Surface texture recognition system based on fabric touch sensor and construction method thereof

Abstract

The invention discloses a surface texture recognition system based on a fabric tactile sensor and a construction method of the surface texture recognition system, and relates to the field of recognition precision regulation and control. The prepared interdigital electrode and the sensing electrode are laminated and assembled into the fabric tactile sensor with the bionic microstructure; the constructed microstructure sensing electrode can realize efficient sensing of static force and mechanical vibration at the same time through microstructure multi-dimensional deformation so as to realize efficient identification of surface fine texture textures; the constructed sensing electrode microstructure is adjustable and controllable, so that the performance is adjustable and controllable, the problems of device flexibility, air and moisture permeability, wearing comfort, recognition accuracy reduction and the like caused by a commonly used template microstructure preparation method are solved, and the problems of difficult signal collection and signal crosstalk caused by a multi-mechanism mixed mode are solved.

Description

Technical Field

[0001] The present invention belongs to the field of recognition accuracy regulation. Specifically, it particularly relates to a surface texture recognition system based on a fabric tactile sensor and a construction method thereof. Background Art

[0002] In recent years, with the rapid development of the fields of humanoid robots and human-computer interaction, higher requirements have been put forward for high-performance tactile perception devices and systems; such sensors can perceive different tactile stimuli, such as surface texture, pressure, shape, etc. by simulating the tactile response of the skin; especially in the field of surface texture recognition, the application of tactile sensors has great potential; traditional surface texture recognition methods mostly rely on vision or high-precision mechanical detection, and it is difficult to simulate real tactile experiences; tactile sensors can efficiently capture fine surface texture information by integrating flexible materials and micro sensing units, and have good adaptability and sensitivity, so they have broad application prospects in fields such as intelligent wearable devices, robot tactile feedback, and virtual reality; in order to obtain more sensitive tactile feedback information, existing technologies generally adopt a multi-modal sensor integration strategy to separately achieve the perception of static force and mechanical vibration (Nat. Electron. 2021, 4, 429; Matter 2022, 5, 1481; Adv. Sci. 2023, 10, 2303949), for example, by using piezoresistive sensors to mimic the slow-adapting bodies of the human body to achieve the perception of static force; by using piezoelectric, capacitive or triboelectric sensors to mimic the fast-adapting bodies of the human body to achieve the perception of high-frequency mechanical vibration;

[0003] However, the multi-modal integration strategy commonly adopted by existing surface texture recognition systems has the following problems: (1) Multi-modal devices require multiple signal collection circuits, increasing the inconvenience during use; (2) The device is composed of multiple layers of electrodes, and shear failure is likely to occur between the electrodes, increasing the failure probability of the device during use; (3) Crosstalk between multiple signals will seriously affect the final recognition accuracy; (4) The mapping relationship between the sensing signal and the surface texture is not yet clear; Therefore, how to construct a piezoresistive tactile sensor based on a single loop to simultaneously achieve efficient perception of static force and dynamic force, and then achieve accurate recognition of surface texture structures, has always been a key challenge. Summary of the Invention

[0004] Aiming at the problems in the related art, the present invention proposes a surface texture recognition system based on a fabric tactile sensor and a construction method thereof to overcome the above-mentioned technical problems existing in the existing related technologies.

[0005] To solve the above-mentioned technical problems, the present invention is realized through the following technical solutions:

[0006] The present invention relates to a method for constructing a surface texture recognition system based on a fabric tactile sensor, comprising the following steps:

[0007] S1. Prepare a fabric sensing electrode with a honeycomb microstructure; specifically, it includes the following steps:

[0008] S11. Deposit the MXene aqueous dispersion and PEDOT:PSS aqueous solution on the surface of cotton yarn through an alternating deposition technique to obtain MXene / PEDOT:PSS composite modified yarn;

[0009] S12. Weave the MXene / PEDOT:PSS composite modified yarn into a fabric sensing electrode with a honeycomb microstructure through a multi-shuttle layered weft insertion weaving process;

[0010] S2. Prepare a fabric interdigital electrode; specifically, it includes the following steps:

[0011] S21. Print conductive silver paste on the surface of a non-woven fabric substrate through a transfer technique to form a fabric interdigital electrode;

[0012] S3. Construct a surface texture recognition system; specifically, it includes the following steps:

[0013] S31. Assemble the fabric sensing electrode with a honeycomb microstructure and the fabric interdigital electrode into a fabric tactile sensor by ultrasonic bonding;

[0014] S32. Construct a surface texture recognition system according to the fabric tactile sensor, microcontroller, transimpedance amplifier, and inverter;

[0015] Weave a microstructure fabric sensing electrode by modifying yarn with a conductive material, and prepare a fabric interdigital electrode through a transfer technique. Stack and assemble the prepared interdigital electrode and sensing electrode into a fabric tactile sensor with a bionic microstructure. Since the fabric structure can be designed during the weaving process, the constructed microstructure sensing electrode can undergo multi-dimensional deformation of the microstructure (vertical compression deformation and shear deformation), and thus can simultaneously achieve efficient perception of static force and mechanical vibration to realize efficient recognition of surface fine texture patterns.

[0016] Preferably, S11 includes the following steps:

[0017] S111. Immerse the cotton yarn in a quaternary ammonium salt cationic modifier at a bath ratio of 1:5 to 1:30 at 20 to 80 °C for 1 to 30 minutes to make the yarn positively charged; then dry it at 60 °C to obtain cation-modified yarn;

[0018] S112. Immerse the cation-modified yarn in the MXene aqueous dispersion at a bath ratio of 1:5 to 1:30 at 20 to 80 °C for 1 to 30 minutes to obtain MXene-modified yarn;

[0019] S113. Repeat the process of S111 for the MXene-modified yarn to obtain a positively charged MXene-modified yarn; then immerse the positively charged MXene-modified yarn in an aqueous PEDOT:PSS solution at a bath ratio of 1:5 to 1:30 at 20 to 80 °C for 1 to 30 minutes for PEDOT:PSS deposition, and then dry at 60 °C.

[0020] S114. Repeat S111, S112, and S113 three times to obtain a MXene / PEDOT:PSS composite-modified yarn.

[0021] Preferably, the mass concentration of the quaternary ammonium salt cationic modifier in S111 is 10 mg / mL.

[0022] Preferably, the mass concentration of the MXene aqueous dispersion in S112 is 10 mg / mL.

[0023] Preferably, the mass concentration of the PEDOT:PSS aqueous solution in S113 is 10 mg / mL.

[0024] Preferably, in the multi-shuttle layer-by-layer weft insertion weaving process in S12, the number of weft shuttles is 2 to 3, the number of weft layers is 2 to 6, the weft insertion angle is 15 to 90°, and the weaving density of the warp and weft yarns is 400 per 10 cm.

[0025] Preferably, the solid content of the conductive silver paste in S21 is 70 wt%, and the transfer loading amount of the conductive silver paste is 5 wt%.

[0026] Preferably, in the ultrasonic bonding method in S31, the ultrasonic power is 2000 W, the ultrasonic frequency is 20 kHz, and the time is 2 minutes.

[0027] Preferably, S32 includes the following steps:

[0028] S321. Power the fabric tactile sensor, convert the current flowing through the fabric tactile sensor into a voltage through the transimpedance amplifier, and then convert the voltage into a positive voltage through an inverter and input it into the microcontroller.

[0029] Preferably, the yarn modification active materials used for modifying the yarn with conductive active materials in S1 include carbon black, acetylene black, graphene, carbon nanotubes, silver nanoparticles, MXene, conductive polymers (such as polyaniline, polypyrrole, polyvinyl thiophene, etc.), etc., and the transfer active materials in the transfer printing technology include conductive silver paste, MXene, graphene, carbon nanotubes, carbon black and other conductive pastes; the fabric substrate includes woven fabrics, knitted fabrics, non-woven fabrics, papers, polymer films, etc.; the yarn modification technologies in S1 include but are not limited to sizing, dip coating, infrared sizing, surface polymerization, etc.

[0030] A construction system of a surface texture recognition system based on a fabric tactile sensor, comprising a fabric sensing electrode preparation module, a fabric interdigital electrode preparation module, and a surface texture recognition system construction module;

[0031] The fabric sensing electrode preparation module is used to prepare a fabric sensing electrode with a honeycomb microstructure;

[0032] The fabric interdigital electrode preparation module is used to print conductive silver paste on the surface of a non-woven fabric substrate through a transfer printing technology to form a fabric interdigital electrode;

[0033] The surface texture recognition system construction module is used to constitute a surface texture recognition system according to the fabric tactile sensor, a microcontroller, a transimpedance amplifier, and an inverter.

[0034] The present invention has the following beneficial effects:

[0035] 1. The present invention proposes a construction method for an efficient, accurate, fabric-based surface texture recognition system; a microstructure sensing electrode is prepared by using a weaving process and a fabric tactile sensor is constructed, and the multi-dimensional deformation effects generated by the bionic microstructure on the surface of the sensing electrode under the action of an external force, including vertical compression deformation and horizontal shear deformation, are utilized to simultaneously realize the efficient perception of static force and mechanical vibration, and finally realize the accurate perception and recognition of surface fine texture.

[0036] 2. In the present invention, various microstructure sensing electrodes can be prepared by a weaving process, and the microstructure of the constructed sensing electrode is adjustable and controllable, so as to realize the adjustable and controllable performance, and avoid the problems such as the reduction of the flexibility, air permeability and moisture permeability, wearing comfort, and recognition accuracy of the device caused by the commonly used template microstructure preparation method.

[0037] Of course, it is not necessary for any product implementing the present invention to simultaneously achieve all the above-mentioned advantages. Description of the Drawings

[0038] To more clearly illustrate the technical solutions of the embodiments of the invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0039] Figure 1 Schematic diagram of the electrode structure parameters in Embodiments 1, 2, 3, and 4 of the present invention;

[0040] Figure 2 Microscope images of the electrodes in Embodiments 1, 2, 3, and 4 of the present invention;

[0041] Figure 3 Three-dimensional microscope image of Embodiment 1 of the present invention;

[0042] Figure 4 SEM image of Embodiment 1 of the present invention;

[0043] Figure 5 Tactile perception performance curve graphs of Embodiments 1, 2, 3, and 4 of the present invention;

[0044] Figure 6 Finite element strain model graphs of Embodiments 1, 2, 3, and 4 of the present invention;

[0045] Figure 7 Tactile sensitivity curve graphs of Embodiments 1, 2, 3, and 4 of the present invention; In the figure, S 1 , S 2 , S 3 , S 4 are the sensitivities of each interval respectively, and R 2 is the fitting linearity of each interval;

[0046] Figure 8 Tactile sensitivity curve graphs of Embodiment 1 of the present invention and Comparative Examples 1, 2, and 3; In the figure, S 1 , S 2 , S 3 , S 4 are the sensitivities of each interval respectively, and R 2 is the fitting linearity of each interval;

[0047] Figure 9 Schematic diagram of the high-frequency mechanical vibration test of the present invention;

[0048] Figure 10 Schematic diagram of the perception performance of Embodiment 1 of the present invention for high-frequency vibration. In the figure, from left to right are the response current curves of the sensor to the dynamic forces of 1Hz, 3Hz, 10Hz, 30Hz, and 100Hz respectively;

[0049] Figure 11Relationship diagram between the number of response peaks and vibration frequency in Embodiment 1 of the present invention

[0050] Figure 12 Structural schematic diagram of the surface texture recognition system of the present invention;

[0051] Figure 13 Physical diagram of the real-time surface texture recognition system of the present invention;

[0052] Figure 14 Schematic diagram of the recognition result in Embodiment 1 of the present invention at a moving speed of 200 mm / S; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0053] Figure 15 Schematic diagram of the recognition result in Embodiment 1 of the present invention at a randomly variable moving speed; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0054] Figure 16 Schematic diagram of the recognition result of Fabric No. 12 in Embodiment 1 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0055] Figure 17 Schematic diagram of the recognition result of Fabric No. 14 in Embodiment 1 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0056] Figure 18 Schematic diagram of the recognition result of Fabric No. 11 in Embodiment 2 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0057] Figure 19 Schematic diagram of the recognition result of Fabric No. 11 in Embodiment 3 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0058] Figure 20 Schematic diagram of the recognition result of Fabric No. 11 in Embodiment 4 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0059] Figure 21 Schematic diagram of the recognition result of the surface texture of a complex curved surface structure in Embodiment 1 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0060] Figure 22 Schematic diagram of the recognition result of the surface texture of a soft substrate in Embodiment 1 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0061] Figure 23 Schematic diagram of the surface texture recognition result of Comparative Example 1 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0062] Figure 24 Schematic diagram of the surface texture recognition result of Comparative Example 2 of the present invention; the abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points;

[0063] Figure 25 Schematic diagram of the surface texture recognition result of Comparative Example 3 of the present invention. The abscissa of the sample signal broken line graph and the template signal broken line graph in the figure is the number of data points.

[0064] Figure 26 Schematic diagram of the neural network structure of the present invention. Detailed implementation manners

[0065] Next, the technical solutions in the embodiments of the invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the invention. Obviously, the described embodiments are only a part of the embodiments of the invention, rather than all the embodiments. Based on the embodiments in the invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the invention.

[0066] Embodiment 1

[0067] This embodiment is a construction method of a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0068] (1) Preparation of bionic microstructured fabric tactile sensor: First, the surface of cotton yarn (metric count: 120) was deposited with transition metal carbide / nitride aqueous dispersion (i.e., MXene aqueous dispersion) and poly(3,4-ethylenedioxythiophene) / poly(styrenesulfonate) aqueous solution (i.e., PEDOT:PSS aqueous solution) by alternating deposition technology: The yarn was first impregnated with a quaternary ammonium cationic modifier of dodecyl dimethyl benzyl ammonium chloride (1227) (10 mg / mL aqueous solution, Sinopharm Group Co., Ltd.) at a bath ratio of 1:20 at 60 °C for 5 minutes to make the yarn positively charged, and the cationic modified yarn was obtained after drying at 60 °C; Subsequently, the cationic modified yarn was impregnated in the MXene aqueous dispersion (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes to obtain the MXene modified yarn; The MXene modified yarn was repeated the above cationic modification process to make it positively charged again, and then impregnated in the PEDOT:PSS aqueous solution (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes for PEDOT:PSS deposition, and then dried at 60 °C; The above steps were repeated for 3 cycles to obtain the MXene / PEDOT:PSS composite modified yarn; Subsequently, the composite modified yarn prepared above was woven into a fabric sensing electrode (MPHMF) with a honeycomb microstructure by a multi-shuttle layer-by-layer weft insertion weaving process, where the number of weft shuttles was 3, the number of weft layers was 6, the weft insertion angle was 15°, and the weaving density of warp and weft yarns was 400 per 10 cm. The structure of the constructed sensing electrode is as shown in Figure 1 a of; The conductive silver paste was printed on the surface of the non-woven fabric (pure cotton, gram weight 100 g / m 2 ) substrate by transfer printing technology, where the solid content of the conductive silver paste was 70 wt%, and the transfer loading amount of the conductive silver paste was 5 wt%, to form a fabric interdigital electrode; The above-prepared MPHMF sensing electrode and the interdigital electrode were assembled into a fabric tactile sensor by ultrasonic bonding (ultrasonic power: 2000 W, ultrasonic frequency: 20 kHz, time: 2 minutes);

[0069] (2) Construction of the surface texture recognition system: The bionic structure fabric-based tactile sensor constructed in step (1), a microcontroller, a transimpedance amplifier, and an inverter are used to form the surface texture recognition system. First, the system generates a 0.5V power supply to power the sensor. The transimpedance amplifier is connected between the negative pole of the sensor and the gnd, and its function is to convert the current flowing through the sensor into a voltage. The amplification factor of the transimpedance amplifier is 330. Since the gain of the transimpedance amplifier is negative, the output of the transimpedance amplifier is connected to an inverter to convert the voltage into a positive voltage that can be input into the microcontroller. The output of the inverter is connected to the microcontroller at the pin of its built-in ADC. In this way, the voltage measured by the microcontroller divided by 330 is the current flowing through the sensor. The constructed surface texture recognition system is used to collect data on various surface textures, establish a classification model database, and then compare the real-time signals collected by the sensor with the classification model database established by machine learning before through machine learning technology, so as to perceive and predict the surface texture. First, the microsensor continuously collects the response current data of the sensor to the surface texture at a sampling frequency of 100Hz, and then the microcontroller sends the data to the computer through the serial port. After the computer receives the data, it will perform data noise reduction processing: When the current magnitude data collected by the sensor reaches a certain threshold of 50×10 -6 data collection starts. When 512 data are collected, the data will be normalized and input into the model. The model structure consists of residual blocks (each residual block contains two convolutional layers, a batch normalization layer, and a ReLU activation function), a shrinkage layer, the main structure of RSNet, a loss function, and an optimizer.

[0070] Example 2

[0071] This example is a construction method of a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0072] (1) Preparation of bionic microstructured fabric tactile sensor: First, the surface of cotton yarn (metric count: 30) is deposited with transition metal carbide / nitride aqueous dispersion (i.e., MXene aqueous dispersion) and poly(3,4-ethylenedioxythiophene) / polystyrene sulfonate aqueous solution (i.e., PEDOT:PSS aqueous solution) by an alternating deposition technique: The yarn is first impregnated with a quaternary ammonium cationic modifier of dodecyl dimethyl benzyl ammonium chloride (1227) (10 mg / mL aqueous solution, Sinopharm Group Co., Ltd.) at a bath ratio of 1:20 at 60 °C for 5 minutes to make the yarn positively charged, and the cation-modified yarn is obtained after drying at 60 °C; Subsequently, the cation-modified yarn is impregnated in the MXene aqueous dispersion (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes to obtain the MXene-modified yarn; The MXene-modified yarn is repeated the above cation-modification process to make it positively charged again, and then impregnated in the PEDOT:PSS aqueous solution (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes for PEDOT:PSS deposition, and then dried at 60 °C. Repeat the above steps for 3 cycles to obtain the MXene / PEDOT:PSS composite-modified yarn; Subsequently, the composite-modified yarn prepared above is woven into a fabric sensing electrode (LHMF) with a honeycomb microstructure by a multi-shuttle layer-by-layer weft insertion weaving process, where the number of weft shuttles is 2, the number of weft layers is 4, the weft insertion angle is 45°, and the weaving density of warp and weft yarns is 400 per 10 cm. The structure of the constructed sensing electrode is as shown in Figure 1 b of; The conductive silver paste is printed on the surface of a non-woven fabric (pure cotton, weight per unit area 100 g / m 2 ) substrate by a transfer printing technique, where the solid content of the conductive silver paste is 70 wt%, and the transfer loading amount of the conductive silver paste is 5 wt%, to form a fabric interdigital electrode; The above-prepared MPHMF sensing electrode and the interdigital electrode are assembled into a fabric tactile sensor by ultrasonic bonding (ultrasonic power: 2000 W, ultrasonic frequency: 20 kHz, time: 2 minutes).

[0073] (2) Construction of the surface texture recognition system: The bionic structure fabric-based tactile sensor constructed in step (1), a microcontroller, a transimpedance amplifier, and an inverter are used to form the surface texture recognition system. First, the system generates a 0.5V power supply to power the sensor. The transimpedance amplifier is connected between the negative electrode of the sensor and the GND, and its function is to convert the current flowing through the sensor into a voltage. The amplification factor of the transimpedance amplifier is 330. Since the gain of the transimpedance amplifier is negative, the output of the transimpedance amplifier is connected to an inverter to convert the voltage into a positive voltage that can be input into the microcontroller. The output of the inverter is connected to the microcontroller at the pin of its built-in ADC. In this way, the voltage measured by the microcontroller divided by 330 is the current flowing through the sensor. The constructed surface texture recognition system is used to collect data on various surface textures, establish a classification model database, and then compare the real-time signals collected by the sensor with the classification model database established by machine learning before through machine learning technology, so as to perceive and predict the surface texture. First, the microsensor continuously collects the response current data of the sensor to the surface texture at a sampling frequency of 100Hz, and then the microcontroller sends the data to the computer through the serial port. After the computer receives the data, it will perform data denoising processing: when the value reaches a certain threshold of 50×10 -6 data collection starts, and when 512 data are collected, the data will be normalized and input into the model. The model specifically uses a model composed of 3 residual blocks. Each residual block consists of convolution, residual pooling, full connection, etc.

[0074] Example 3

[0075] This example is a construction method of a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0076] (1) Preparation of bionic microstructured fabric tactile sensor: First, the surface of cotton yarn (metric count: 120) is deposited with transition metal carbide / nitride aqueous dispersion (i.e., MXene aqueous dispersion) and poly(3,4-ethylenedioxythiophene) / poly(styrenesulfonate) aqueous solution (i.e., PEDOT:PSS aqueous solution) by an alternating deposition technique: The yarn is first impregnated with a quaternary ammonium cationic modifier of dodecyl dimethyl benzyl ammonium chloride (1227) (10 mg / mL aqueous solution, Sinopharm Group Co., Ltd.) at a bath ratio of 1:20 at 60 °C for 5 minutes to make the yarn positively charged, and then dried at 60 °C to obtain cation-modified yarn; Subsequently, the cation-modified yarn is impregnated in the MXene aqueous dispersion (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes to obtain MXene-modified yarn; The MXene-modified yarn is repeated the above cation-modification process to make it positively charged again, and then impregnated in the PEDOT:PSS aqueous solution (10 mg / mL) at a bath ratio of 1:20 at 60 °C for 5 minutes for PEDOT:PSS deposition, and then dried at 60 °C. Repeat the above steps for 3 cycles to obtain MXene / PEDOT:PSS composite-modified yarn; Subsequently, the composite yarn prepared above is woven into a fabric sensing electrode (LTHMF) with a honeycomb microstructure by a multi-shuttle layer-by-layer weft insertion weaving process, where the number of weft shuttles is 2, the number of weft layers is 2, the weft insertion angle is 30°, and the weaving density of warp and weft yarns is 400 per 10 cm. The structure of the constructed sensing electrode is as shown in Figure 1 c of; The conductive silver paste is printed on the surface of the non-woven fabric substrate by a transfer printing technique, where the solid content of the conductive silver paste is 70 wt%, and the transfer loading of the conductive silver paste is 5 wt%, to form a fabric interdigital electrode; The above-prepared MPHMF sensing electrode and the interdigital electrode are assembled into a fabric tactile sensor by ultrasonic bonding (ultrasonic power: 2000 W, ultrasonic frequency: 20 kHz, time: 2 minutes);

[0077] (2) Construction of the surface texture recognition system: The bionic structure fabric-based tactile sensor constructed in step (1), a microcontroller, a transimpedance amplifier, and an inverter are used to form the surface texture recognition system. First, the system generates a 0.5V power supply to power the sensor. The transimpedance amplifier is connected between the negative pole of the sensor and the ground. Its function is to convert the current flowing through the sensor into a voltage. The amplification factor of the transimpedance amplifier is 330. Since the gain of the transimpedance amplifier is negative, the output of the transimpedance amplifier is connected to an inverter to convert the voltage into a positive voltage that can be input into the microcontroller. The output of the inverter is connected to the pin of the built-in ADC of the microcontroller. In this way, the voltage measured by the microcontroller divided by 330 is the current flowing through the sensor. The constructed surface texture recognition system is used to collect data on various surface textures, establish a classification model database, and then compare the real-time signals collected by the sensor with the classification model database established by machine learning before through machine learning technology, so as to perceive and predict the surface texture. First, the microsensor continuously collects the response current data of the sensor to the surface texture at a sampling frequency of 100Hz. Then, the microcontroller sends the data to the computer through the serial port. After the computer receives the data, it will perform data denoising processing: when the value reaches a certain threshold of 50×10 -6 data collection starts. When 512 data are collected, the data will be normalized and input into the model. The model specifically uses a model composed of 3 residual blocks. Each residual block consists of convolution, residual pooling, fully connected, etc.

[0078] Example 4

[0079] This example is a construction method of a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0080] (1) Preparation of the bionic microstructured fabric tactile sensor:

[0081] First, the surface of the cotton yarn (metric count: 80s) was deposited with a transition metal carbide / nitride aqueous dispersion (i.e., MXene aqueous dispersion) and a poly(3,4-ethylenedioxythiophene) / polystyrene sulfonate aqueous solution (i.e., PEDOT:PSS aqueous solution) by an alternate deposition technique: The yarn was first cationically modified with a quaternary ammonium cationic modifier of dodecyl dimethyl benzyl ammonium chloride (1227) (10 mg / mL aqueous solution, Sinopharm Group Co., Ltd.) at a bath ratio of 1:20 and impregnated at 60 °C for 5 minutes to make the yarn positively charged, and the cationically modified yarn was obtained after drying at 60 °C; Subsequently, the cationically modified yarn was impregnated in the MXene aqueous dispersion (10 mg / mL) at a bath ratio of 1:20 and 60 °C for 5 minutes to obtain the MXene-modified yarn; The MXene-modified yarn was repeated the above cationic modification process again to make it positively charged, and then it was impregnated in the PEDOT:PSS aqueous solution (10 mg / mL) at a bath ratio of 1:20 and 60 °C for 5 minutes for PEDOT:PSS deposition, and then dried at 60 °C. The above steps were repeated for 3 cycles to obtain the MXene / PEDOT:PSS composite-modified yarn; Subsequently, the composite yarn prepared above was woven into a fabric sensing electrode (SSHMF) with a honeycomb microstructure by a multi-shuttle layer-by-layer weft insertion weaving process, where the number of weft shuttles was 2, the number of weft layers was 2, the weft insertion angle was 90°, and the weaving density of the warp and weft yarns was 400 per 10 cm. The structure of the constructed sensing electrode was as shown in Figure 1 d of; The conductive silver paste was printed on the surface of the non-woven fabric substrate by a transfer printing technique, where the solid content of the conductive silver paste was 70 wt%, and the transfer loading of the conductive silver paste was 5 wt%, to form a fabric interdigital electrode; The above-prepared MPHMF sensing electrode and the interdigital electrode were assembled into a fabric tactile sensor by ultrasonic bonding (ultrasonic power: 2000 W, ultrasonic frequency: 20 kHz, time: 2 minutes);

[0082] (2) Construction of the surface texture recognition system: The bionic structure fabric-based tactile sensor constructed in step (1), a microcontroller, a transimpedance amplifier, and an inverter are used to form the surface texture recognition system. First, the system generates a 0.5V power supply to power the sensor. The transimpedance amplifier is connected between the negative electrode of the sensor and the GND. Its function is to convert the current flowing through the sensor into a voltage. The amplification factor of the transimpedance amplifier is 330. Since the gain of the transimpedance amplifier is negative, the output of the transimpedance amplifier is connected to an inverter to convert the voltage into a positive voltage that can be input to the microcontroller. The output of the inverter is connected to the microcontroller at the pin of its built-in ADC. In this way, the voltage measured by the microcontroller divided by 330 is the current flowing through the sensor. The constructed surface texture recognition system is used to collect data on various surface textures, establish a classification model database, and perceive and predict surface textures through machine learning techniques. First, the microsensor continuously collects the response current data of the sensor to the surface texture at a sampling frequency of 100Hz. Then, the microcontroller sends the data to the computer through the serial port. After the computer receives the data, it will perform data noise reduction processing: When the value reaches a certain threshold of 50×10 -6 data collection starts. When 512 data are collected, the data will be normalized and input into the model. The model specifically uses a model composed of 3 residual blocks. Each residual block consists of convolution, residual pooling, fully connected, etc.

[0083] Example 5

[0084] This example discloses a construction system for a surface texture recognition system based on a fabric tactile sensor. The system can implement the method of the above example, including a fabric sensing electrode preparation module, a fabric interdigital electrode preparation module, and a surface texture recognition system construction module;

[0085] The fabric sensing electrode preparation module is used to prepare a fabric sensing electrode with a honeycomb microstructure;

[0086] The fabric interdigital electrode preparation module is used to print conductive silver paste on the surface of the non-woven fabric substrate through a transfer printing technique to form a fabric interdigital electrode;

[0087] The surface texture recognition system construction module is used to form a surface texture recognition system according to the fabric tactile sensor, the microcontroller, the transimpedance amplifier, and the inverter.

[0088] Comparative Example 1

[0089] This comparative example discloses a construction method for a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0090] (1) Preparation of the bionic microstructure fabric tactile sensor: The specific method is the same as that in Example 1, except that the weaving structure is a non-microstructure plain weave structure;

[0091] (2) Construction of the surface texture recognition system: The specific method is the same as that in Example 1.

[0092] Comparative Example 2

[0093] This comparative example discloses a method for constructing a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0094] (1) Preparation of the bionic microstructure fabric tactile sensor: The specific method is the same as that in Example 1, except that in the preparation of the sensing electrode, the weaving technique is not used, but the honeycomb structure fabric is directly used to alternately deposit MXene and PEDOT:PSS active materials to prepare the sensing electrode (MPHMF);

[0095] (2) Construction of the surface texture recognition system: The specific method is the same as that in Example 1.

[0096] Comparative Example 3

[0097] This comparative example discloses a method for constructing a surface texture recognition system based on a fabric tactile sensor, including the following steps:

[0098] (1) Preparation of the bionic microstructure fabric tactile sensor: The specific method is the same as that in Example 1, and the active material is a single MXene material;

[0099] (2) Construction of the surface texture recognition system: The specific method is the same as that in Example 1.

[0100] The transfer printing technology of the present invention is a prior art. Specifically, first, an interdigital electrode stamp is engraved, and then it is dipped in conductive silver paste and printed on the fabric surface.

[0101] The trade name of the conductive silver paste of the present invention is Weixiulao NANO20. Of course, other prior arts can also be used.

[0102] The establishment of the classification model database of the present invention includes repeatedly collecting the response signals of the samples to the samples through the prepared sensors for the samples, and collecting the signals of the samples (the present invention samples at a sampling frequency of 100 Hz, and the sample signals contain 512 data points) to obtain the database of this sample. This process is repeated for other samples, and then the classification model database is obtained.

[0103] The machine learning technology of the present invention preferably adopts the technology of neural network learning.

[0104] The structural models of the sensing electrodes in Examples 1 to 4 are as Figure 1 shown, and are composed of Figure 1It can be seen that there are grid microstructures of a certain size on the surfaces of the sensing electrodes in the embodiments. Embodiments 1 to 3 are both double-sided microstructures, while Embodiment 4 is a single-sided microstructure; the thicknesses of the sensing electrodes in Embodiments 1 to 4 are 1.3 mm, 1.6 mm, 0.56 mm, and 0.4 mm respectively. The relatively high electrode thicknesses of Embodiments 1 and 2 are mainly due to the relatively large height of the double-sided grids, while Embodiment 4 has the smallest thickness due to the single-sided microstructure. The differences in the microstructural dimensions of the sensing electrodes in Embodiments 1 to 4 are mainly due to the differences in the weaving process parameters of multi-shuttle layer-by-layer weft insertion during the weaving process.

[0105] The observation results of the sensing electrodes of Embodiments 1 to 4 through an optical microscope are as Figure 2 shown. From Figure 2 it can be seen that the sensing electrode of Embodiment 1 has the smallest honeycomb size (1.5 * 1.5 mm) (there is a scale bar in the lower right corner of each figure, and the size can be calculated according to the scale bar), and from its three-dimensional structure diagram, it can be seen that the honeycomb microstructure is mainly formed by the interweaving of yarns to form a grid structure ( Figure 3 ). The three-dimensional honeycomb structure of Embodiment 1 is also further confirmed in the SEM images, as Figure 4 shown.

[0106] The tactile perception performances of Embodiments 1 to 4 are as Figure 5 shown. From Figure 5 it can be seen that Embodiment 1 has the widest monitoring range (0 - 400 kPa) and the largest relative current response (ΔI / I 0 ). Through finite element simulation analysis of the microstructure deformation of Embodiments 1 to 4 under the same pressure, the results are as Figure 6 shown. From Figure 6 it can be seen that under the same pressure, the compression deformations of the sensing electrodes of Embodiments 1 to 4 are 12.3%, 5.6%, 9.5%, and 7.5% respectively. Therefore, the better tactile sensing performance of Embodiment 1 is mainly due to its larger deformation area under external forces. The further linear fitting results of the current response curves of Embodiments 1 to 4 are as Figure 7 shown. From Figure 7 it can be seen that the sensitivities of Embodiment 1, Embodiment 2, Embodiment 3, and Embodiment 4 are 2486.9 kPa -1 , 677.7 kPa -1 , 25.1 kPa -1 , and 87.8 kPa -1 respectively. The highest sensitivity of Embodiment 1 is mainly due to the fact that the surface microstructure of its MPHMF is more likely to deform under external forces, which is also confirmed in the Figure 6 finite element model analysis

[0107] The tactile perception curves of Embodiment 1 and Comparative Examples 1 to 3 are as Figure 8As shown, from Figure 8 it can be seen that the sensitivities of Example 1 are 35.5 times, 3.1 times, and 4.3 times that of Comparative Example 1, Comparative Example 2, and Comparative Example 3, respectively (in all cases, the highest sensitivities in the figures are taken for comparison. For example, the sensitivity of Example 1 is 2486.9 kPa -1 , that of Comparative Example 1 is 70.1 kPa -1 , that of Comparative Example 2 is 792.6 kPa -1 , and that of Comparative Example 3 is 573.7 kPa -1 ). Example 1 has the most excellent tactile perception performance compared with the comparative examples. This is mainly because in Comparative Example 1, due to the lack of microstructures on the surface of the sensing electrode, the deformation area of the microstructures under external force stimulation is small; in Comparative Example 2, since it is for the surface modification of the microstructured fabric, the active material is difficult to penetrate into the internal yarns of the fabric, and it is difficult to establish a complete conductive path inside the yarns, deteriorating the electron transfer process during tactile perception; in Comparative Example 3, since a single MXene material is used, the conductivity is lower than that of Example 1, so the electrode interface resistance is larger, and thus the tactile perception performance is poor.

[0108] For Example 1, according to the test method of high-frequency mechanical vibration as Figure 9 shown, the test results are as Figure 10 and Figure 11 shown. From Figure 10 it can be seen that the maximum perceivable frequency of Example 1 for mechanical vibration can be as high as 100 Hz; and the number of electrical signal response peaks shows an obvious linear relationship with the vibration frequency ( Figure 11 ), indicating that Example 1 of the present invention has excellent perception ability for high-frequency mechanical vibration. This is mainly because Example 1 has rich surface microstructures and a fast response recovery rate, and its perception ability for mechanical vibration far exceeds that of piezoresistive tactile sensors reported in other literatures.

[0109] The schematic diagram of the surface texture recognition system constructed by the present invention is as Figure 12 shown, mainly including a tactile sensor, a microcontroller, a transimpedance amplifier, and an inverter. The surface texture is recognized and predicted by machine learning. The surface texture recognition system collects data for training to extract texture features, and then establishes a template signal. During the repeated sliding of the sensor in the surface texture recognition system on the sample surface, the real-time current response signal of the sample texture is collected, and the collected texture signal is compared with the template signal obtained by machine learning, thereby realizing the perception of the surface texture.

[0110] The physical diagram of the surface texture recognition system constructed by the present invention is as Figure 13 shown. The recognition results of Example 1 for the surface texture at a constant sliding speed (20 cm / s) and a random speed are as Figure 14 and 15As shown. It can be seen from the figure that whether it is constant high-speed sliding or randomly variable speed, Example 1 shows an identification accuracy of more than 99.7%, demonstrating excellent identification accuracy and ultra-high identification speed. Both the identification speed and accuracy are better than those of human skin. The results of Example 1 for the identification of other surface textures are as Figure 16 and 17 shown. It can be seen from the figure that the identification accuracy of Example 1 for other complex surface textures is also above 99.5%, demonstrating excellent universality.

[0111] The identification results of Example 2, Example 3, and Example 4 for the same surface texture are respectively as Figure 18 , 19 , and 20 shown. It can be seen from the figure that the identification accuracies of Example 2, Example 3, and Example 4 for the same surface texture are 80.7%, 70.8%, and 60.5% respectively. The identification accuracy of Example 2 to 4 is much lower than that of Example 1. The lower identification accuracy of Examples 2-4 compared to Example 1 is mainly due to the fact that Example 1 has a more optimal surface microstructure size, which is more conducive to the perception of static force and high-frequency vibration, and its more excellent tactile perception ability.

[0112] The identification result of Example 1 for the surface texture of a complex curved surface is as Figure 21 shown. It can be seen from Figure 21 that the identification accuracy of Example 1 for the surface texture of the curved surface structure is 99.7%. The identification result of Example 1 for the surface texture of a soft structure is as Figure 22 shown. It can be seen from the figure that the identification accuracy of Example 1 for the surface texture of the soft structure is 95.7%, demonstrating excellent robustness and wide adaptability.

[0113] The identification results of Comparative Example 1, Comparative Example 2, and Comparative Example 3 for the surface texture are respectively as Figure 23 , 24 , and 25 shown. It can be seen from the figure that the identification accuracies of Comparative Example 1, Comparative Example 2, and Comparative Example 3 for the same surface texture are 40.7%, 70.7%, and 62.8% respectively. The lowest surface texture identification accuracy of Comparative Example 1 is mainly due to the lack of surface microstructure of the sensor, making it difficult to efficiently perceive static force and mechanical vibration; the lower identification accuracy of Comparative Example 2 is mainly due to the use of surface deposition technology for its sensing electrode, and it is difficult for the active material to penetrate into the electrode interior, resulting in poor tactile perception performance of the device; the lower identification accuracy of Comparative Example 3 is mainly due to the use of a single MXene material for the electrode active material, resulting in lower tactile perception performance of the device.

[0114] In summary, the present invention has developed an efficient surface texture recognition system based on a fabric-based tactile sensor based on knitting technology and transfer technology, with the following advantages:

[0115] (1) Strong technical promotion; there are many choices for the conductive active materials used in the present invention, and the flexible substrate can be a textile substrate or other flexible films;

[0116] (2) The process is simple and suitable for mass production;

[0117] (3) The present invention uses a weaving process to prepare the microstructured sensing electrode. The electrode microstructure is adjustable, controllable, and diverse in structure, and the electrode structure is stable;

[0118] (4) The present invention uses a transfer printing technology to prepare the interdigital electrode. The interdigital electrode has good conductivity, no penetration between electrodes, and has a high preparation accuracy;

[0119] (5) Utilize the multi-dimensional deformation effects generated by the bionic microstructures on the surface of the sensing electrode under external forces, including vertical compression deformation and horizontal shear deformation, so as to simultaneously achieve efficient perception of static forces and mechanical vibrations. The sensitivity to static force perception is as high as 2486.9 kPa -1 , and the response frequency to mechanical vibration is as high as 100 Hz, far exceeding that of the piezoresistive tactile sensors reported in the literature;

[0120] (6) The tactile sensor constructed by the present invention has an ultra-fast response speed to static forces and mechanical vibrations. The response times are 10 mS and 3 mS respectively, which are better than the response speed of human skin;

[0121] (7) Due to the efficient perception ability of the tactile sensor constructed by the present invention for static forces and mechanical vibrations, it can achieve precise perception and recognition of surface fine textures. The surface texture recognition system constructed by the present invention has ultra-high recognition accuracy and recognition speed. The recognition accuracy for more than 21 kinds of textures is above 98.9%, and the recognition speed is as high as 30 cm / s, far exceeding the recognition speed and accuracy of human skin;

[0122] (8) The surface texture recognition system constructed by the present invention has excellent robustness and has super adaptability to complex shapes and curved surface structures, and all have ultra-high recognition accuracy;

[0123] (9) The bionic multi-level structure flexible pressure sensor prepared by the present invention not only has good air permeability, moisture permeability and wearing comfort, but also has good pressure sensing performance;

[0124] (10) The performance of the tactile sensor prepared by the present invention is adjustable and controllable, and the performance of the device and the recognition accuracy of the system can be adjusted by adjusting the microstructure model of the sensing electrode;

[0125] (11) Wide application range; the surface texture recognition system constructed by the present invention can be used in fields such as humanoid robots and human-computer interaction.

[0126] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples.

[0127] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principle and practical application of the invention, so that those skilled in the art can well understand and utilize the invention.

Claims

1. A method for constructing a surface texture recognition system based on a fabric tactile sensor, characterized in that: The following steps are involved: S1. Preparation of a fabric sensing electrode with a honeycomb microstructure; The specific steps include: S11, depositing MXene aqueous dispersion and PEDOT:PSS aqueous solution on the surface of cotton yarn by alternating deposition technology to obtain MXene / PEDOT:PSS composite modified yarn; S12, weaving the MXene / PEDOT:PSS composite modified yarn into a fabric sensing electrode with a honeycomb microstructure through a multi-shuttle layered weft insertion weaving process; S2, preparing a fabric interdigital electrode; specifically comprising the following steps: S21, printing the conductive silver paste on the surface of the non-woven fabric substrate by transfer printing technology to form a fabric interdigital electrode; S3, constructing a surface texture recognition system; specifically comprising the following steps: S31, assembling the fabric sensing electrode with the honeycomb microstructure and the fabric interdigital electrode into a fabric tactile sensor by ultrasonic bonding; S32: construct a surface texture recognition system according to the fabric tactile sensor, microcontroller, transimpedance amplifier and inverter.

2. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 1, characterized in that: The S11 comprises the following steps: S111, immersing the cotton yarn in a quaternary ammonium salt cationic modifier at a bath ratio of 1:5 to 1:30 at 20 to 80° C. for 1 to 30 minutes to make the yarn positively charged; and then drying at 60° C. to obtain a cationic modified yarn; S112, immersing the cationic modified yarn in a MXene aqueous dispersion at a bath ratio of 1:5 to 1:30 for 1 to 30 minutes at 20 to 80° C. to obtain a MXene modified yarn; S113, repeating the process of S111 with the MXene modified yarn to obtain a positively charged MXene modified yarn; then immersing the positively charged MXene modified yarn in a PEDOT:PSS aqueous solution at a bath ratio of 1:5 to 1:30 at 20 to 80° C. for 1 to 30 minutes to perform PEDOT:PSS deposition, and then drying at 60° C.; S114. Repeat S111, S112 and S113 three times to obtain MXene / PEDOT:PSS composite modified yarn.

3. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: The mass concentration of the quaternary ammonium salt cationic modifier described in S111 is 10 mg / mL.

4. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: The mass concentration of the MXene aqueous dispersion in S112 is 10 mg / mL.

5. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: The mass concentration of the PEDOT:PSS aqueous solution in S113 is 10 mg / mL.

6. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: In the weaving process of multi-shuttle layered weft insertion described in S12, the number of weft shuttles is 2 to 3, the number of weft layers is 2 to 6, the weft insertion angle is 15 to 90°, and the weaving density of the warp and weft yarns is 400 yarns / 10 cm.

7. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: The solid content of the conductive silver paste in S21 is 70 wt %, and the transfer loading amount of the conductive silver paste is 5 wt %.

8. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 3, characterized in that: The ultrasonic bonding method in S31 has an ultrasonic power of 2000 W, an ultrasonic frequency of 20 kHz and a duration of 2 minutes.

9. The method for constructing a surface texture recognition system based on a fabric tactile sensor according to claim 2, characterized in that: The S32 comprises the following steps: S321, supplying power to the fabric tactile sensor, converting the current flowing through the fabric tactile sensor into a voltage through the transimpedance amplifier, and then converting the voltage into a positive voltage through an inverter and inputting it into a microcontroller.

10. A system for implementing the method for constructing a surface texture recognition system based on a fabric tactile sensor as claimed in any one of claims 1 to 9.

Images