A kind of human-like skin type super-resolution tactile sensor and tactile reconstruction method

Abstract

The application discloses a kind of human-like skin type super-resolution tactile sensor and tactile reconstruction method.The sensor draws lessons from the bionic mechanism that human skin embeds different types of receptors in soft tissue, realizes the functional bionics of various human receptors (such as Pacini corpuscle, Meissner corpuscle, etc.) by embedding different types and different arrangement modes of sensing units in soft silicone material, and then realizes large-area and super-resolution perception of forward pressure, shear force and torsional force, etc. through tactile reconstruction method. The characteristics of the proposed sensor are focused on the bionics of human skin tissue function rather than just the bionics of human receptor function, that is, by selecting soft silicone material with the largest force traction coupling effect, the same external contact can cause as many sensing units as possible to respond, thereby realizing human-like skin type large-area and super-resolution tactile perception.

Description

TECHNICAL FIELD

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

[0002] Tactile, as a parallel perception mode with vision and hearing, can provide robots with rich information about the contacted environment, such as the shape, surface texture, roughness, temperature and humidity of the contacted object, the position where the robot contacts the external object, and the size of the contact force, whether the robot finger slips with the external object during the grabbing process, and the like. Tactile information plays a crucial role in tasks such as robot grabbing and human-computer interaction, which require direct physical contact with the external environment.

[0003] At present, in order to obtain tactile information, various principles of tactile sensors have been developed in the field of robots, mainly including piezoresistive, capacitive, photoelectric, and visual tactile types. The essential principle of piezoresistive tactile sensor is based on the change of resistivity of single crystal silicon material when it is extruded by external force, which can be converted into pressure change through calibration technology; however, it cannot be flexibly attached to a large area, and cannot accurately measure shear force. The capacitive pressure sensor is usually composed of two parallel electrodes, and when the sensor receives external pressure, the deformation will cause the change of the capacitance value, and the pressure size can be sensed by detecting the change of the capacitance; however, it has poor anti-interference ability, and it is difficult to accurately measure shear force and torsional force. Piezoelectric sensor is based on piezoelectric effect, that is, when piezoelectric material is subjected to pressure, positive and negative charges will be generated on the two opposite surfaces, and the amount of generated charge is positively correlated with the pressure, and by detecting the voltage value of the two surfaces of the material, the change relationship between voltage and external pressure can be established; although the sensitivity of piezoelectric sensor is high, its manufacturing cost and complexity are also higher, it cannot detect static pressure, and vibration and high-frequency noise will have a great impact on it. The above tactile sensors are usually analogs of single receptor function in the skin, so in order to improve the spatial resolution performance, the traditional method is usually to reduce the size of the sensing unit or increase the density of the sensor array. However, the traditional method cannot exceed the physical resolution of the array, and cannot realize the human-like skin type infinite resolution.

[0004] The visual tactile sensor currently more concerned by many scholars and researchers is a sensor that designs mark points on the surface of a soft elastic layer and captures the motion state of the mark points through a camera to infer the information of contact force. The development of the visual tactile sensor needs to utilize the visual imaging technology of a miniature camera and a data processing method to increase the measurement dimension of tactile information, and the visual tactile sensor has superior performance, the ability to perceive complex tactile forces such as shear force and sliding force, and the ability of texture recognition. However, the visual tactile sensor also has obvious shortcomings, such as the difficulty in conforming to different surface shapes, the difficulty in miniaturization, and the unsuitability for large-area laying. SUMMARY

[0005] The present application aims to solve the problems of insufficient spatial resolution of the current tactile sensor, the way to improve the resolution depending on increasing the density of the sensing array, and the difficulty in miniaturization, and proposes a kind of human skin-like super-resolution tactile sensor and tactile reconstruction method.

[0006] The purpose of the present application is achieved by the following technical scheme: a kind of human skin-like super-resolution tactile sensor, which is obtained by installing from top to bottom a soft silicone layer, a bottom adhesive layer and a hard shell layer;

[0007] The soft silicone layer has a spatial structure arranged sensing array embedded therein, and the bottom adhesive layer includes an array type sensing unit horizontally attached to the inner bottom surface of the hard shell layer.

[0008] The soft silicone layer with a spatial structure arranged sensing array embedded therein is placed in the hard shell with the horizontally arrayed sensing unit attached.

[0009] Further, the spatial structure arrangement of the sensing array in the soft silicone layer includes a three-dimensional geometric feature arrangement mode in the form of bionic clover, petal or circular ring.

[0010] The bionic clover arrangement mode includes arranging four flexible thin film pressure sensors in a tilted manner at the four edges of a sensing point, and for each flexible thin film pressure sensor, the center of the sensing point is slightly higher, and the center of the sensing point is slightly lower.

[0011] The bionic petal arrangement mode includes arranging eight flexible thin film pressure sensors around a sensing point, four of which are arranged in a bionic clover mode, and the other four are embedded in the four corners of a rectangle formed by the four clover modes, and the embedding mode is: first, the direction of the flexible thin film is consistent with the direction around the sensing point, and then the flexible thin film is tilted on this basis, and the tilting direction of the flexible thin film symmetrically about the center of the sensing point is consistent.

[0012] The circular arrangement mode comprises: arranging eight flexible thin film pressure sensors around a sensing point in a tilted manner, and the flexible thin film pressure sensors are arranged at equal intervals and form a circular arrangement around the clock or counterclockwise.

[0013] Further, the soft silicone layer has a thickness and Shore hardness that have an optimal force traction coupling effect, and by applying a fixed normal force to the soft silicone layer, the pressure unit can generate a corresponding maximum distance as an index for selection.

[0014] On the other hand, the present specification also provides a haptic reconstruction method based on the sensor, which comprises: normal force reconstruction by the horizontal array sensor of the bottom adhesive layer, shear force reconstruction and torsional force reconstruction by the spatially arranged sensor array.

[0015] Further, the normal force reconstruction comprises: the normal force forms a bivariate Gaussian shape pressure distribution on the surface of the soft silicone layer.

[0016]

[0017] wherein x and y represent the position in the surface coordinate plane of the haptic sensor, μ x and μ y represent the mean value in the x and y directions, σ x and σ y represent 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, and ρ = 0, then the pressure distribution is simplified as follows:

[0019]

[0020] From the above pressure distribution model, it can be obtained that the normal force contact center coordinates are (μ x , μ y ), and the normal pressure size is When the normal force is applied to the surface of the soft silicone layer, due to the optimal force traction coupling effect, not less than two sensor units in the bottom pressure sensing layer will generate a coupling response signal; the single-chip microcomputer reads the coupling response signal and sends it to the computer end through the serial port, and the pressure change curve is displayed in real time on the computer end; according to the received coupling response signal, the optimal Gaussian distribution parameters are fitted by using the least square method, and the size and position estimation problem of the normal force is converted into an optimization problem with the minimum error.

[0021]

[0022] Wherein, according to the optimal parameters obtained by optimization, the contact position coordinates (mu x , mu y ) are displayed in real time through visual processing, and the contact simulation effect is drawn.

[0023] Further, 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 circular ring shape arrangement, and the shear force reconstruction modes of different sensor arrangement arrays are as follows:

[0024] The shear force acting in a certain direction on the surface of the soft silicone layer produces a response signal of the embedded sensor array, according to the fixed position and fixed angle of the pressure sensing unit producing the response signal, using a supervised learning method of artificial labeling, combining a multilayer perception machine to learn the nonlinear relationship between the sensor array electric signal and the actual force, extracting the regularity and similarity features from the large-scale data collection, training a shear force reconstruction model, and using the trained model to obtain the size and direction of the shear force.

[0025] Further, the torsion force reconstruction includes torsion force reconstruction of a bionic four-leaf clover arrangement array, torsion force reconstruction of a bionic petal arrangement array and torsion force reconstruction of a circular ring shape arrangement, and the shear force reconstruction modes of different sensor arrangement arrays are as follows:

[0026] When a torsion force is applied at a certain point on the surface of the soft silicone layer, the torsion force will radiate to the surrounding through the soft silicone layer with the best force traction coupling effect, and the three-dimensional structure pressure sensor array embedded in the internal layer will produce regular response signals; using a supervised learning method of artificial labeling, combining a multilayer perception machine to learn the nonlinear relationship between the sensor array electric signal and the actual force, extracting the regularity and similarity features from the large-scale data collection, training a torsion force reconstruction model, and using the trained model to obtain the size and direction of the torsion force.

[0027] Further, the supervised learning method of artificial labeling uses mean square error as a loss function to measure the difference between the predicted value and the actual value.

[0028] The beneficial effects of the present application are as follows:

[0029] 1.The present application has simple structure, small volume and light weight, and can be applied to the tactile perception of gripper, manipulator and other devices used in operation, thereby improving the safety and reliability of operation. The present application can effectively respond to the shear force parallel to the contact surface through a specific spatial arrangement mode, simulate the function of human skin soft tissue, and overcome the physical resolution capability of the sensor array, thereby realizing the super-resolution tactile perception of human-like skin. The present application can realize the perception of multi-dimensional force information by combining different sensor arrangement modes, which is simple and effective. The present application can adjust the surface structure of the soft silicone layer according to the actual contact surface requirements, and adapt to the requirements of different occasions. The present application can reproduce the state information of the contact force in the visual window in real time, thereby improving the intuitive perception of the user for the contact state. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 A structural schematic diagram of a human-like skin type large-area, super-resolution tactile sensor according to the present application is shown in the figure.

[0031] Figure 2 The arrangement modes of horizontal array, bionic four-leaf clover shape, bionic petal shape and circular shape according to the present application are shown in the figure.

[0032] Figure 3 A data acquisition curve diagram in a specific example of the present application is shown in the figure.

[0033] Figure 4 A visual interface of the contact position center estimated according to the tactile reconstruction algorithm of the present application is shown in the figure.

[0034] Figure 5 A contact force effect diagram estimated according to the tactile reconstruction algorithm of the present application is shown in the figure.

[0035] Figure 6 A representative response curve when a shear force is applied to the bionic four-leaf clover arrangement mode is shown in the figure.

[0036] Figure 7 A representative curve when a shear force is applied to the bionic petal arrangement mode is shown in the figure.

[0037] Figure 8 A representative curve when a torsional force is applied to the circular arrangement mode is shown in the figure. DETAILED DESCRIPTION

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

[0039] The present application learns from the bionic mechanism of human skin embedding different types of receptors in soft tissue, realizes the functional bionics of various human receptors (such as Pacini corpuscle, Meissner corpuscle, etc.) by embedding different types and different arrangement modes of sensing units in soft silicone material, and then realizes large-area and super-resolution perception of forward pressure, shear force and torsional force through a touch reconstruction algorithm based on a physical model and machine learning.

[0040] As shown in Figure 1 The present application provides a kind of human skin type's super-resolution tactile sensor, which is obtained by installing soft silicone layer 1, pressure sensing layer 2 and rigid shell layer 3;

[0041] The pressure sensing layer 2 includes the bottom adhering layer 4 and the internal embedded layer 5 in the soft silicone layer.

[0042] The internal embedded layer 5 is a spatially structured sensing array, and the bottom adhering layer 4 includes an array of sensing units horizontally attached to the inner bottom surface of the rigid shell layer 3.

[0043] The soft silicone layer embedded with the spatially structured sensing array is placed in the rigid shell attached with the horizontal array of sensing units.

[0044] The spatial structure of the sensing array in the soft silicone layer includes a bionic clover, petal shape or circular arrangement pattern with three-dimensional geometric features.

[0045] The bionic clover arrangement pattern includes arranging four flexible thin film pressure sensors in a tilted manner around the four edges of a sensing point, with each flexible thin film pressure sensor slightly higher near the center of the sensing point and slightly lower away from the center of the sensing point.

[0046] The bionic petal shape arrangement pattern includes arranging eight flexible thin film pressure sensors around a sensing point, with four arranged in a bionic clover pattern and the other four embedded in the four corners of a rectangle formed by the four clover patterns. The embedding method is to first align the flexible thin film direction with the direction around the sensing point, and then tilt on this basis. The tilting directions of the flexible thin films symmetrically about the center of the sensing point are consistent.

[0047] The circular arrangement pattern includes arranging eight flexible thin film pressure sensors in a tilted manner around a sensing point, with each flexible thin film pressure sensor spaced equally, and arranged in a circular shape clockwise or counterclockwise.

[0048] The soft silicone layer 1 is obtained by mixing food-grade silicone A and B according to a mass or volume ratio of 1:1, stirring uniformly, and then directly pouring into a mold for curing and forming. The pressure sensing unit embedded in the inner layer 5 is arranged and embedded in the space structure before the liquid silicone is cured and formed. The shape of the soft silicone layer 1 can be designed according to the actual requirements of the model pouring, for example, the shape can be designed as the size and structure of a given mechanical hand.

[0049] In order to use the soft silicone layer 1 with the best force traction coupling effect, the related parameters of the prepared soft silicone layer 1 need to be determined before the model design, including but not limited to thickness, Shore hardness, etc. For the soft silicone layer 1 with different thickness and Shore hardness, the pressure sensing unit is placed at the bottom of the soft silicone layer 1, and a fixed normal force is applied on the soft silicone layer 1, so that the maximum distance of the pressure sensing unit producing a response is used as an evaluation index, and the thickness and Shore hardness of the soft silicone layer 1 with the best force traction coupling effect are selected.

[0050] The bottom adhering layer 4 is composed of a 3x3 array of sensing units, which is directly horizontally attached to the inner bottom surface of the hard shell layer 3. The hard shell is 3D printed. Then, the soft silicone layer 1 which is cured and embedded with a spatially arranged sensing array is placed in the hard shell with horizontally arrayed sensing units, and each sensing unit is connected to the pressure conversion module through wired connection, and then connected to the single-chip microcomputer for data reading and processing.

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

[0052] Based on the aforementioned embodiment of the human-like skin type super-resolution tactile sensor, the present application also provides an embodiment of a tactile reconstruction method. The method realizes normal force reconstruction through the horizontal array sensor of the bottom adhering layer 4, shear force reconstruction and torsional force reconstruction through the spatially arranged sensing array.

[0053] First part: taking a 3x3 horizontal array as an example, the size and position estimation of the normal force is realized through a binary Gaussian physical model.

[0054] The implementation process is as follows:

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

[0056]

[0057] Where x and y represent the position in the coordinate plane of the surface of the tactile sensor, μ x and μ y represent the mean value in the x and y directions, σ x and σ ywhere σx and σy represent the standard deviation in x and y direction, and p represents the correlation between x and y. Based on the isotropy of the silicone material, and also for the sake of simplifying the model, let σx = σy = σ, and p = 0, then the pressure distribution is simplified as follows: x = σ y = σ, p = 0, then the pressure distribution is simplified as follows:

[0058]

[0059] From the above pressure distribution model, it can be obtained that the normal force contact center coordinates are (μ x , μ y ), and the normal pressure size is When the normal force is applied on the surface of the soft silicone layer 1, due to the best force traction coupling effect, the bottom adhesive layer 4 will generate a coupling response signal, that is, for the same normal contact force, there are not less than two sensing units to generate a response signal. The coupling signal is read by the single-chip microcomputer and sent to the computer end through the serial port, and the pressure change curve is displayed in real time on the computer end, as shown in Figure 3 Here, the sensor design uses a total of nine channels, and the data of the nine channels is transmitted in real time. According to the received nine channel data, and using the least square method to fit the best Gaussian distribution parameters, the size and position estimation problem of the normal force is converted into the optimization problem of the minimum error:

[0060]

[0061] wherein, 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 i-th normal force, and Err(·) represents the sum of squares of the error between the actual value and the estimated value of the normal force. According to the optimal parameters obtained by optimization, the contact position coordinates (μ x , μ y ) are displayed in real time through visual processing, and the contact simulation effect is drawn, as shown in Figure 4 , 5 .

[0062] The second part: through the spatial arrangement structure of the sensor array to realize the estimation of the size and direction of the shear force and the torsion force.

[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 of different directions and sizes will cause pressure sensing units distributed at different positions and angles to produce 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 magnitude 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 pattern, 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 As shown in the response curve; for the sensor array arranged in a bionic petal pattern, 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 7 The response curves shown in Figure 2 are from different arrangement patterns. The response curves for the same shear force exhibit clear regularity in the generated response curve signals, and the curves between the coupled response sensing units are similar. Using supervised learning methods with human labeling and large-scale data collection, these regularities and similarities can be extracted, enabling shear force detection across the entire tactile sensor surface.

[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 surrounding areas through the soft silicone layer 1 with the optimal force traction coupling effect, so that the three-dimensional pressure sensor array of the internal embedded layer 5 will produce 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 a circular sensor array, when a torsional force is applied to the lower left of the center of the array, the sensor units located near the contact center will respond. Multiple response curves show similar trends, but with varying amplitudes. Similarly, using labeled data (response curves, sensor positions, sensor arrangement angles, etc.) as input, a multi-layer perceptron (MLP) method is used to generate an output model for the tactile sensor. This allows the magnitude and location of the torsional force to be estimated.

[0066] It is very difficult to build a physical model similar to Gaussian distribution for the sensor array with spatial arrangement structure, so according to the universal approximation theorem, a multi-layer perceptron (MLP) is chosen to obtain the mapping relationship between the sensor electrical signal and the actual force information. MLP is a typical feedforward artificial neural network, which is composed of an input layer, several hidden layers and an output layer. Each layer is composed of several neurons, and the neurons are connected through weighted connections. MLP has strong non-linear modeling capability and can effectively handle the mapping problem between complex input and output. The number of input layer nodes corresponds to the sampling data dimension of the sensor array, i.e. time t i , the response value F i of each sensor, the position (x i , y i ) and angle θ i ; two hidden layers are used, the first hidden layer includes 64 neurons, and the second hidden layer includes 16 neurons, and the activation function uses the Sigmoid function; the output layer includes two dimensions of force information, i.e. the size and direction of 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 non-linear relationship between the sensor array electrical signal 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 environment.

[0067] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0068] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and are not restrictive of the application. The application is not limited to the precise construction described above and illustrated in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims

1. A human-like skin type super-resolution tactile sensor, characterized by, The sensor is installed by a soft silica gel layer from top to bottom, a bottom adhering layer and a hard shell layer; The soft silica gel layer is internally embedded with a spatial structure arranged sensor array, the bottom adhering layer includes arrayed sensor units and is horizontally adhered to the inner bottom surface of the hard shell layer; The spatial structure arrangement of the sensor array in the soft silica gel layer includes a three-dimensional geometric feature arrangement mode in the form of a bionic four-leaf clover, a bionic petal shape or a circular ring shape; The arrangement mode of the bionic four-leaf clover includes arranging four flexible thin film pressure sensors in a tilted manner at the four edges of a sensing point, and for each flexible thin film pressure sensor, the center of the sensing point is slightly higher and the center of the sensing point is slightly lower, The arrangement mode of the bionic petal shape includes arranging eight flexible thin film pressure sensors around a sensing point, four of which are arranged in a bionic four-leaf clover mode, and the other four are embedded in the four corners of a rectangle formed by the four-leaf clover mode, the embedding method is to first make the flexible thin film direction consistent with the direction around the sensing point, and then tilt on this basis, the tilt direction of the flexible thin film symmetrically arranged about the center of the sensing point is consistent, The arrangement mode of the circular ring shape includes arranging eight flexible thin film pressure sensors in a tilted manner around a sensing point, and the interval between each flexible thin film pressure sensor is the same, and the circular ring shape is arranged in a clockwise or counterclockwise direction. The soft silica gel layer embedded with the spatial structure arranged sensor array is placed in the hard shell with the horizontally arrayed sensor units adhered.

2. A human-like skin type ultrahigh-resolution tactile sensor according to claim 1, characterized in that, The soft silica gel layer has a thickness and Shore hardness that provides an optimal force traction coupling effect, and by applying a fixed normal force to the soft silica gel layer, the pressure unit can generate a corresponding maximum distance as an indicator for selection.

3. A method of haptic reconstruction based on the sensor of any of claims 1-2, characterized in that, The method includes normal force reconstruction by the horizontal array sensor of the bottom adhering layer, shear force reconstruction and torsional force reconstruction by the spatial structure arranged sensor array.

4. The haptic reconfiguration method of claim 3, wherein, The normal force reconstruction includes forming a binary Gaussian pressure distribution of the normal force along the surface of the soft silica gel layer: ； wherein, and represents the position of the tactile sensor surface in the coordinate plane, and denotes and the mean value in the direction, and denotes and the standard deviation in the direction, denotes and the correlation between. Let , Then the pressure distribution simplifies to the form: ； The normal force contact center coordinate is , and the normal pressure size is ; when the normal force is applied to the surface of the soft silicone layer, the optimal force traction coupling effect causes not less than two sensing units in the bottom adhesive layer to generate a coupling response signal; the single-chip microcomputer reads the coupling response signal and sends it to the computer end through the serial port, and the pressure change curve is displayed in real time on the computer end; according to the received coupling response signal, the best Gaussian distribution parameters are fitted by using the least square method, and the size and position estimation problem of the normal force is converted into an optimization problem with the minimum error: ； wherein, The meaning of the parameter is the actual value of the normal force of the i-th measurement, is the i-th normal force estimate value, the meaning of Err(, ) is the sum of squares of the error between the actual value and the estimated value of the normal force, according to the optimal parameter obtained by optimization, the contact position coordinates are displayed in real time through visual processing , and the contact simulation effect is drawn at the same time.

5. The tactile reconfiguration method of claim 3, wherein, The shear force reconstruction process includes shear force reconstruction of the bionic four-leaf clover arrangement array, shear force reconstruction of the bionic petal arrangement array and shear force reconstruction of the circular ring shape arrangement, and the shear force reconstruction mode of different sensor arrangement arrays is: The shear force in a certain direction acts on the surface of the soft silica gel layer, and the internally embedded sensor array generates a response signal, according to the fixed position and fixed angle of the pressure sensor unit generating the response signal, using a supervised learning method with artificial labeling, combining a multi-layer perception machine to learn the nonlinear relationship between the sensor array electric signal and the actual force, through large-scale data collection, the regularity and similarity features are extracted, a shear force reconstruction model is trained, and the size and direction of the shear force are obtained using the trained model.

6. The haptic reconfiguration method of claim 5, wherein, The torsional force reconstruction includes torsional force reconstruction of the bionic four-leaf clover arrangement array, torsional force reconstruction of the bionic petal arrangement array and torsional force reconstruction of the circular ring shape arrangement, and the shear force reconstruction mode of different sensor arrangement arrays is: When a torsion force is applied at a certain point on the surface of the soft silicone layer, the torsion force will radiate to the surrounding through the soft silicone layer with the best force traction coupling effect, and the three-dimensional structure pressure sensing array of the embedded layer generates regular response signals; using a human-labeled supervised learning method, combining a multi-layer perception machine to learn the nonlinear relationship between the sensor array electric signal and the actual force, through large-scale data collection, the regularity and similarity features are extracted, a torsion force reconstruction model is trained, and the size and direction of the torsion force are obtained using the trained model.

7. The tactile reconfiguration method of claim 5, wherein, In the human-labeled supervised learning method, the mean square error is used as a loss function to measure the difference between the predicted value and the actual value.

Images