Multilayer fabric type sensing glove based on carbon black-silica gel system, preparation method and gesture reconstruction device

Abstract

The invention relates to the field of flexible wearable sensing equipment, in particular to a multi-layer fabric type sensing glove based on a carbon black-silica gel system, a preparation method and a gesture reconstruction device. The sensing glove comprises a fabric substrate which is sewn and printed with a carbon black-silica gel composite material, and a silver fiber line and a button electrode are assembled on the fabric substrate; the fabric substrate comprises an upper-layer fabric substrate and a lower-layer fabric substrate; printing a carbon black-silica gel composite material on the fabric substrate to form a printing layer which is used as a sensing layer and a conductive layer; the printing layer comprises an upper outer side printing layer, an upper inner side printing layer and a lower outer side printing layer; the upper outer side printing layer and the upper inner side printing layer are connected through a via hole, and the upper outer side printing layer and the lower outer side printing layer are connected through a silver fiber wire; the sensing layer is arranged on the printing layer on the outer side of the upper layer to form a plurality of strain sensors. According to the sensing glove, the number of rigid electronic components is reduced, and the integration, the stability and the wearing comfort of the sensing glove are improved.

Description

Technical Field

[0001] The present invention relates to the field of flexible wearable sensing devices, and particularly to a multi-layer fabric sensing glove based on a carbon black-silica gel system, a preparation method and a gesture reconstruction device. Background Art

[0002] With the progress of technologies in various fields such as material science, medicine and electronics, intelligent wearable devices are gradually changing people's lifestyles and working methods. Among them, the hand, as the most frequently used body part of people, plays an important role in daily life, and full-hand pose reconstruction and real-time tracking are required in fields such as virtual reality and medical rehabilitation. These application scenarios pose the following challenges to wearable sensing gloves: (1) The pose reconstruction angle of the hand should be accurate enough; (2) The movement of the hand must be acquired and tracked in real time. (3) Have a certain reliability and stability; (4) The glove must be worn in a non-invasive manner and cannot affect the wearer's normal activities and work; (5) Can be manufactured in a simple method and in large quantities at low cost.

[0003] Currently, the research on gesture reconstruction at home and abroad mainly adopts three technical solutions: (1) Gesture reconstruction based on computer vision: This method captures hand images through a depth camera, and then uses computer vision technology to calculate the positions of hand bones or the angles of joints, so as to reconstruct the hand pose. Usually, the solution based on computer vision is vulnerable to occlusion, light and background interference, and multiple cameras or optical markers need to be set, which limits its scope of application. (2) Gesture reconstruction based on an inertial measurement unit (IMU) sensing glove: This method calculates the attitude angles measured by the IMUs fixed on the glove, solves the rotation matrices between each coordinate system, and finally obtains the relative position relationship, so as to calculate the hand pose. Due to the zero drift problem of the IMU, the sensing glove based on the IMU has the problem of reduced accuracy during long-term use, and the coordinate system of the IMU is not aligned with the coordinate system of the body or joint, and the glove needs to be calibrated in advance, and the rigidity and bulky size of the IMU itself make the sensing glove have a complicated appearance and is not conducive to wearing. (3) Gesture reconstruction based on a strain sensor glove: This method calculates the finger bending angle by measuring the strain generated on the finger back when the finger joint bends through a flexible strain sensor, so as to reconstruct the hand pose. Currently, in the manufacturing methods of sensing gloves based on flexible strain sensors, most of them manufacture the sensors separately and then fix the sensors on the glove or clothing through various processes. This method will cause instability of the glove structure and discomfort in wearing. A few methods directly integrate the sensors into the glove fabric. However, the improvement of this performance often comes at the cost of complex manufacturing processes and high metal material costs. Summary of the Invention

[0004] To solve the problems existing in the prior art, the present invention proposes a multi-layer fabric-type sensing glove based on a carbon black-silica gel system, a preparation method, and a gesture reconstruction device. The strain sensor array is integrated on the glove with a double-layer fabric substrate, which has a simple process and low cost. The obtained fabric-type sensing glove can complete accurate hand posture reconstruction while ensuring the comfort of the wearer.

[0005] In an embodiment of the present invention, a multi-layer fabric-type sensing glove based on a carbon black-silica gel system includes a fabric substrate printed with a carbon black-silica gel composite material that is sewn together, and silver fiber wires and button electrodes are assembled on the fabric substrate. The fabric substrate includes an upper fabric substrate and a lower fabric substrate; a carbon black-silica gel composite material is printed on the fabric substrate to form a printed layer, which serves as a sensing layer and a conductive layer. The printed layer includes an upper outer printed layer, an upper inner printed layer, and a lower outer printed layer; the upper outer printed layer and the upper inner printed layer are connected through vias, and the upper outer printed layer and the lower outer printed layer are connected through silver fiber wires; the sensing layer is arranged on the upper outer printed layer to form a plurality of strain sensors. As the finger bends, the sensor undergoes strain and the resistance increases accordingly.

[0006] In an embodiment of the present invention, the method for preparing the above multi-layer fabric-type sensing glove specifically includes the following steps: S1. Using carbon black as a conductive medium, prepare a carbon black-silica gel conductive composite paste, and print the paste on both sides of a fabric substrate and one side of another fabric substrate through a designed screen printing mold respectively. S2. Punch holes at the via positions and the button electrode positions. The via positions are the column electrode centers of the first row of sensors and the column electrode centers of the third row of sensors. S3. Apply the composite paste on both sides of the via and cure it. The composite paste is in full contact with the printed layer to form a stable electrical connection. S4. Sew silver fiber wires on the protruding areas of the upper outer printed layer and the lower outer printed layer, and achieve electrical connection through the contact between the silver fiber wires and the printed layer; finally, sew the two fabric substrates with knitting threads to make a complete glove.

[0007] In an embodiment of the present invention, a gesture reconstruction device for a fabric-type sensing glove includes: The above multi-layer fabric-type sensing glove, which is used to obtain the knuckle bending angle information of the finger and map the knuckle bending angle information into a resistance change signal; A signal acquisition and transmission circuit, which is used to read the voltage signal caused by the resistance change of the strain sensor on the fabric-type sensing glove and send the read voltage signal to the hand posture calculation module; The hand gesture calculation module is used to receive the voltage signal sent by the signal acquisition and transmission circuit, and calculate in real time the resistance values of the sensors in the strain sensor, estimate the finger bending angles according to the resistance values, and then predict the bending angles of each finger joint through a deep learning network to reconstruct the hand gesture.

[0008] As can be seen from the above technical solutions, the present invention avoids the instability problems caused by the traditional mechanical connection method of sensors and gloves, reduces the number of rigid electronic components, improves the integration, stability and wearing comfort of the sensing glove, and has important reference value for the perception of human movement in the field of wearable sensing devices.

[0009] Compared with the prior art, the beneficial effects achieved by the present invention include: 1. The sensing glove of the present invention does not contain any rigid electronic components, has a non-invasive all-textile form, does not affect the normal activities of the wearer, and has good wearing comfort.

[0010] 2. The sensing glove of the present invention adopts a printed integration scheme, has high stability and integration, is not easy to cause sensor detachment during long-term use, and has a thinner and lighter volume.

[0011] 3. The present invention provides an efficient scheme for integrating a strain sensor array of a carbon black-silica gel system onto a fabric glove, reduces the number of additional connection components and interface electrodes, and optimizes the manufacturing process.

[0012] 4. The fabric-type sensing glove of the present invention can generate stable response signals according to hand movements and can calculate hand postures through deep learning algorithms. Description of the Drawings

[0013] Figure 1 is the structural diagram of the fabric-type sensing glove in the embodiment of the present invention; Figure 2 is the schematic diagram of the distribution of three rows of strain sensors on the upper outer printing layer; Figure 3 is the row and column electrode connection method of the strain sensor array in the upper outer printing layer and the upper inner printing layer and the corresponding button electrode positions, where A, B, C, D, and E are the positions of the button electrodes of the column electrodes, and F, G, and H are the positions of the button electrodes of the row electrodes; Figure 4 is the equivalent circuit diagram of the strain sensor array; Figure 5 is the screen printing pattern of the three-layer printing layer; Figure 6 is the test result of the absolute resistance change rate of the resistance of each block with strain; Figure 7It is a curve showing the change of the sensor resistance with hand movements. Detailed implementation manners

[0014] The technical solutions of the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, but the implementation manners of the present invention are not limited thereto. Embodiment

[0015] Due to the special structure of the human hand, a general double-layer sensor array based on a resistance matrix cannot be directly deployed on a fabric sensing glove, because the sensors between each finger can only be connected proximally, and direct connections in the middle and distal parts will affect the normal movement of the fingers. For this reason, this embodiment provides a printed multi-layer fabric-type sensing glove based on a carbon black-silicone system, with sensors arranged at finger joints to sense the bending changes of the fingers, so as to detect hand posture signals, with simple process and low cost. Specifically, a 3×5 strain sensor array is integrated on a glove with a double-layer fabric substrate, reducing the number of rigid electronic components and additional connection components. The fabric-type sensing glove can complete accurate hand posture reconstruction while ensuring the comfort of the wearer.

[0016] Refer to Figure 1 , the fabric-type sensing glove of this embodiment has a multi-layer sensing-conductive integrated fabric structure, which specifically includes two layers of knitted fabrics printed with carbon black-silicone composite materials and sewn together, and male buttons of silver fiber wires and conductive button electrodes (such as metal button electrodes) are assembled on the knitted fabrics. Among them, the knitted fabric serves as the fabric substrate, including an upper-layer (dorsal layer) fabric substrate and a lower-layer (palm layer) fabric substrate; a carbon black-silicone composite material is printed on the fabric substrate to form a printed layer, which serves as both a sensing layer and a conductive layer. In this embodiment, three printed layers are printed on the two fabric substrates, namely an upper-layer outer printed layer, an upper-layer inner printed layer, and a lower-layer outer printed layer; partial structural electrical connection is achieved between the upper-layer outer printed layer and the upper-layer inner printed layer through vias, and electrical connection is achieved between the upper-layer outer printed layer and the lower-layer outer printed layer through silver fiber wires.

[0017] This embodiment designs an adaptable printed pattern according to the shape and structure of the human hand. The printed pattern includes a connection part and a sensing part; the sensing part is designed in a serpentine shape to increase the resistance; the connection part is designed in a strip shape according to the space situation. Among them, the sensing part is arranged on the upper-layer outer printed layer to form multiple strain sensors. As the finger bends, the sensors are strained and the resistance increases accordingly.

[0018] Combine Figure 2, 15 strain sensors are printed on the outer upper printing layer, with 3 strain sensors printed on each finger, specifically including the first row of sensors arranged at the distal phalanx position, the second row of sensors arranged at the middle phalanx position, and the third row of sensors arranged at the proximal phalanx position. The two poles of each sensor are the row electrode and the column electrode respectively; among them, the distal end of the first row of sensors is the row electrode, and the proximal end is the column electrode. The distal ends of the second row of sensors and the third row of sensors are both column electrodes, and the proximal ends are both row electrodes.

[0019] The row electrodes of the first row of sensors are electrically connected through silver fiber wires and the outer lower printing layer, thereby realizing the interconnection between the row electrodes of the first row of sensors. The column electrodes of the first row of sensors and the second row of sensors are designed to be directly connected through the same connecting part, and the first row of sensors and the second row of sensors form a symmetric arrangement with opposite electrodes. The column electrodes of the first and second rows of sensors are electrically connected to the column electrodes of the third row of sensors through vias and the inner upper printing layer. The row electrodes and column electrodes of all sensors are centrally led to the electrode lead-out area on the back of the hand through the connecting parts of the printing layer, and are directly connected to the flexible circuit board through button electrodes without the need to install additional connecting wires.

[0020] In this embodiment, protruding areas are provided on the outer upper printing layer and the outer lower printing layer for sewing silver fiber wires; that is, the silver fiber wires are sewn in the protruding areas of the outer upper printing layer and the outer lower printing layer to realize the electrical connection between the outer upper printing layer and the outer lower printing layer, and further realize the electrical connection of the row electrodes of the first row of sensors; among them, the protruding areas are located at the fingertips of the fingers and the extensor / flexor tendons of the little finger, as Figure 1 shown. The outer upper printing layer and the inner upper printing layer are electrically connected through the Figure 1 vias shown, thereby realizing the electrical connection of the column electrodes of the second row of sensors and the third row of sensors. The vias are filled with the same paste as the printing layer (i.e., the paste of the same carbon black-silica composite material), and the paste needs to fill the entire via and make full contact with the coating of the column electrode. The male button electrode of the metal button electrode is installed in the electrode lead-out area designed on the printing layer, which can be conveniently directly connected to the female button on the flexible circuit board.

[0021] Combined with Figure 3 and Figure 4 , it shows the connection methods of the row and column electrodes of the strain sensor array on the outer upper printing layer and the inner upper printing layer, as well as the corresponding positions of the button electrodes and the equivalent circuit. The three printing layers, vias and silver fiber wires form a 3×5 row-column shared resistance matrix, and only 10 vias and 6 groups of silver fiber wires are required respectively. The column electrodes of the third row of sensors are designed as circles with a diameter of 4 mm to reduce the occupation of the space of the connection part of the row electrodes of the second row of sensors.

[0022] One button electrode is led out from each row and each column of the row-column shared resistance matrix, and a total of 3 rows and 5 columns of button electrodes are obtained. Among the 8 button electrodes, 4 button electrodes are in contact with the upper outer printing layer, and the other 4 button electrodes are in contact with the upper inner printing layer to avoid conflicts in different electrode connection areas. The column electrode buttons of the index finger, middle finger, ring finger and little finger are all in contact with the upper inner printing layer to achieve electrical connection, and all row electrode buttons and the column electrode button of the thumb are in contact with the upper outer printing layer to achieve electrical connection.

[0023] In one embodiment, the design pattern of the printing layer is as Figure 5 shown, where the width of the sensing area is not higher than 1 mm and the length is not lower than 73 mm; the minimum width of the connection area is not lower than 8 mm and the maximum length is not higher than 16 mm. The pattern design rules of the sensing area and the connection area follow the size effect theory, and the effective aspect ratio of the connection area of each part is related to the shape and current path. Among them, the upper inner printing layer and the lower outer printing layer both serve as connection areas, and the upper outer printing layer is provided with a sensing area and a connection area.

[0024] That is to say, in this embodiment, the upper outer printing layer mirrors and flips the first row of sensors, with the distal end as the row electrode and the proximal end as the column electrode. The row electrode of the first row of sensors forms an interconnection with the lower outer printing layer through a silver fiber wire and is led out to the position of the button electrode. The column electrode of the first row of sensors is directly connected to the column electrode of the second row of sensors and is connected to the column electrode of the third row of sensors through vias and the upper inner printing layer. The row electrodes and column electrodes of all sensors are led to the electrode lead-out area on the back of the hand through the connection area of the printing layer, and then directly connected to the flexible circuit board through the button electrode without the need to install additional wires.

[0025] In this embodiment, the preparation method of the fabric-type sensing glove mainly includes the following steps: S1. Using carbon black as the conductive medium, prepare a carbon black-silicone conductive composite paste, and print the paste-like composite paste on both sides of a fabric substrate and one side of another fabric substrate through a designed screen printing mold respectively; a rectangular frame line is designed around the pattern of the screen printing mold to achieve alignment of different printing layers.

[0026] Specifically, carbon black (3 g), silica gel (30.3 g), and silicone oil (45 g) are mixed, and 1.8 g of a curing agent is added. The mixture is stirred with a blender for 30 minutes. The stirred composite slurry is poured onto a customized screen mold, and the slurry is printed on a fabric substrate using a rubber brush board. The fabric substrate is a composite knitted fabric of polyester and spandex. The printed fabric is placed in a vacuum drying oven and heated and cured for 1 hour. Two-layer fabrics with three printed layers are prepared respectively by the above method. For the upper fabric, rectangular outer frames need to be designed on the screen mold as marks for alignment on both printed layers.

[0027] S2. Use an electric punch to punch round holes at the designed via hole positions and button electrode positions. The via hole positions are the centers of the column electrodes (square) of the first row of sensors and the centers of the column electrodes (round) of the third row of sensors.

[0028] S3. Apply the composite slurry on both sides of the via holes and cure it. The composite slurry needs to be in full contact with the printed layer to form a stable electrical connection. Install the male button electrode of the metal button electrode at the round hole of the button.

[0029] In this step, the via holes and the two-sided printed layers around them are filled with the composite slurry and cured. The male button electrode is installed at the button electrode position by the bite of two metal sheets and is in contact with the outer printed layer of the upper layer to form a stable electrical connection.

[0030] S4. Sew silver fiber wires on the protruding areas of the outer printed layer of the upper layer and the outer printed layer of the lower layer, and achieve electrical connection through the contact between the silver fiber wires and the printed layer; finally, sew the two fabric substrates with ordinary knitting threads to make a complete glove.

[0031] This embodiment also provides a gesture reconstruction device based on a multi-layer fabric type sensing glove, including: The above-mentioned fabric type sensing glove, which is used to obtain the bending angle information of 15 finger joints of 5 fingers and map the finger joint bending angle information into a resistance change signal; A signal acquisition and transmission circuit, which is used to read the voltage signal caused by the resistance change of the strain sensor on the fabric type sensing glove and wirelessly transmit the read voltage signal to the hand gesture calculation module in real time; A hand gesture calculation module, which is used to receive the voltage signal sent by the signal acquisition and transmission circuit, and calculate in real time the resistance values of the sensors in the strain sensor, estimate the finger bending angle according to the resistance values, and then predict the bending angles of each finger joint through a deep learning network to reconstruct the hand gesture to control the manipulator or the hand model in virtual reality (VR).

[0032] Among them, the hand gesture calculation module can be implemented by a host computer, including a data reception and gesture reconstruction module and an interactive display module; the data reception and gesture reconstruction module receives and unpacks the voltage signals (i.e., Bluetooth data) sent by the acquisition and transmission circuit, and maps the normalized resistance values of 100 frames of strain sensor arrays to 15 knuckle bending angles through a trained deep learning network; the interactive display module includes a manipulator control system and / or a VR hand model control interface, and realizes precise control of the manipulator and / or the hand model according to the joint bending angles predicted by the deep learning network. The deep learning network is a long short-term memory (LSTM) network.

[0033] The signal acquisition and transmission circuit includes a readout circuit and a Bluetooth data sending module. The readout circuit includes a microcontroller, which reads and calculates the resistance values of a 3×5 strain sensor array using the resistance matrix method (RMA). The row line scanning signal is controlled by the IO port of the microcontroller, and the high and low levels are loaded onto the button electrodes corresponding to the row electrodes on the circuit board. Then, the voltage on the column electrodes is obtained through the analog-to-digital conversion (ADC) peripheral inside the microcontroller. Finally, the resistance values of 15 strain sensors are calculated locally by the microcontroller according to the RMA calculation formula; the Bluetooth data sending module communicates with the microcontroller through the serial port and sends the resistance values of the sensors to the host computer at a fixed frequency.

[0034] The signal acquisition and transmission circuit is made of a flexible printed circuit board (FPCB). The flexible printed circuit board is designed with a customized button electrode package, which consists of double-sided rectangular pads and a central round hole. The row electrodes of each sensor on the fabric-type sensing glove are connected to the IO ports of the microcontroller (STM32F103RCT6), the column electrodes are connected to the ADC channels, and the Bluetooth data sending module is connected to the serial port. Among them, the electrode is designed with a customized PCB package according to the size of the metal button electrode. The PCB package consists of square double-sided pads and round vias. After installing the female button of the metal button electrode on the PCB, it is further soldered with solder to achieve a stable electrical connection. After installing the female button of the metal button electrode on the button electrode package, it is soldered with solder to form a stable electrical connection. After installing the metal electrode button, a stable electrical-mechanical double connection between the flexible printed circuit board and the fabric-type sensing glove can be directly achieved through the electrode button.

[0035] The readout circuit performs dynamic row scanning (each row level is pulled low respectively, and the other rows are pulled high), obtains the voltage of each column through the ADC channel, obtains the resistance values of each sensor by the microcontroller according to the resistance matrix method calculation formula, and sends them to the host computer at a frame rate of 30Hz through Bluetooth. The host computer is written in the Python language. After reading the Bluetooth-received data, it unpacks and filters the data, and draws a curve graph of the resistance change of 15 sensors over time, and saves the resistance data at the same time. All functions of the host computer are run in real time through a program called at regular intervals.

[0036] In this embodiment, the calculation formula of the resistance matrix method is as follows: ; Wherein, R ij is the resistance of each sensor, V ij is the voltage on the j-th column when scanning the i-th row, V rj is the voltage on the j-th column when scanning the reference resistance, V cc is the power supply voltage of the microcontroller, R rj is the reference resistance value.

[0037] Furthermore, in addition to real-time plotting of the resistance-time curve, the host computer also inputs the resistance data sequence into the trained LSTM network. The sequence consists of 15 resistance data for 100 frames. The LSTM network maps the resistance data sequence into 15 finger joint angles, and then inputs them into the script in Unity to control the actions of the hand model in the virtual scene; or sends them to the Arduino control board of the manipulator via Bluetooth to control the manipulator to perform corresponding actions. The training dataset consists of sensor data and labels. The labels use the joint angle data collected by a commercially available gesture reconstruction glove (MANUS) as the gold standard. When collecting data, first wear the fabric-based sensing glove of the present invention, and then put on the MANUS gesture reconstruction glove outside.

[0038] The performance test of the fabric-based sensing glove and its gesture reconstruction device in this embodiment is as follows: Combined with Figure 6 , the fabric printed with the conductive material is divided into 41 blocks. Each block is stretched to 60% strain with a universal testing machine, and the resistance is measured with a Keithley digital multimeter to obtain the resistance-strain relationship diagram, and the absolute resistance change rate of the resistance of each block with strain is calculated by derivation; the results show that the resistance change rate of the sensing area is much larger than that of the connection area, that is, the resistance change generated in the sensing area under the same strain is much higher than the resistance change in the connection area, proving that the designed printed layer pattern meets the design requirements of the size effect.

[0039] Combined with Figure 7, test the response signal of the gesture reconstruction device based on the fabric-type sensing glove to hand movements. After the user wears the glove and connects to the signal acquisition and transmission circuit, each finger performs individual bending movements and combined bending movements, and the upper computer records the curve of the resistance change of each sensor over time. In the figure, the first letter of the English letter abbreviation represents the finger name: T - thumb, I - index finger, M - middle finger, R - ring finger, L - little finger; the following letters are the abbreviations of the finger joint names, CMC - the first metacarpophalangeal joint, MCP - metacarpophalangeal joint, IP - interphalangeal joint, PIP - proximal interphalangeal joint, DIP - distal interphalangeal joint. When the finger performs different movements, the calculated resistance signal has a stable and low-crosstalk waveform. It should be noted that the slight crosstalk shown between the ring finger and the little finger is due to the inability to complete completely independent movements between the two, rather than crosstalk between the sensors.

[0040] The above embodiments are the preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A multi-layer fabric sensing glove based on a carbon black-silicone system, characterized in that: The invention comprises a fabric substrate printed with a carbon black-silicon composite material and stitched together, and a silver fiber line and a button electrode are arranged on the fabric substrate; The fabric substrate comprises an upper fabric substrate and a lower fabric substrate; a carbon black-silica gel composite material is printed on the fabric substrate to form a printing layer, which serves as a sensing layer and a conductive layer; The printing layer includes an upper outer printing layer, an upper inner printing layer and a lower outer printing layer; the upper outer printing layer and the upper inner printing layer are connected through vias, and the upper outer printing layer and the lower outer printing layer are connected through silver fiber lines; the sensing layer is arranged on the upper outer printing layer to form multiple strain sensors. As the finger bends, the sensor is strained and the resistance increases accordingly.

2. The multi-layer fabric sensor glove according to claim 1, characterized in that: The upper outer printing layer prints three strain sensors on each finger, specifically including a first row of sensors arranged at the distal phalanx position, a second row of sensors arranged at the middle phalanx position, and a third row of sensors arranged at the proximal phalanx position, and the two poles of each sensor are a row electrode and a column electrode respectively; The far ends of the sensors in the first row are row electrodes, and the near ends are column electrodes; the far ends of the sensors in the second row and the third row are both column electrodes, and the near ends are both row electrodes.

3. The multi-layer fabric sensor glove according to claim 2, characterized in that: The row electrodes of the first row of sensors are connected to the lower outer printed layer through silver fiber lines to achieve interconnection between the row electrodes of the first row of sensors; the column electrodes of the first row of sensors and the second row of sensors are designed to be directly connected through the same connecting part, and the first row of sensors and the second row of sensors form a symmetrical arrangement of electrodes with their backs to each other; the column electrodes of the first and second rows of sensors are connected to the column electrodes of the third row of sensors through vias and the upper inner printed layer; the row electrodes and column electrodes of all sensors are concentrated and led to the electrode lead-out area on the back of the hand through the connecting part of the printed layer.

4. The multi-layer fabric sensor glove according to claim 1, characterized in that: The upper outer printing layer and the lower outer printing layer are provided with protruding areas for sewing the silver fiber thread; the protruding areas are located at the fingertips and the little finger extensor / flexor tendons.

5. The multi-layer fabric sensor glove according to claim 1, characterized in that: The via hole is filled with the same carbon black-silica gel composite material as the printed layer.

6. A method for preparing the multi-layer fabric sensor gloves according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1. Using carbon black as a conductive medium, a carbon black-silicone conductive composite slurry is prepared, and the composite slurry in a paste state is printed on both sides of a fabric substrate and one side of another fabric substrate through a designed screen printing mold; S2, drilling holes at the via hole positions and button electrode positions, where the via hole positions are the center of the column electrode of the first row of sensors and the center of the column electrode of the third row of sensors; S3, applying composite slurry on both sides of the via and curing them, so that the composite slurry is in full contact with the printed layer to form a stable electrical connection; S4. Silver fiber threads are sewn on the protruding areas of the upper outer printed layer and the lower outer printed layer to achieve electrical connection through the contact between the silver fiber threads and the printed layer; finally, two fabric bases are sewn with knitted thread to make a complete glove.

7. A gesture reconstruction device for a fabric sensor glove, characterized in that: include: The multi-layer fabric sensor glove according to any one of claims 1 to 5, used to obtain the bending angle information of the finger joints, and map the bending angle information of the finger joints into a resistance change signal; A signal acquisition and transmission circuit is used to read the voltage signal caused by the resistance change of the strain sensor on the fabric sensing glove, and send the read voltage signal to the hand posture solution module; The hand posture calculation module is used to receive the voltage signal sent by the signal acquisition and transmission circuit, and calculate the resistance value of each sensor in the strain sensor in real time, estimate the finger bending angle based on the resistance value, and then predict the bending angle of each finger joint through a deep learning network to reconstruct the hand posture.

8. The gesture reconstruction device according to claim 7, characterized in that: The hand posture solution module includes a data receiving and gesture reconstruction module and an interactive display module; the data receiving and gesture reconstruction module receives and unpacks the voltage signal sent by the acquisition transmission circuit, and maps the resistance value of the strain sensor to the bending angle of the finger joints through the trained deep learning network; the interactive display module includes a manipulator control system and / or a VR hand model control interface, and realizes precise control of the manipulator and / or hand model according to the joint bending angle predicted by the deep learning network.

9. The gesture reconstruction device according to claim 7, characterized in that: The signal acquisition and transmission circuit includes a readout circuit and a Bluetooth data sending module; the readout circuit includes a microcontroller, which uses a resistance matrix method to read and calculate the sensor resistance of the strain sensor array. The row line scanning signal is controlled by the IO port of the microcontroller, and the high and low levels are loaded onto the button electrodes corresponding to the row electrodes. The voltage on the column electrode is then obtained through the analog-to-digital conversion peripheral inside the microcontroller, and finally the sensor resistance is calculated according to the resistance matrix method calculation formula; the Bluetooth data sending module communicates with the microcontroller through the serial port, and sends the sensor resistance value to the hand posture solution module at a fixed frequency.

10. The gesture reconstruction device according to claim 9, characterized in that: The signal acquisition and transmission circuit is made of a flexible printed circuit board.

Images