Material identification sensor
By using a multimodal three-dimensional force and material recognition sensor with a biomimetic papillary and microneedle array layered structure, the difficulties of traditional sensors in multi-parameter detection and signal decoupling are solved, and high-precision multimodal material recognition and stable sensing are achieved.
Patent Information
- Application Number
- CN202511524102.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-23
- Publication Date
- 2025-12-23
AI Technical Summary
Traditional flexible tactile sensors cannot simultaneously detect multiple parameters such as three-dimensional force, temperature, material type, roughness, or slip state. They also suffer from problems such as difficulty in signal decoupling, bulky structure, complex wiring, and susceptibility to noise and crosstalk.
A multimodal three-dimensional force and material recognition sensor employing a biomimetic papillary and microneedle array layered structure separates force sensing from surface characteristic sensing through physical layering design. Combining a three-dimensional force sensing layer, an elastic filling layer, and a material sensing layer, it utilizes an embedded microneedle array, papillary array, and signal acquisition components to achieve simple decoupling and high-precision recognition of multimodal signals.
It achieves a 30% or more improvement in the accuracy of three-dimensional force measurement, a material recognition accuracy of 98.3%, a compact structure that is easy to integrate, and improved sensing sensitivity and grasping stability, which are significantly better than single-modal recognition methods.
Smart Images

Figure CN121185476A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present disclosure relates to the technical field of material recognition sensor, in particular, to a multi-modal three-dimensional force and material recognition sensor based on bionic papilla and microneedle array layered structure. BACKGROUND
[0002] At present, the traditional mainstream flexible tactile sensor mostly adopts piezoresistive or capacitive principle. The traditional sensor is often limited to single or a few modalities, and can only detect single parameters such as force or temperature. For example, the force sensor based on strain gauge or capacitive principle can detect the size of contact force, and the temperature sensor based on thermistor can sense temperature change. These sensors cannot simultaneously detect multi-parameter information such as three-dimensional force, temperature, material type, roughness or slip state, and it is difficult to achieve comprehensive recognition of materials.
[0003] Although the traditional flexible tactile sensor based on triboelectric effect principle can realize multi-dimensional force perception, under complex stress, the mixed single-layer sensor is still a great challenge in signal decoupling. Some traditional flexible tactile sensors realize multi-modal by stacking different functional sensing layers, but this often leads to thick structure, complex wiring, and introduces interlayer interference, resulting in low repeatability and reliability, and also increases the volume and complexity of the system due to the need for external signal processing circuit. Long signal path is easy to cause noise and crosstalk, which further reduces the recognition accuracy.
[0004] Although related research attempts to increase the number of sensors or optimize the layout to improve the perception ability, these methods still do not solve the problem of multi-modal high-precision material recognition. Therefore, people expect to obtain a material recognition sensor which can simply decouple multi-modal signals and prevent crosstalk.
[0005] The above information disclosed in the background section is only intended to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY
[0006] In view of the above, the present disclosure provides a material recognition sensor to eliminate the problems in the prior art, comprising: a support frame having the shape of one or more bones of a human body or an animal and a hollow cavity for arranging a signal line; a three-dimensional force sensing layer, the back of which is tightly wrapped on the support frame to be supported and thus has a shape profile shaped with the support frame, and the surface of which is arranged with an embedded micro-needle array for directly decomposing a three-dimensional force vector at the physical level; an elastic filling layer, the back of which is wrapped on the three-dimensional force sensing layer to be supported and thus has a shape profile shaped with the three-dimensional force sensing layer, so as to buffer and conduct the externally applied force to the embedded micro-needle array; and a material sensing layer, the back of which is wrapped on the elastic filling layer to be supported and thus has a shape profile shaped with the elastic filling layer, and the surface of which is periodically arranged with a papilla array.
[0007] According to the material recognition sensor of the present disclosure, the three-dimensional force sensing layer comprises an array of a plurality of sensing units, adjacent sensing units have a predetermined spacing, and each sensing unit contains a central vertical micro-needle and four micro-needles tilted 45° in a pyramid shape.
[0008] According to the material recognition sensor of the present disclosure, the micro-needles are made of carbon nanotube doped piezoresistive silicone rubber, and when receiving external force conducted from the outside, the deformation of the central micro-needle causes the resistance to represent the normal force of the external force, and the resistance change of the four tilted micro-needles is combined to represent the tangential force in two directions of the external force.
[0009] According to the material recognition sensor of the present disclosure, the papillae in the periodically arranged papilla array on the surface of the material sensing layer have a predetermined height, and each papilla is integrated with a micro-thermistor and a triboelectric nanogenerator unit.
[0010] According to the material recognition sensor of the present disclosure, the micro-thermistor is a thin film thermistor for obtaining the temperature curve slope of the thin film thermistor in the heat flow transient response at the moment when the external object contacts the papilla surface film.
[0011] According to the material recognition sensor of the present disclosure, the triboelectric nanogenerator unit is composed of a PTFE film covered on the surface of the papilla and a copper electrode below it, so as to generate a triboelectric signal corresponding to the contact material when the PTFE film is touched.
[0012] According to the material recognition sensor of the present disclosure, the signal acquisition component is used to acquire the electrical signals generated by the embedded microneedle array in the three-dimensional force perception layer and the micro-thermistor and triboelectric nanogenerator unit in the material perception layer when contacted by an external object, and transmit the acquired signals to the material recognition component, so that the material recognition component identifies the contacted material based on the pre-trained recognition model.
[0013] According to the material recognition sensor of the present disclosure, the recognition model is trained based on the contact feature data samples of predetermined material samples, and the similarity probability percentage of the feature data samples is set to identify specific materials.
[0014] According to the material recognition sensor of the present disclosure, the recognition model is a pre-trained lightweight graph neural network model.
[0015] According to the material recognition sensor of the present disclosure, the elastic filling layer is polydimethylsiloxane with a predetermined hardness, which conducts and buffers forces between the three-dimensional force perception layer and the material perception layer, and isolates external high-frequency vibrations to stabilize the perception of the material perception layer.
[0016] The material recognition sensor of the present disclosure physically separates macroscopic force perception and microscopic surface property perception by physical layering design, fundamentally solving the signal crosstalk problem, and improving the three-dimensional force measurement accuracy by more than 30% compared with mixed single-layer sensors. Further, high integration and bionic design: three modalities of three-dimensional force, thermal conductivity, and triboelectricity are realized in a single skeletal form (such as a fingertip), the structure is compact, and it is easier to integrate on a dexterous hand. The bionic papilla design enhances the sensitivity of perception and the stability of grasping. Moreover, the present disclosure combines the features of two physical dimensions of heat and triboelectricity, with strong information complementarity, and the recognition accuracy for easily confused materials (such as different types of plastics) can reach 98.3%, which is significantly better than single-modality recognition methods, so the material recognition accuracy is high.
[0017] It should be understood that the above general description and the following detailed description are only exemplary and do not limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS
[0018] The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings. The drawings described below are only some embodiments of the present disclosure, and other drawings can be obtained from these drawings by those of ordinary skill in the art without creative labor.
[0019] Figure 1 is a structural schematic diagram of a material recognition sensor according to an exemplary embodiment.
[0020] Figure 2 is a flowchart illustrating a process of identifying a material using a material identification sensor according to an exemplary embodiment. DETAILED DESCRIPTION
[0021] Example embodiments now will be described more fully hereinafter with reference to the accompanying drawings; however, these embodiments should not be construed as limiting all example embodiments. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example embodiments to those skilled in the art. Like reference numerals refer to like elements throughout.
[0022] Moreover, described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the aspects of the disclosure can be practiced without one or more of the specific details, or with other methods, components, materials, and so forth. In other instances, well-known structures, devices, systems, and so forth have not been shown or described in order to avoid obscuring aspects of the disclosure.
[0023] The block diagrams in the drawings show functions and functionality as they can be implemented in software / firmware. In other words, the block diagrams show a function specified in software or firmware. Should such be implemented in software, the functions will be then implemented once when the software is run. In the block diagrams, each block can represent both hardware (e.g., an application specific integrated circuit) and / or software (e.g., the software being run on a general purpose processor or a microcontroller); the software implementation can be stored in any type of non-transitory storage medium (e.g., random access memory, flash memory, read only memory, etc.).
[0024] It should be understood that although the terms first, second, third, etc. can be used herein to describe various elements, components, regions, layers and / or sections, these elements, components, regions, layers and / or sections should not be limited by these terms. These terms are only used to distinguish one element, component, region, layer or section from another element, component, region, layer or section. Thus, a first element discussed below could be termed a second element without departing from the teachings of the present disclosure. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.
[0025] Those skilled in the art can understand that the modules or processes in the drawings are only schematic diagrams of example embodiments, and are not necessarily essential for implementing the present disclosure, and therefore cannot be used to limit the protection scope of the present disclosure.
[0026] Figure 1 is a structural diagram of a material identification sensor according to an exemplary embodiment. As shown, a material identification sensor 100 according to the present disclosure includes a support frame 110, a three-dimensional force sensing layer 120, an elastic filling layer 130, and a material sensing layer 140. Figure 1 is a structural diagram of a material identification sensor according to an exemplary embodiment. As shown, a material identification sensor 100 according to the present disclosure includes a support frame 110, a three-dimensional force sensing layer 120, an elastic filling layer 130, and a material sensing layer 140.
[0027] As Figure 1 shown, the support frame 110 has one or more bone shapes of the human or animal body and a hollow cavity for arranging signal lines. For the convenience of description, the present disclosure takes the shape of a conventional finger phalanx as an example. It is obvious that the shape of a toe or other parts of the human or animal body can be used as a carrier for other components of the present application. As Figure 1 shown, the support frame 110 adopts a finger shape, which is hollow inside. The finger-shaped support frame 110 is a rigid skeleton, which can be made of 3D printed polylactic acid (PLA) material, and the hollow cavity provides a channel for wiring.
[0028] The back of the three-dimensional force sensing layer 120 is tightly wrapped on the support frame to be supported and thus has a shape profile imitating the support frame, and its surface is arranged with an embedded microneedle array for directly decomposing the three-dimensional force vector in the physical layer. The embedded microneedle array has a predetermined spacing, for example, 1.5 mm, between adjacent sensing units 121. Each sensing unit contains a central vertical microneedle and four pyramidal-shaped inclined microneedles at 45°. The microneedles are made of carbon nanotube-doped piezoresistive silicone rubber, and when receiving external force conducted from the outside, the deformation of the central microneedle causes the resistance to represent the normal force of the external force, and the combined resistance changes of the four inclined microneedles decompose the two directions of the tangential force of the external force. Three-dimensional force sensing is a precision measurement technology that can detect force information (Fx, Fy, Fz) in three-dimensional space (X, Y, Z axis) at the same time, which is widely used in robots, industrial automation, biomechanics and other fields. Its core principle is to convert mechanical signals into electrical signals through strain gauge resistance change or elastomer deformation, combined with decoupling algorithm to achieve high-precision measurement. Therefore, the specific content of three-dimensional force sensing in the present disclosure will not be described here, and three-dimensional force sensing or detection itself is not the technical content concerned in the present disclosure. Unlike the prior art, the sensing unit of the present disclosure can more accurately obtain the component force of the contact surface normal by adding a central microneedle.
[0029] Further, the microneedles of the sensing unit 121 are made of carbon nanotube-doped piezoresistive silicone rubber, and when receiving external force conducted from the outside, the deformation of the central microneedle causes the resistance to represent the normal force of the external force, and the combined resistance changes of the four inclined microneedles decompose the two directions of the tangential force of the external force. Specifically, when subjected to external force, the resistance change of the central microneedle mainly reflects the normal force (Fz); the combined resistance change of the four inclined microneedles can accurately decompose the two directions of the tangential force (Fx, Fy). This structure realizes the physical decomposition of three-dimensional force. It should be noted that Figure 1The adjacent sensing units 121 of the embedded microneedle array in the three-dimensional force sensing layer 120 are displayed in an omitted manner, where the dashed lines represent the continuity of the arrangement of the adjacent sensing units 121.
[0030] As shown in Figure 1 The back of the elastic filling layer 130 is wrapped around the three-dimensional force sensing layer 120 to support it and thus has a shape profile that is shaped like the three-dimensional force sensing layer 120, thereby buffering and conducting externally applied forces to the embedded microneedle array. The elastic filling layer 130 is a polydimethylsiloxane of a predetermined hardness, which conducts and buffers forces in the three-dimensional force sensing layer and the material sensing layer, and isolates external high-frequency vibrations to stabilize the sensing of the material sensing layer.
[0031] Finally, the back of the material sensing layer 140 is wrapped around the elastic filling layer 130 to support it and thus has a shape profile that is shaped like the elastic filling layer 130, and its surface is periodically arranged with a papillary array 141. The papillae in the papillary array 141 on the surface of the material sensing layer 140 have a predetermined height, for example, 50 um. Each papilla is integrated with a micro-thermistor and a triboelectric nanogenerator unit. It should be noted that Figure 1 The papillary array 141 in the material sensing layer 140 is displayed in an omitted manner, where the dashed lines represent the continuity of the arrangement of the adjacent papillae.
[0032] The micro-thermistor is a thin-film thermistor that is used to obtain the temperature curve slope of the thin-film thermistor in response to the heat flow transient response in the instant when the external object contacts the papillary surface film. Moreover, the triboelectric nanogenerator unit is composed of a PTFE film covering the papillary surface and a copper electrode below it, so that when the PTFE film is touched, a triboelectric signal corresponding to the contact material is generated. Specifically, the micro-thermistor is used to measure the heat flow transient response in the instant of contact. The thermal conductivity of different materials (such as metal, wood, plastic) differs greatly, which will cause the temperature curve slope of the thermistor to be significantly different. Therefore, by the temperature curve slope value generated by the heat conduction when different materials contact the thin-film thermistor, the heat conduction characteristic value of different materials for the thin-film thermistor is formed, which can be used for subsequent model training of the characteristic values of the material recognition model. The papillary surface of the triboelectric nanogenerator unit is covered with a layer of polytetrafluoroethylene (PTFE) film, which forms a triboelectric unit with the copper electrode below it. When in contact with different object surfaces, due to the difference in material electron affinity, a unique triboelectric signal (voltage waveform, amplitude) is generated, and these data can be collected in advance for different materials as characteristic values of specific materials for subsequent model training of the characteristic values of the material recognition model.
[0033] As shown in Figure 1As shown, the three-dimensional force sensing layer 120 and the material sensing layer 140 generate three-dimensional force, thermal conductivity data and triboelectric signals when the sensor 100 of the present disclosure is contacted by different materials, which are introduced into the signal acquisition assembly 150 via the signal lines arranged in the hollow cavity, and then the collected signals are transmitted to the material identification assembly 160. The material identification assembly 160 identifies the contacted material based on a pre-trained identification model.
[0034] It should be noted that the identification model is trained based on the contact feature data samples of predetermined material samples, thereby forming the comprehensive feature values of the feature data that can be collected by the material identification sensor 100 of the present disclosure. To this end, in order to identify the material, the present disclosure sets the similar probability percentage of the feature data samples to identify the specific material. That is, when the comprehensive feature values of the collected actual various data reach the similar probability percentage of the predetermined feature data samples, it can be determined that the material corresponding to the predetermined feature data samples. The probability percentage can be between 70-100%, and the similar probability corresponding to different materials can be different. The identification model is a pre-trained lightweight graph neural network model.
[0035] In addition, in order to increase the friendliness of the material identification sensor 100 to the skin, a flexible packaging layer such as skin-friendly silica gel can be wrapped outside the material sensing layer 140. For a finger-shaped sensor, the surface of the flexible packaging layer has a simulated fingerprint texture, which increases the gripping friction and protects the internal structure.
[0036] Figure 2 is a flowchart showing the identification of materials using the material identification sensor of the present disclosure according to an exemplary embodiment. As shown in FIG. 8, the material identification sensor 100 of the present disclosure is used to identify the material. First, the material identification sensor 100 is contacted by the material to be identified, and the three-dimensional force sensing layer 120 and the material sensing layer 140 generate three-dimensional force, thermal conductivity data and triboelectric signals. The signals are introduced into the signal acquisition assembly 150 via the signal lines arranged in the hollow cavity, and then the collected signals are transmitted to the material identification assembly 160. The material identification assembly 160 identifies the contacted material based on a pre-trained identification model. Figure 2As shown, in using the material identification sensor 100 of the present disclosure, first at step S210, synchronous signal acquisition is performed as in the conventional multi-modal signal acquisition. The signal acquisition component 150 synchronously acquires the resistance signal of the piezoresistive microneedle array from the three-dimensional force perception layer 120 and the temperature curve of the micro-thermistor and the voltage signal of the triboelectric nanogenerator from the material perception layer 140 at a sampling rate of 1-1.5 kHz or other frequencies. Subsequently at step S220, the material identification component 160 performs three-dimensional force calculation, that is, the piezoresistive signal is calculated directly through the pre-calibrated mechanical model, and the [Fx, Fy, Fz] vector is output in real time. Fz = f_z(AR_center), [Fx, Fy]=f_xy(AR_tilt1, …, AR_tilt4). Subsequently, at step S230, the material identification component 160 performs material feature extraction, including: extracting thermal features, extracting the initial slope and steady-state temperature difference of the temperature curve within 50 ms after contact; extracting triboelectric features, extracting the peak value, half-peak width and decay time constant of the voltage signal during contact / separation. After extracting the material feature values, at step S240, the material identification component 160 performs multi-modal fusion and material identification. Specifically, the extracted thermal features and triboelectric features are spliced into a feature vector, that is, the aforementioned comprehensive feature value, and the feature vector is input into a pre-trained lightweight graph neural network (GNN). The GNN can effectively learn the correlation (that is, the similarity probability) between different physical features, and finally output the classification probability of 12 common materials (such as wood, glass, ABS plastic, nylon, fabric, etc.). The present disclosure can use existing deep learning methods to assist in realizing the hybrid tactile sensor identification of the present disclosure.
[0037] By using the material identification sensor of the present disclosure, the macroscopic force perception and the microscopic surface property perception are physically separated by physical layering design, which fundamentally solves the signal crosstalk problem, and the three-dimensional force measurement accuracy is improved by more than 30% compared with the hybrid single-layer sensor. Further, high integration and bionic design: three-dimensional force, thermal conductivity, and triboelectricity modal perception are realized in a single skeletal form (such as a fingertip), the structure is compact, and it is easier to integrate on a dexterous hand. The bionic papilla design enhances the sensitivity of perception and the stability of grasping. Moreover, the present disclosure combines the features of thermal and triboelectric two physical dimensions, the information complementarity is strong, and for materials that are easy to confuse (such as different types of plastics), the recognition accuracy can reach 98.3%, which is significantly better than single-modal recognition methods, so the material identification accuracy is high.
[0038] Those skilled in the art can easily understand, through the above description of the embodiments, that the example embodiments described herein can be implemented by software or by software in combination with necessary hardware. Therefore, the operation process of the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product. The software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.) or on a network, and includes a number of instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to perform the method according to the embodiments of the present disclosure.
[0039] The example embodiments of the present disclosure are specifically shown and described above. It should be understood that the present disclosure is not limited to the detailed structure, arrangement or implementation method described herein; on the contrary, the present disclosure is intended to cover various modifications and equivalent arrangements within the spirit and scope of the appended claims.
Claims
1. A material identification sensor, comprising: A support frame having an external profile resembling one or more skeletons of a human or animal and a hollow cavity for arranging signal lines. The three-dimensional force sensing layer is tightly wrapped around the support frame on its back side to support it and thus has a shape contour that mimics the support frame. The surface of the layer is arranged with an embedded microneedle array for directly decomposing the three-dimensional force vector at the physical level. An elastic filler layer, whose back side wraps around the three-dimensional force-sensing layer to support it and thus has a shape contour that mimics the three-dimensional force-sensing layer, thereby buffering externally applied forces and transmitting them to the embedded microneedle array. as well as The material-sensing layer is supported by an elastic filler layer on its back and thus has a shape contour that mimics the elastic filler layer, and its surface is periodically arranged with an array of papillae.
2. The material identification sensor according to claim 1, wherein the three-dimensional force sensing layer comprises an array of multiple sensing units, with a predetermined spacing between adjacent sensing units, and each sensing unit comprises a central vertical microneedle and four pyramid-shaped microneedles tilted at 45°.
3. The material identification sensor according to claim 2, wherein the microneedle is made of piezoresistive silicone rubber doped with carbon nanotubes, and when receiving an external force transmitted from the outside, the resistance caused by the deformation of the central microneedle represents the normal force of the external force, and the combined decomposition of the resistance changes of the four tilted microneedles represents the tangential force in two directions of the external force.
4. The material identification sensor according to claim 1, wherein the papillae in the periodically arranged papillary array on the surface of the material sensing layer have a predetermined height, and each papilla integrates a micro thermistor and a triboelectric nanogenerator.
5. The material identification sensor according to claim 4, wherein the miniature thermistor is a thin-film thermistor, used to obtain the temperature curve slope of the thin-film thermistor in response to the transient thermal flow when an external object contacts the thin film on the surface of the papilla.
6. The material identification sensor according to claim 4, wherein the triboelectric nanogenerator unit is composed of a PTFE film covering the papillary surface and a copper electrode below it, so as to generate a triboelectric signal corresponding to the contact material when the PTFE film is touched.
7. The material identification sensor according to claim 4 further includes a signal acquisition component and a material identification component. The signal acquisition component is used to acquire the electrical signals generated by the embedded microneedle array in the three-dimensional force sensing layer and the micro thermistors and triboelectric nanogenerators in the material sensing layer when they are contacted by an external object, and transmit the acquired signals to the material identification component so that the material identification component can identify the contacted material based on a pre-trained identification model.
8. The material identification sensor according to claim 7, wherein the identification model is trained based on contact feature data samples of a predetermined material sample, and the similarity probability percentage of the feature data samples is set to identify a specific material.
9. The material identification sensor according to claim 7, wherein the identification model is a pre-trained lightweight graph neural network model.
10. The material identification sensor according to claim 1, wherein the elastic filling layer is a polydimethylsiloxane of predetermined hardness, which transmits and buffers force in the three-dimensional force sensing layer and the material sensing layer, and isolates external high-frequency vibrations to stabilize the sensing of the material sensing layer.
