Fish-scale-like gaps help soft robotic grippers sense bending, texture, and firmness through small contact changes.
(Nanowerk Spotlight) A fruit-picking robot can see a fruit before it picks it up, but vision cannot reliably tell how the fruit will respond to a squeeze. Firmness, subtle surface texture, and the first signs of slipping appear only during contact. Human fingers read those cues through tiny skin deformations and vibrations during motion. For robotic systems that handle delicate or variable objects, useful touch sensors must capture similar mechanical detail while surviving repeated grasping, sliding, and bending.
That requirement creates a difficult materials problem. The sensor has to be soft enough to conform to curved objects, sensitive enough to detect small deformations, fast enough to follow texture during sliding, and robust enough to keep working after repeated contact. Many flexible tactile sensors improve one of those traits by sacrificing another, a challenge also reflected in work on flexible sensors that combine touch and body-position sensing. Fine microstructures can raise sensitivity but add fragility, while complex stacks can make dense arrays harder to manufacture.
One common way to balance those demands is capacitive sensing, a method that measures touch by watching how an electrical signal changes when a material bends, compresses, or moves. Many capacitive sensors use two electrodes facing each other, with a soft layer between them. That layout works well for pressure, but it is less suited to bending because most of the signal comes from squeezing the layer thinner.
Another layout places the electrodes side by side on a flexible surface, which is easier to manufacture as an array. The drawback is that this side-by-side design usually produces a weaker signal when it bends. A new paper in Advanced Materials (“Fish‐Scale‐Inspired Giant Piezocapacitive Sensors for Human‐Level Touch Perception”) tackles that weak-signal problem directly by keeping the manufacturable layout and changing what sits above it.
Bio-inspired design and mechanism of the giant piezocapacitive sensor (GPCS) for robotic touch perception. (a) Design concept inspired by the fish-scale/skin composite structure: rigid PZT flakes are integrated onto a soft silicone substrate to mimic the ‘rigid scale–soft dermis’ architecture. (b) (i) Finite element analysis and optical micrographs illustrating the strain-isolation mechanism, where deformation is predominantly accommodated by gap expansion while the rigid scales remain strain-free. (ii) The scale-like electric-field gating film (SEGF) under complex deformations that combine stretching, twisting and knife puncturing. (c) (i) Mechanism of the electric-field gating effect, showing the modulation of fringing electric fields. (ii) Simulation results of the normalized capacitance (C/C0) with different air-gap widths. (d) Schematic overview of the GPCS-integrated robotic system for multi-modal tasks, including texture recognition and fruit ripeness assessment. (Image: Reproduced with permission from Wiley-VCH Verlag) (click on image to enlarge)
The researchers call the result a giant piezocapacitive sensor. Piezocapacitive means that mechanical deformation changes capacitance, the electrical property the sensor measures. Giant refers to the unusually large capacitance change produced by the structured film. The film sits above the electrodes and changes how the electrical field moves during bending.
The film resembles overlapping fish scales, with rigid high-permittivity pieces separated by tiny air gaps. High permittivity means the pieces help concentrate the electrical field. As the gaps open and close, they regulate the field and turn small bends into large capacitance changes. In tests, the sensor resolved bending changes of 0.005° across ±90°, responded in 0.6 ms, and maintained performance after 100 000 bending cycles.
The design uses rigid lead zirconate titanate, or PZT, scales on a soft silicone substrate. The PZT pieces provide the electrical effect, while the silicone gives the film the compliance needed for bending and contact with curved surfaces. The important action happens between the rigid pieces. When the device bends, neighboring scales do not need to stretch much. Instead, the air gaps between them change shape.
Those gaps turn a structural discontinuity into the sensing mechanism. When the sensor bends inward, the gaps narrow, less electrical field leaks through them, and capacitance increases. When the sensor bends outward, the gaps widen, more field leaks out, and capacitance decreases. The device can therefore read both bending direction and bending magnitude through controlled field leakage rather than through compression alone.
The comparison with simpler devices shows why the gating film matters. A bare side-by-side electrode produced only a small capacitance change during bending. A version covered with a soft high-permittivity composite raised the starting capacitance but still did little to improve bending response. Adding the scale-like gating film increased the response by roughly 177 times compared with the bare electrode.
That structure also gives the sensor a route to durability. Because the rigid scales avoid most of the stretching, repeated deformation concentrates in the narrow gaps where opening and closing can occur again and again. The paper reports stable output through prolonged bending tests. It also shows that the sensor continued to function after holes were punched through part of the sensing region, a useful sign for devices that may face wear or local damage.
Fast response turns the bending signal into a richer form of touch. Texture does not appear to a sliding fingertip as a fixed pattern. It becomes a sequence of small, rapid deformations. The sensor followed vibrations up to 1000 Hz and distinguished small frequency differences around 500 Hz, suggesting that it can track the dynamic signals that arise when a soft finger moves across a surface.
The researchers tested that idea by mounting a smaller sensor on a soft robotic fingertip and sliding it over raised test patterns. As the fingertip crossed each feature, the surface bent the sensor and produced a capacitance trace. The device resolved micrometer-scale relief, including printed toner lines only 1.8 µm thick on paper, a height the paper describes as generally imperceptible to human touch.
Fabric sensing provided a more varied test of texture perception. The fingertip slid across sixteen textiles with different woven and knitted structures. Each fabric generated a repeatable signal shaped by its surface height, stiffness, and repeating pattern. By extracting the signal amplitude and dominant spacing in the pattern, the researchers separated all sixteen samples, showing that the sensor captured texture information in a compact electrical form.
This result fits a broader push to give robots tactile information that vision cannot supply. Nanowerk has covered related work on robots learning to feel what vision misses, including efforts to infer material properties through physical interaction. The new sensor follows that direction, but its central contribution lies in the material structure that amplifies small contact-induced deformations.
The fruit-sorting experiment shows how that capability could matter during grasping. The team attached four sensors to a fin-ray robotic gripper and monitored how the fingers bent while holding kiwis. Softer fruit caused larger gripper deformation and stronger capacitive responses. After training on repeated grasps of unripe, intermediate, and ripe fruit, the system classified ripeness with 92 % overall accuracy.
That demonstration should not be read as a complete agricultural sorting system. It used defined ripeness classes, controlled grasping, and a trained classifier. Its value lies in the integration of sensing into the grasp itself. The gripper did not need a separate firmness test. It extracted useful mechanical information while performing the normal handling action, then used the result to sort or hand over the fruit.
The same practical need appears in other soft robotic handling systems, including Nanowerk coverage of soft robot grippers that assess fruit ripeness by touch. Gentle manipulation requires more than compliant fingers. The robot also needs feedback that tells it how the object changes under contact, especially when visual appearance fails to reveal firmness, damage, or ripeness.
The paper also identifies limits that still matter for deployment. Capacitive sensors can respond to humidity and nearby objects, so shielding will need to reduce environmental interference and proximity effects. Manufacturing precision also remains important. More controlled patterning of the scale geometry and gap widths could improve consistency between devices and support larger sensor arrays.
The result shows how a common weakness in flexible sensors can become useful. Small gaps and cracks usually signal damage or fabrication limits. Here, they become the part of the device that amplifies touch. If the approach can be made consistent across large arrays, it could help soft grippers and tactile skins detect bending, texture, and firmness through repeated contact with real objects.
For authors and communications departmentsclick to open
Lay summary
Prefilled posts
ORCID information
Hongbo Wang (University of Science and Technology of China)
, 0000-0002-6546-1241 corresponding author
Nanowerk Newsletter
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
