Human skin-like super-resolution tactile sensor and tactile reconstruction method

Abstract

The invention discloses a human skin-like super-resolution tactile sensor and a tactile reconstruction method. According to the sensor, by referring to the bionic mechanism that different types of receptors are embedded in soft tissues of human skin, functional bionics of various human receptors (such as Pareny bodies, Meisner bodies and the like) are realized by embedding different types of sensing units in different arrangement modes in soft silica gel materials, and further, the sensor can be used for realizing the functional bionics of various human receptors (such as Pareny bodies, Meisner bodies and the like) through a tactile reconstruction method. And large-area and super-resolution sensing of forward pressure, shearing force, torsional force and the like is realized. The sensor is characterized in that the sensor focuses on bionics of human skin tissue functions rather than bionics of human body receptor functions, that is to say, by selecting a soft silica gel material with the maximum force traction coupling effect, the same external contact can cause as many sensing units to generate responses as possible; therefore, human skin-like large-area and super-resolution tactile perception is realized.

Description

Technical Field

[0001] The present invention relates to the field of robot tactile perception, and in particular to a human skin-like super-resolution tactile sensor and a tactile reconstruction method. Background Art

[0002] As a perceptual modality parallel to vision and hearing, touch can provide robots with irreplaceable and rich information about the environment they are in contact with, such as the shape, surface texture, roughness, temperature and humidity of the object being contacted, the location where the robot contacts the external object and the magnitude of the contact force, whether the robot's fingers slip on the external object during the grasping process, etc. Tactile information plays a vital role in tasks that require direct physical contact with the external environment, such as robot grasping and human-machine interaction.

[0003] At present, in order to obtain tactile information, tactile sensors based on various principles have been developed in the field of robotics, mainly piezoresistive, capacitive, photoelectric, visual tactile, etc. The essential principle of piezoresistive tactile sensors is that when single-crystal silicon materials are squeezed by external forces, the resistivity changes, which can be converted into pressure changes through calibration technology; however, it cannot be flexibly attached to a large area, and cannot directly measure the shear force accurately. Capacitive pressure sensors are usually composed of two parallel electrodes. When the sensor is subjected to external pressure, the deformation will cause the capacitance value to change, and the pressure is sensed by detecting the capacitance change; however, its anti-interference ability is poor, and it is difficult to accurately measure the shear force and torsion force. Piezoelectric sensors are based on the piezoelectric effect, that is, when piezoelectric materials are subjected to pressure, positive and negative charges will be generated on their two opposite sides, and the amount of charge generated is positively correlated with the pressure. By detecting the voltage values ​​on both sides of the material, the relationship between the voltage and the external pressure can be established; although piezoelectric sensors have higher sensitivity, their manufacturing cost and complexity are also higher, they cannot detect static pressure, and vibration and high-frequency noise will have a greater impact on them. The above-mentioned tactile sensors are usually simulations of the functions of a single receptor in the skin. Therefore, in order to improve the performance of spatial resolution, the traditional method is often to reduce the size of the sensor unit or increase the density of the sensor array. However, the traditional method can never surpass the physical resolution of the array and cannot achieve infinite resolution similar to human skin.

[0004] The visual tactile sensor, which is currently attracting more attention from many scholars and researchers, designs marking points on the surface of the soft elastic layer and captures the movement state of the marking points through a camera to infer the information of the contact force. The development of visual tactile sensors requires the use of visual imaging technology and data processing methods of micro cameras to increase the measurement dimension of tactile information. It has superior performance and has the ability to perceive complex tactile forces such as shear force and sliding force and the ability to recognize textures. However, its disadvantages are also very obvious, such as difficulty in fitting to different surface shapes and difficulty in miniaturization, and is not suitable for large-area paving. Summary of the invention

[0005] The purpose of the present invention is to propose a human skin-like super-resolution tactile sensor and a tactile reconstruction method to address the problems of insufficient spatial resolution of current tactile sensors, the reliance on increasing the density of sensor arrays to improve resolution, and the difficulty in miniaturization.

[0006] The object of the present invention is achieved through the following technical solutions: a human skin-like super-resolution tactile sensor, which is obtained by installing a soft silicone layer, a bottom bonding layer and a hard outer shell layer from top to bottom;

[0007] The soft silicone layer has a sensor array arranged in a spatial structure embedded therein, and the bottom bonding layer includes an array sensor unit, which is horizontally attached to the inner bottom surface of the hard outer shell layer;

[0008] The soft silicone layer embedded with the spatially structured sensor array is placed in a hard shell with horizontal array sensor units attached.

[0009] Furthermore, the spatial structural arrangement of the sensor array inside the soft silicone layer includes: an arrangement pattern with three-dimensional geometric features in the shape of a bionic four-leaf clover, petals or a ring;

[0010] The arrangement mode of the bionic four-leaf clover includes: arranging four flexible film pressure sensors in an inclined manner on four sides of a sensing point, and for each flexible film pressure sensor, the position close to the center of the sensing point is slightly higher and the position far from the center of the sensing point is slightly lower.

[0011] The arrangement pattern of the bionic petal shape includes: arranging eight flexible film pressure sensors around a sensing point respectively, wherein four of the flexible film pressure sensors are arranged in a bionic four-leaf clover pattern, and the other four are embedded in the four vertices of a rectangle formed by the four-leaf clover pattern, and the embedding method is: firstly, the direction of the flexible film is consistent with the direction surrounding the sensing point, and then tilted on this basis, and the tilt direction of a group of flexible films symmetrical about the center of the sensing point is consistent,

[0012] The circular ring arrangement mode includes: arranging eight flexible film pressure sensors around a sensing point in an inclined manner, with the flexible film pressure sensors spaced the same distance apart, and arranging them in a clockwise or counterclockwise circular ring shape.

[0013] Furthermore, the soft silicone layer has a thickness and a Shore hardness for an optimal force traction coupling effect, and by applying a normal force of a fixed magnitude to the soft silicone layer, the maximum distance that the pressure unit can generate is selected as an indicator.

[0014] On the other hand, the present specification also provides a tactile reconstruction method based on the sensor, the method comprising: normal force reconstruction through a horizontal array sensor of a bottom bonding layer, shear force reconstruction and torsional force reconstruction through a sensor array arranged in a spatial structure.

[0015] Furthermore, the normal force reconstruction includes: the normal force forms a binary Gaussian pressure distribution along the surface of the soft silicone layer:

[0016]

[0017] Where x and y represent the position of the tactile sensor surface in the coordinate plane, μ x and μ y represents the mean in the x and y directions, σ x and σ y represents the standard deviation in the x and y directions, and ρ represents the correlation between x and y;

[0018] Let σ x =σ y =σ, ρ = 0, the pressure distribution is simplified to the following form:

[0019]

[0020] From the above pressure distribution model, we can get: the normal force contact center coordinate is (μ x , μ y ), the normal pressure is When the normal force is applied to the surface of the soft silicone layer, due to the optimal force traction coupling effect, no less than two sensing units in the bottom pressure sensing layer will generate coupling response signals; the coupling response signal is read by the single chip microcomputer and sent to the computer via the serial port, and the pressure change curve is displayed in real time on the computer. According to the received coupling response signal, the optimal Gaussian distribution parameters are fitted using the least squares method, and the normal force size and position estimation problem is transformed into an optimization problem with the minimum error:

[0021]

[0022] Among them, according to the optimal parameters obtained by optimization, the contact position coordinates (μ x , μ y ) and plot the contact simulation effect at the same time.

[0023] Furthermore, the shear force reconstruction process includes shear force reconstruction of a bionic four-leaf clover arrangement array, shear force reconstruction of a bionic petal arrangement array, and shear force reconstruction of a ring shape arrangement. The shear force reconstruction modes of different sensor arrangement arrays are:

[0024] A shear force along a certain direction acts on the surface of the soft silicone layer, and the embedded sensor array generates a response signal. According to the fixed position and fixed angle of the pressure sensing unit that generates the response signal, a supervised learning method with artificial labeling is used, combined with a multi-layer perceptron to learn the nonlinear relationship between the electrical signal of the sensor array and the actual force. Through large-scale data collection, the regularity and similarity features are extracted, and the shear force reconstruction model is trained. The trained model is used to obtain the magnitude and direction of the shear force.

[0025] Furthermore, the torsional force reconstruction includes torsional force reconstruction of a bionic four-leaf clover arrangement array, torsional force reconstruction of a bionic petal arrangement array, and torsional force reconstruction of a ring shape arrangement. The shear force reconstruction modes of different sensor arrangement arrays are:

[0026] When a torsional force is applied to a certain point on the surface of the soft silicone layer, the torsional force will radiate to the surroundings through the soft silicone layer with the optimal force traction coupling effect, and the three-dimensional structural pressure sensor array of the internal embedded layer will generate a regular response signal; the supervised learning method with artificial labeling is combined with the multi-layer perceptron learning sensor array The nonlinear relationship between the electrical signal and the actual force, the regularity and similarity features are extracted through large-scale data collection, and the torsional force reconstruction model is trained. The trained model is used to obtain the magnitude and direction of the torsional force.

[0027] Furthermore, the mean square error is used as a loss function in the supervised learning method of manual labeling to measure the difference between the predicted value and the actual value.

[0028] Beneficial effects of the present invention:

[0029] 1. The present invention has a simple structure, small size, and light weight. It can be applied to grippers, manipulators, and the like for tactile perception during operation, thereby improving the safety and reliability of operation. The present invention can effectively respond to shear forces parallel to the contact surface through a specific spatial arrangement pattern, simulating the function of human skin soft tissue, surpassing the physical resolution capability of the sensor array, and realizing human skin-like super-resolution tactile perception. The present invention realizes the perception of multi-dimensional force information by combining different sensor arrangement patterns, and the method is simple and effective. The present invention can adjust the surface structure of the soft silicone layer according to the actual contact surface requirements to meet the needs of different occasions. The present invention can reproduce the state information of the contact force in real time in the visualization window, thereby improving the user's intuitive perception of the contact state. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a schematic diagram of the structure of a large-area, super-resolution tactile sensor similar to human skin proposed by the present invention;

[0031] Figure 2 The arrangement modes are the horizontal array, the bionic four-leaf clover shape, the bionic petal shape and the circular shape of the present invention;

[0032] Figure 3 It is a data collection curve diagram in a specific example of the present invention;

[0033] Figure 4 A visualization interface of the contact position center estimated by the tactile reconstruction algorithm of the present invention;

[0034] Figure 5 A contact force effect diagram estimated by the tactile reconstruction algorithm of the present invention;

[0035] Figure 6 The representative response curve when shear force is applied to the bionic four-leaf clover arrangement pattern;

[0036] Figure 7 The representative curves when shear force is applied to the bionic petal arrangement pattern;

[0037] Figure 8 This is a representative curve when a torsional force is applied to a circular arrangement pattern. DETAILED DESCRIPTION

[0038] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings.

[0039] The present invention draws on the bionic mechanism of embedding different types of receptors in soft tissues of human skin, and realizes the functional bionics of various human receptors (such as Pacinian corpuscles, Meissner corpuscles, etc.) by embedding sensor units of different types and different arrangement patterns in soft silicone materials. Then, through a tactile reconstruction algorithm based on physical models and machine learning, large-area and super-resolution perception of normal pressure, shear force, torsional force, etc. is achieved.

[0040] like Figure 1 As shown, the present invention provides a human skin-like super-resolution tactile sensor, which is obtained by installing a soft silicone layer 1, a pressure sensing layer 2 and a hard outer shell layer 3;

[0041] The pressure sensing layer 2 includes a bottom bonding layer 4 and an internal embedding layer 5 within the soft silicone layer.

[0042] The inner embedded layer 5 is a sensor array arranged in a spatial structure, and the bottom bonding layer 4 includes an array of sensor units, which are horizontally attached to the inner bottom surface of the hard outer shell layer 3;

[0043] The soft silicone layer embedded with the spatially structured sensor array is placed in a hard shell with horizontal array sensor units attached.

[0044] The spatial structural arrangement of the sensor array inside the soft silicone layer includes: an arrangement pattern with three-dimensional geometric features in the shape of a bionic four-leaf clover, a petal or a ring;

[0045] The arrangement mode of the bionic four-leaf clover includes: arranging four flexible film pressure sensors in an inclined manner on four sides of a sensing point, and for each flexible film pressure sensor, the position close to the center of the sensing point is slightly higher and the position far from the center of the sensing point is slightly lower.

[0046] The arrangement pattern of the bionic petal shape includes: arranging eight flexible film pressure sensors around a sensing point respectively, wherein four of the flexible film pressure sensors are arranged in a bionic four-leaf clover pattern, and the other four are embedded in the four vertices of a rectangle formed by the four-leaf clover pattern, and the embedding method is: firstly, the direction of the flexible film is consistent with the direction surrounding the sensing point, and then tilted on this basis, and the tilt direction of a group of flexible films symmetrical about the center of the sensing point is consistent,

[0047] The circular ring arrangement mode includes: arranging eight flexible film pressure sensors around a sensing point in an inclined manner, with the flexible film pressure sensors spaced the same distance apart, and arranging them in a clockwise or counterclockwise circular ring shape.

[0048] The soft silicone layer 1 is obtained by mixing food-grade silicone A and B available on the market in a mass or volume ratio of 1:1, stirring evenly, and then directly pouring into a mold for curing and forming. Among them, the pressure sensing unit of the internal embedding layer 5 is embedded in the liquid silicone according to the spatial structure arrangement before curing and forming. The shape of the soft silicone layer 1 can be designed and cast according to actual needs, for example, its shape can be designed to be the size structure of a given manipulator.

[0049] In order to use the soft silicone layer 1 with the best force traction coupling effect, the relevant parameters of the prepared soft silicone layer 1, including but not limited to thickness, Shore hardness, etc., need to be determined before the casting model is designed. For soft silicone layers 1 of different thicknesses and Shore hardnesses, a pressure sensing unit is placed at the bottom thereof, and a normal force of a fixed magnitude is applied to the soft silicone layer 1. The maximum distance that can make the pressure sensing unit respond is used as an evaluation index to select the thickness and Shore hardness of the soft silicone layer 1 with the best force traction coupling effect.

[0050] The bottom bonding layer 4 is composed of a 3×3 array of sensor units, which is directly horizontally attached to the inner bottom surface of the hard shell layer 3. The hard shell is made by 3D printing. Subsequently, the soft silicone layer 1, which is solidified and embedded with a sensor array arranged in a spatial structure, is placed in the hard shell with the horizontal array of sensor units attached, and each sensor unit is connected to the pressure conversion module by wired connection, and then connected to the single-chip microcomputer for data reading and processing.

[0051] After the hardware preparation process is completed, super-resolution and multi-dimensional force perception can be achieved through tactile reconstruction methods.

[0052] Based on the above-mentioned human skin-like super-resolution tactile sensor embodiment, the present invention also provides an embodiment of a tactile reconstruction method. The method realizes normal force reconstruction through the horizontal array sensor of the bottom bonding layer 4, and shear force reconstruction and torsional force reconstruction through the sensor array arranged in a spatial structure.

[0053] Part 1: Taking a 3×3 horizontal array as an example, the magnitude and position of the normal force are estimated through a binary Gaussian physical model.

[0054] The implementation process is as follows:

[0055] The normal force forms a binary Gaussian pressure distribution along the surface of the soft silicone layer 1:

[0056]

[0057] Where x and y represent the position of the tactile sensor surface in the coordinate plane, μ x and μ y represents the mean in the x and y directions, σ x and σ yrepresents the standard deviation in the x and y directions, and ρ represents the correlation between x and y. Based on the isotropy of silicone materials and to simplify the model, let σ x =σ y =σ, ρ = 0, the pressure distribution is simplified to the following form:

[0058]

[0059] From the above pressure distribution model, we can get: the normal force contact center coordinate is (μ x , μ y ), the normal pressure is When the normal force is applied to the surface of the soft silicone layer 1, the bottom bonding layer 4 will generate a coupled response signal due to the optimal force traction coupling effect, that is, for the same normal contact force, there are no less than two sensing units that generate response signals. The coupling signal is read by the microcontroller and sent to the computer via the serial port, where the pressure change curve is displayed in real time, such as Figure 3 As shown. The sensor design here uses a total of nine channels, and the data of these nine channels is transmitted in real time. Based on the received nine-channel data, the least squares method is used to fit the optimal Gaussian distribution parameters, so that the normal force size and position estimation problem is transformed into an optimization problem with the minimum error:

[0060]

[0061] in, The meaning of the parameter is the actual value of the normal force measured for the i-th time, P(x i ,y i ) is the estimated value of the normal force for the ith time, and Err(,) means the sum of squares of the errors between the actual value of the normal force and the estimated value. Based on the optimal parameters obtained through optimization, the contact position coordinates (μ x , μ y ), and draw the contact simulation effect at the same time, such as Figure 4 , 5 shown.

[0062] Part 2: Estimation of the magnitude and direction of shear force and torsional force through a sensor array with a spatially arranged structure.

[0063] The implementation process is as follows:

[0064] First, when a shear force in a certain direction acts on the surface of the soft silicone layer 1, the shear force will cause the soft silicone layer 1 to deform, and under the action of the traction coupling effect, the shear force will be transmitted to the pressure sensing unit of the internal embedded layer 5, thereby generating a response signal. Therefore, shear forces in different directions and sizes will cause pressure sensing units at different positions and angles to generate different but regular response signals. According to the fixed position and fixed angle of the pressure sensing unit that generates the response signal, combined with the multi-layer perceptron (MLP) method, the size and direction of the shear force can be obtained through large-scale data collection. For example, for the sensor array arranged in a bionic four-leaf clover, when a shear force is applied at a 45-degree angle to the upper right along the surface of the soft silicone layer 1, the following will be generated: Figure 6 Response curve shown; for the sensor array arranged in bionic petals, when a shear force is applied along the surface of the soft silicone layer 1 at an angle of 45 degrees to the upper right, the following will be generated Figure 7 The response curves shown. From the response curves of different arrangement modes, it can be found that for the same shear force, the generated response curve signals have obvious regularity, and the curves between the coupled response sensing units are similar. Using the supervised learning method with artificial labeling, through large-scale data collection, the regularity and similarity features can be extracted, thereby realizing the shear force detection along the entire surface of the tactile sensor.

[0065] Secondly, when a torsional force is applied to a certain point on the surface of the soft silicone layer 1, the torsional force will radiate to the surroundings through the soft silicone layer 1 with the best force traction coupling effect, so that the three-dimensional structure pressure sensor array of the internal embedded layer 5 generates a regular response signal. The sensor array arranged in a ring is more suitable for torsional force reconstruction, for example Figure 8 As shown in the figure, for the sensor array arranged in a ring, when a torsional force is applied at the lower left position of the center of the array, the sensor unit located near the contact center will respond, showing that the change trends of multiple response curves are basically the same, with different amplitudes and similarities. Similarly, the labeled data (response curve, sensor position, sensor arrangement angle, etc.) is used as input to obtain the output model of the tactile sensor through the multi-layer perceptron (MLP) method. In this way, the magnitude and position of the torsional force can be estimated.

[0066] It is very difficult to construct a physical model similar to the Gaussian distribution for the sensor array with a spatial arrangement structure. Therefore, according to the universal approximation theorem, a multilayer perceptron is chosen to obtain the mapping relationship between the sensor electrical signal and the actual force information. The multilayer perceptron is a typical feedforward artificial neural network, which consists of an input layer, several hidden layers, and an output layer. Each layer consists of several neurons, and the neurons are connected by weighted connections. MLP has powerful nonlinear modeling capabilities and can effectively handle complex mapping problems between input and output. The number of nodes in the input layer corresponds to the sampling data dimension of the sensor array, that is, time t i , the response value F of each sensor i , position(x i ,y i ) and angle θ i ; Two hidden layers are used, the first hidden layer includes 64 neurons, the second hidden layer includes 16 neurons, and the activation function uses the Sigmoid function; the output layer contains two dimensions of force information, namely the magnitude and direction of the force; the mean square error (MSE) is used as the loss function to measure the difference between the predicted value and the actual value. Through the above process, the MLP network can learn the complex nonlinear relationship between the electrical signal of the sensor array and the actual force, thereby providing accurate force estimation for the tactile sensor, significantly improving the accuracy and reliability of the sensor system, especially in complex multi-dimensional force coupling environments.

[0067] Those skilled in the art will readily appreciate other embodiments of the present application after considering the description and practicing the contents disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The description and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the claims.

[0068] It should be understood that the above general description and the detailed description below are only exemplary and explanatory and cannot limit the present application. The present application is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is limited only by the attached claims.

Claims

1. A human skin-like super-resolution tactile sensor, characterized in that: The sensor is installed from top to bottom by a soft silicone layer, a bottom bonding layer and a hard outer shell layer; The soft silicone layer has a sensor array arranged in a spatial structure embedded therein, and the bottom bonding layer includes an array sensor unit, which is horizontally attached to the inner bottom surface of the hard outer shell layer; The soft silicone layer embedded with the spatially structured sensor array is placed in a hard shell with horizontal array sensor units attached.

2. The human skin-like super-resolution tactile sensor according to claim 1, characterized in that: The spatial structural arrangement of the sensor array inside the soft silicone layer includes: an arrangement pattern with three-dimensional geometric features in the shape of a bionic four-leaf clover, a petal or a ring; The arrangement mode of the bionic four-leaf clover includes: arranging four flexible film pressure sensors in an inclined manner on four sides of a sensing point, and for each flexible film pressure sensor, the position close to the center of the sensing point is slightly higher and the position far from the center of the sensing point is slightly lower. The arrangement pattern of the bionic petal shape includes: arranging eight flexible film pressure sensors around a sensing point respectively, wherein four of the flexible film pressure sensors are arranged in a bionic four-leaf clover pattern, and the other four are embedded in the four vertices of a rectangle formed by the four-leaf clover pattern, and the embedding method is: firstly, the direction of the flexible film is consistent with the direction surrounding the sensing point, and then tilted on this basis, and the tilt direction of a group of flexible films symmetrical about the center of the sensing point is consistent, The circular ring arrangement mode includes: arranging eight flexible film pressure sensors around a sensing point in an inclined manner, with the flexible film pressure sensors spaced the same distance apart, and arranging them in a clockwise or counterclockwise circular ring shape.

3. The human skin-like super-resolution tactile sensor according to claim 1, characterized in that: The soft silicone layer has a thickness and a Shore hardness for an optimal force traction coupling effect, and is selected by applying a normal force of a fixed magnitude to the soft silicone layer, and the maximum distance that the pressure unit can produce is used as an indicator.

4. A tactile reconstruction method based on the sensor according to any one of claims 1 to 3, characterized in that: The method comprises: normal force reconstruction realized by a horizontal array sensor of a bottom bonding layer, shear force reconstruction and torsional force reconstruction realized by a sensor array arranged in a spatial structure.

5. The tactile reconstruction method according to claim 4, characterized in that: The normal force reconstruction includes: the normal force forms a binary Gaussian pressure distribution along the surface of the soft silicone layer: Where x and y represent the position of the tactile sensor surface in the coordinate plane, μ x and μ y represents the mean in the x and y directions, σ x and σ y represents the standard deviation in the x and y directions, and ρ represents the correlation between x and y; Let σ x =σ y =σ, ρ = 0, the pressure distribution is simplified to the following form: From the above pressure distribution model, we can get: the normal force contact center coordinate is (μ x , μ y ), the normal pressure is When the normal force is applied to the surface of the soft silicone layer, the optimal force traction coupling effect causes no less than two sensing units in the bottom bonding layer to generate a coupling response signal. The coupling response signal is read by the single-chip microcomputer and sent to the computer via the serial port. The pressure change curve is displayed in real time on the computer. Based on the received coupling response signal, the optimal Gaussian distribution parameters are fitted using the least squares method, and the normal force size and position estimation problem is transformed into an optimization problem with the minimum error: in, The meaning of the parameter is the actual value of the normal force measured for the i-th time, P(x i ,y i ) is the estimated value of the normal force for the ith time, Err(,) means the sum of squares of the errors between the actual value of the normal force and the estimated value. According to the optimal parameters obtained by optimization, the contact position coordinates (μ x , μ y ) and plot the contact simulation effect at the same time.

6. The tactile reconstruction method according to claim 4, characterized in that: The shear force reconstruction process includes shear force reconstruction of a bionic four-leaf clover arrangement array, shear force reconstruction of a bionic petal arrangement array, and shear force reconstruction of a ring shape arrangement. The shear force reconstruction modes of different sensor arrangement arrays are: A shear force along a certain direction acts on the surface of the soft silicone layer, and the embedded sensor array generates a response signal. According to the fixed position and fixed angle of the pressure sensing unit that generates the response signal, a supervised learning method with artificial labeling is used, combined with a multi-layer perceptron to learn the nonlinear relationship between the electrical signal of the sensor array and the actual force. Through large-scale data collection, the regularity and similarity features are extracted, and the shear force reconstruction model is trained. The trained model is used to obtain the magnitude and direction of the shear force.

7. The tactile reconstruction method according to claim 6, characterized in that: The torsional force reconstruction includes torsional force reconstruction of bionic four-leaf clover arrangement array, torsional force reconstruction of bionic petal arrangement array and torsional force reconstruction of ring shape arrangement. The shear force reconstruction modes of different sensor arrangement arrays are: When a torsional force is applied to a certain point on the surface of the soft silicone layer, the torsional force will radiate to the surroundings through the soft silicone layer with the optimal force traction coupling effect, and the three-dimensional structural pressure sensor array of the internal embedded layer will generate a regular response signal; the supervised learning method with artificial labeling is combined with the multi-layer perceptron learning sensor array The nonlinear relationship between the electrical signal and the actual force, the regularity and similarity features are extracted through large-scale data collection, and the torsional force reconstruction model is trained. The trained model is used to obtain the magnitude and direction of the torsional force.

8. The tactile reconstruction method according to claim 6, characterized in that: The human-labeled supervised learning method uses mean square error as the loss function to measure the difference between the predicted value and the actual value.

Images