d. Available from: https://www.tandfonline.com/doi/full/10.1080/00218460601102878 [Last accessed: January 24, 2024].

38.

, Mayekar

, Aguasvivas

, et al. Electro-hydraulic rolling soft wheel: Design, hybrid dynamic modeling, and model predictive control. IEEE Trans Robot, 2022; 38(5):3044–3063.

39.

Eshaghi

, Rogers

, Nejat

, et al. Closed-Loop motion control of robotic swarms – a tether-based strategy. IEEE Trans Robot, 2022; 38(6):3564–3581; doi: 10.1109/TRO.2022.3181055

40.

Cacucciolo

, Shintake

, Kuwajima

, et al. Stretchable pumps for soft machines. Nature, 2019; 572(7770):516–519; doi: 10.1038/s41586-019-1479-6

41.

, Wang

, Zhang

, et al. A full-device autonomous self-healing stretchable soft battery from self-bonded eutectogels. Adv Mater, 2023; 35(6):e2208392; doi: 10.1002/adma.202208392

42.

Jiang

, Luo

, Li

, et al. A novel scaffold-reinforced actuator with tunable attitude ability for grasping. IEEE Trans Robot, 2023; 39(2):1164–1177; doi: 10.1109/TRO.2022.3200550