Contact force detection method and magnetic haptic sensor

Abstract

This invention provides a contact force detection method and a magnetic tactile sensor, relating to the field of magnetic tactile sensors. The method is used in a magnetic tactile sensor comprising multiple magnets. The portion of the magnetic tactile sensor excluding the magnets is constructed of a deformable material. All magnets are placed on the same horizontal plane, and the magnetization direction of each magnet is opposite to that of its adjacent magnets, perpendicular to the horizontal plane. The method includes: obtaining position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet's detection point in a preset three-dimensional coordinate system and a contact force position classification model; obtaining magnitude parameters of the contact force based on the three-dimensional magnetic flux density change vector, position parameters, and a contact force magnitude regression model; and obtaining the contact force detection result based on the position and magnitude parameters. This invention's contact force detection method, by obtaining the position and magnitude of the contact force based on the contact force classification model and the contact force magnitude regression model, improves the accuracy of contact force detection.

Description

Technical Field

[0001] This invention relates to the field of magnetic tactile sensors, and more specifically, to a contact force detection method and a magnetic tactile sensor. Background Technology

[0002] When a magnetic tactile sensor detects contact force, it detects the change in the magnetic field caused by the contact force to determine the magnitude and location of the force. The greater the contact force, the greater its impact on the magnetic field, resulting in a larger change in the magnetic field and more accurate measurements of the force's magnitude and location. Conversely, smaller contact forces have less impact on the magnetic field, leading to smaller changes and less satisfactory results in detecting the force's magnitude and location.

[0003] In existing technologies, magnetic tactile sensors are used to detect small contact forces. However, since small contact forces have little impact on the magnetic field and the magnetic field changes little, the detection accuracy is insufficient, including the accuracy of detecting the size and position of the contact force. Summary of the Invention

[0004] The main technical problem addressed by this invention is how to improve the accuracy of contact force detection by a magnetic tactile sensor. To solve this problem, this invention provides a contact force detection method for a magnetic tactile sensor. The magnetic tactile sensor includes multiple magnets, the portion of which, excluding the magnets, is made of a deformable material. All magnets are placed on the same horizontal plane, and the magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The contact force detection method includes:

[0005] Based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model, the position parameters of the contact force are obtained.

[0006] The three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point along the three coordinate axes of the preset three-dimensional coordinate system.

[0007] The magnitude parameter of the contact force is obtained based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model.

[0008] The detection result of the contact force is obtained based on the position parameter and the magnitude parameter.

[0009] Optionally, it also includes:

[0010] The preset three-dimensional coordinate system is established based on the magnet, wherein the three-dimensional coordinate system includes an XY plane, and the XY plane is the horizontal plane;

[0011] The step of obtaining the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet at a detection point in a preset three-dimensional coordinate system and the contact force position classification model includes:

[0012] The three-dimensional magnetic flux density change vector is input into the contact force location classification model;

[0013] The position parameters of the contact force are obtained based on the output of the contact force position classification model.

[0014] Optionally, obtaining the magnitude parameter of the contact force based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model includes:

[0015] Based on the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system and the position parameter, a four-dimensional contact force vector is obtained;

[0016] The four-dimensional contact force vector is input into the contact force magnitude regression model to obtain the magnitude parameter of the contact force.

[0017] Optionally, before obtaining the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet's detection point in a preset three-dimensional coordinate system and the contact force position classification model, the method further includes:

[0018] Obtain the magnetic flux density at the detection point;

[0019] When the magnet is subjected to the contact force, the difference between the magnetic flux density of the detection point on each coordinate axis when the contact force is applied and the magnetic flux density of the detection point on each coordinate axis when the contact force is not applied is obtained, and the difference is used as the magnetic flux density change vector of the coordinate axis.

[0020] Optionally, obtaining the magnetic flux density at the detection point includes:

[0021] Obtain the regional magnetic flux density of each of the magnets at the detection point;

[0022] The magnetic flux density of the detection point is obtained based on the magnetic flux density of the region of each magnet at the detection point.

[0023] Optionally, obtaining the magnetic flux density at the detection point based on the magnetic flux density of each magnet in the region at the detection point includes:

[0024] The magnetic flux density at the detection point is obtained based on the magnetic flux density of each magnet in the region at the detection point and the superposition formula;

[0025] The superposition formula is:

[0026] Among them, B C The magnetic flux density at the detection point. The magnetic flux density of the region at the detection point for each of the magnets, where n is the number of the magnets.

[0027] Optionally, it also includes:

[0028] Select an initial classification model;

[0029] When the magnet is pressed by the contact force, multiple sets of data are obtained, including the three-dimensional magnetic flux density change vector of the detection point in the preset three-dimensional coordinate system, and the contact force position parameters of the contact force.

[0030] Each set of three-dimensional magnetic flux density change vectors is used as sample data for the initial classification model, and the position parameters of the contact force are used as sample labels for the initial classification model to train the initial classification model.

[0031] The trained initial classification model is used as the contact force location classification model.

[0032] Optionally, obtaining multiple sets of data, such as the three-dimensional magnetic flux density change vector of the detection point in the preset three-dimensional coordinate system and the contact force position parameters, when the magnet is pressed by the contact force, includes:

[0033] Based on the contact force, each magnet's corresponding area is pressed a preset number of times to obtain the initial magnetic flux density change of the detection point on each coordinate axis;

[0034] Based on the initial magnetic flux density change on each of the coordinate axes, the three-dimensional magnetic flux density change vector is obtained by eliminating a preset initial offset;

[0035] The contact force position parameters are obtained based on the corresponding region of each magnet.

[0036] Optionally, it also includes:

[0037] Select an initial regression model;

[0038] Multiple sets of three-dimensional magnetic flux density change vectors and contact force magnitude parameters of the detection point are obtained when the magnet is pressed by the contact force, wherein each set of three-dimensional magnetic flux density change vectors corresponds to a contact force position parameter and a contact force magnitude parameter;

[0039] The three-dimensional magnetic flux density change vector and the contact force position parameter are used as inputs to the initial regression model, and the contact force magnitude parameter is used as the output of the initial regression model to train the initial regression model.

[0040] The trained initial regression model is used as the regression model for the magnitude of the contact force.

[0041] The contact force detection method of the present invention inputs the magnetic flux density change vector detected by the detection point in the magnetic tactile sensor into the contact force position classification model, thereby obtaining the position parameters of the contact force according to the trained contact force classification model; then, according to the position parameters and the magnetic flux density change vector, the magnitude parameters of the contact force are obtained through the trained contact force magnitude regression model. This enables contact force detection with the same accuracy for any size of contact force according to the two models, thereby improving the accuracy of detecting smaller contact forces.

[0042] The present invention also provides a magnetic tactile sensor, which includes a plurality of magnets. The portion of the magnetic tactile sensor excluding the magnets is made of a deformable material. All the magnets are placed on the same horizontal plane. The magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The magnetic tactile sensor includes:

[0043] The contact force position detection unit is used to obtain the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model.

[0044] The three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point along the three coordinate axes of the preset three-dimensional coordinate system.

[0045] The contact force magnitude detection unit is used to obtain the magnitude parameter of the contact force based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model.

[0046] The processing unit is used to obtain the detection result of the contact force based on the position parameter and the size parameter.

[0047] The magnetic tactile sensor of the present invention inputs the magnetic flux density change vector detected by the detection point in the magnetic tactile sensor into the contact force position classification model, thereby obtaining the position parameters of the contact force according to the trained contact force classification model; then, according to the position parameters and the magnetic flux density change vector, the magnitude parameters of the contact force are obtained through the trained contact force magnitude regression model. This enables the detection of contact force with the same accuracy according to the two models for any size of contact force, thereby improving the accuracy of detecting smaller contact forces. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the magnetic tactile sensor structure in an embodiment of the present invention;

[0049] Figure 2 This is a flowchart of the contact force detection method in an embodiment of the present invention;

[0050] Figure 3 This is a flowchart of the contact force detection method in an embodiment of the present invention;

[0051] Figure 4 This is a flowchart of the contact force detection method in an embodiment of the present invention;

[0052] Figure 5 This is a flowchart of the contact force detection method in an embodiment of the present invention;

[0053] Figure 6 This is a flowchart of the contact force detection method in an embodiment of the present invention;

[0054] Figure 7 This is a schematic diagram of a magnetic tactile sensor unit in an embodiment of the present invention;

[0055] Figure 8 This is a schematic diagram of a magnetic tactile sensor unit in an embodiment of the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] The present invention discloses a contact force detection method for a magnetic tactile sensor. The magnetic tactile sensor includes multiple magnets, all placed on the same horizontal plane. The magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The portion of the magnetic tactile sensor excluding the magnets is made of a deformable material. Figure 1 As shown, with Figure 1Taking a magnetic tactile sensor as an example, the magnetic tactile sensor has four magnets, each of which is set in each square unit. The magnetic tactile sensor has a total of 4 units, each of which is a cuboid with a square cross-section. The magnetization directions of O1, O2, and O4 are opposite, while the magnetization directions of O1 and O3 are the same. The bottom surface of the magnetic tactile sensor is a horizontal plane. When the magnetic tactile sensor is placed on a horizontal plane, the symmetrical distribution of the magnets of the magnetic tactile sensor is achieved. The deformable material of the magnetic tactile sensor, except for the magnets, can be soft cast foam.

[0058] Combination Figure 2 As shown, the contact force detection method includes:

[0059] S1: Based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model, the position parameters of the contact force are obtained;

[0060] The three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point along the three coordinate axes of the preset three-dimensional coordinate system.

[0061] Specifically, the three-dimensional magnetic flux density change vector is the magnetic flux density change vector of the detection point along the three coordinate axes of a preset three-dimensional coordinate system. This change is detected at the detection point in the magnetic tactile sensor using a magnetic encoder or other magnetic flux density detection device. When a contact force is applied to a deformable material, the material deforms, thus affecting the magnetic flux density. The change vector of the magnetic flux density along the three coordinate axes of the preset three-dimensional coordinate system after the magnetic tactile sensor is affected by the contact force is measured. Based on the change vector of the magnetic flux density along the three coordinate axes, a contact force position classification model is used to obtain the position parameters of the contact force. These position parameters represent the area on the magnetic tactile sensor where the contact force acts; this area can be represented by a number. Figure 1 For example, the numbering can be based on the magnets in the magnetic tactile sensor, which are divided into 1, 2, 3, and 4 to represent the area corresponding to each magnet. The contact force position classification model is used as a classifier, which is obtained by training the K-nearest neighbor classification model to classify the position of the contact force.

[0062] S2: Based on the magnetic flux density change vector, the position parameters, and the contact force magnitude regression model, the magnitude parameter of the contact force is obtained;

[0063] Specifically, after obtaining the position parameters output by the above contact force position classification model, that is, after obtaining the corresponding position number and the change vector of magnetic flux density of the three coordinate axes measured at the detection point, the magnitude of the contact force is obtained through the contact force magnitude regression model based on the above two parameters. The contact force magnitude regression model is a regressor obtained by training a feedforward neural network to regress the contact force magnitude parameter and obtain a more accurate value.

[0064] S3: Obtain the detection result of the contact force based on the position parameter and the magnitude parameter;

[0065] Specifically, once the position and magnitude parameters of the contact force are obtained, the magnitude of the contact force and the location where the contact force acts can be determined. Once the magnitude and location of the contact force are obtained, the detection of the contact force can be completed.

[0066] The contact force detection method of the present invention inputs the magnetic flux density change vector detected by the detection point in the magnetic tactile sensor into the contact force position classification model, thereby obtaining the position parameters of the contact force according to the trained contact force classification model; then, according to the position parameters and the magnetic flux density change vector, the magnitude parameters of the contact force are obtained through the trained contact force magnitude regression model. This enables contact force detection with the same accuracy for any size of contact force according to the two models, thereby improving the accuracy of detecting smaller contact forces.

[0067] Combination Figure 3 As shown, in this embodiment of the invention, it further includes:

[0068] The preset three-dimensional coordinate system is established based on the magnet, wherein the three-dimensional coordinate system includes an XY plane, and the XY plane is the horizontal plane;

[0069] S1: Based on the three-dimensional magnetic flux density change vector of each magnet at a detection point in a preset three-dimensional coordinate system and the contact force position classification model, the position parameters of the contact force are obtained, including:

[0070] S11: Input the three-dimensional magnetic flux density change vector into the contact force location classification model;

[0071] S12: Obtain the position parameters of the contact force based on the output of the contact force position classification model.

[0072] In this embodiment, a three-dimensional coordinate system needs to be established first for each magnet in the magnetic tactile sensor. The XY plane is taken as the horizontal plane, and each magnet is placed on the horizontal plane. Since the magnetic tactile sensor has multiple magnets but only one detection point, the detection point is located at a different position in the three-dimensional coordinate system of each magnet. Based on the detection point, the change vector of magnetic flux density is detected, and the components of the change vector of magnetic flux density in the three axes of the preset three-dimensional coordinate system are obtained, namely the magnetic flux density change components, which include the components of the change vector of magnetic flux density in the X-axis, Y-axis and Z-axis respectively. The three components are input into the trained contact force position classification model to classify the contact force position and obtain the specific application position of the contact force, which is represented by the corresponding region number.

[0073] The contact force detection method of the present invention calculates the magnetic flux density change vector detected by the magnetic tactile sensor to obtain the component of each axis in the three-dimensional coordinate system. Since the magnetic flux density change in the Z-axis component has the most important overall response, the position of the contact force can be obtained by a classifier based on the change of the most important component. The detection accuracy of the contact force position is improved by using a trained classification model.

[0074] Combination Figure 4 As shown, in this embodiment of the invention, step S2: obtaining the magnitude parameter of the contact force based on the magnetic flux density change vector, the position parameter, and the contact force magnitude regression model, includes:

[0075] S21: Based on the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system and the position parameter, a four-dimensional contact force vector is obtained;

[0076] S22: Input the four-dimensional contact force vector into the contact force magnitude regression model to obtain the magnitude parameter of the contact force.

[0077] In this embodiment, after obtaining the components of the magnetic flux density change vector in three axes, it is combined with the contact force position parameters output by the contact force position classification model to form a four-dimensional vector including the components and the contact force corresponding region number. This four-dimensional vector is then input into the trained contact force magnitude regression model, which outputs the corresponding contact force magnitude parameters.

[0078] The contact force detection method of the present invention uses a trained feedforward neural network to regress the magnitude of the contact force based on the change in magnetic flux density and the position of the contact force. The contact force can then be detected based on the magnitude and position parameters.

[0079] In this embodiment of the invention, before obtaining the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet's detection point in a preset three-dimensional coordinate system and the contact force position classification model, the method further includes:

[0080] Obtain the magnetic flux density at the detection point;

[0081] When the magnet is subjected to the contact force, the difference between the magnetic flux density of the detection point on each coordinate axis when the contact force is applied and the magnetic flux density of the detection point on each coordinate axis when the contact force is not applied is obtained, and the difference is used as the magnetic flux density change vector of the coordinate axis.

[0082] In this embodiment, the magnetic flux density at the detection point can be obtained by a magnetic encoder. Since the magnetic flux density does not change when there is no contact force, the magnetic flux density changes when a contact force applies pressure to the magnet. The magnetic flux density change vector is obtained by the difference between the magnetic flux density before the contact force interference and the magnetic flux density after the change.

[0083] The contact force detection method of the present invention detects magnetic flux density using a magnetic flux density detection device and calculates the change in magnetic flux density, which facilitates the subsequent calculation of contact force based on the change.

[0084] In this embodiment of the invention, obtaining the magnetic flux density at the detection point includes:

[0085] Obtain the regional magnetic flux density of each of the magnets at the detection point;

[0086] The magnetic flux density of the detection point is obtained based on the magnetic flux density of the region of each magnet at the detection point.

[0087] In this embodiment, since each magnet will experience a change in magnetic flux density due to contact force, in order to obtain the overall magnetic flux density change vector of the detection point, it is first necessary to analyze each magnet to obtain the regional magnetic flux density generated by each magnet at the detection point, and then obtain the magnetic flux density of the detection point based on the regional magnetic flux density of all magnets at the same detection point.

[0088] The contact force detection method of the present invention calculates the magnetic flux density of each magnet at the detection point, thereby obtaining the value of the magnetic flux density detected at the actual detection point, which facilitates subsequent judgment of the change value of magnetic flux density.

[0089] In this embodiment of the invention, obtaining the magnetic flux density of the detection point based on the regional magnetic flux density of each magnet at the detection point includes:

[0090] The magnetic flux density at the detection point is obtained based on the magnetic flux density of each magnet in the region at the detection point and the superposition formula;

[0091] The superposition formula is:

[0092] Among them, B C The magnetic flux density at the detection point. The magnetic flux density of the region at the detection point for each of the magnets, where n is the number of the magnets.

[0093] In this embodiment, since a magnetic tactile sensor contains multiple magnets arranged on the same plane, it can... Figure 1Taking a magnetic tactile sensor as an example, the magnets are symmetrically distributed. Therefore, the magnetic flux density detected at the detection point can be determined by superimposing the magnetic flux density of each magnet at that point. This includes the superposition of the magnetic flux densities of the four magnets at that point. Therefore, the magnetic flux density at the detection point can be calculated using the superposition formula. Figure 1 Taking a magnetic tactile sensor as an example, if there are four magnets, then This represents the magnetic flux density of each magnet at the detection point.

[0094] The contact force detection method of the present invention calculates the magnetic flux density of each magnet at the detection point by superposition formula, and then obtains the value of the magnetic flux density detected at the actual detection point, which facilitates subsequent judgment of the change value of magnetic flux density.

[0095] Combination Figure 5 As shown, in embodiments of the present invention, it further includes:

[0096] S41: Select the initial classification model;

[0097] S42: Obtain multiple sets of magnetic flux density change vectors at the detection points in the preset three-dimensional coordinate system when the magnet is pressed by the contact force, wherein each set of magnetic flux density change vectors corresponds to a contact force position parameter;

[0098] S43: Use each group of three-dimensional magnetic flux density change vectors as sample data for the initial classification model, and use the position parameters of the contact force as sample labels for the initial classification model to train the initial classification model;

[0099] S44: Use the trained initial classification model as the contact force location classification model.

[0100] In this embodiment, before using the contact force location classification model for the first time, it needs to be trained to ensure that the model's output meets the requirements before actual contact force detection can be performed. Therefore, an initial classification model needs to be selected first. Each set of three-dimensional magnetic flux density change vectors is used as sample data for the initial classification model, and the contact force location parameters are used as sample labels. The initial classification model is trained by inputting sample data and outputting sample labels. The trained initial classification model is then used as the contact force location classification model. When the test results of the initial classification model show a classification accuracy of 99.5% or higher, the initial classification model is considered to have completed training.

[0101] The contact force detection method of the present invention obtains a contact force position classification model through an initial classification model, and then detects the position of any contact force applied to the magnetic tactile sensor. The classification model improves the accuracy of contact force position detection.

[0102] In this embodiment of the invention, obtaining multiple sets of three-dimensional magnetic flux density change vectors of the detection point in the preset three-dimensional coordinate system and contact force position parameters of the contact force when the magnet is pressed by the contact force includes:

[0103] Based on the contact force, each magnet's corresponding area is pressed a preset number of times to obtain the initial magnetic flux density change of the detection point on each coordinate axis;

[0104] Based on the initial magnetic flux density change on each of the coordinate axes, the three-dimensional magnetic flux density change vector is obtained by eliminating a preset initial offset;

[0105] The contact force position parameters are obtained based on the corresponding region of each magnet.

[0106] In this embodiment, the change in magnetic flux density is obtained by controlling the contact force to press the magnetic tactile sensor a preset number of times, and the components of the change in magnetic flux density on three coordinate axes are calculated. The change value of each axis component is standardized by subtracting a fixed preset offset to obtain the final component, which is then used as the input to the initial classification model.

[0107] In a preferred embodiment of the present invention, in terms of data processing, a low-pass infinite impulse response (IIR) filter is used to process the collected data, with the passband and stopband frequencies set to 3Hz and 10Hz, respectively, and the passband ripple and stopband attenuation set to 1dB and 60dB, respectively. The data filtered using the Chebyshev Type I method is cut by local peaks and minimums to form the loading and unloading datasets for each pressing cycle. The collected datasets are standardized according to the mean and standard deviation of the saturation data, and some cycle data that suddenly changed significantly due to accidents during the experiment are deleted and labeled as training data with the corresponding position categories. The resulting datasets are divided into training, testing, and validation datasets in a ratio of 70:15:15. In order to identify patterns with respect to different contact positions, a K-nearest neighbor classification model is trained using the Minkowski distance metric, where the number of nearest neighbors is set to 1.

[0108] In a preferred embodiment of the present invention, regarding data collection, combined with Figure 1Taking a magnetic tactile sensor as an example, the sensor was tested using a three-axis compression platform. The positions of the X and Y platforms were controlled by two micrometer manual linear platforms, while the Z-axis was controlled by a servo motor and a motion controller. A force sensor was mounted on the Z-platform, with a probe with a rounded tip connected below it. In some preferred embodiments, the force sensor could be an ATI-Nano17 six-axis force sensor. The prototype was placed on a laboratory jack and fixed to a base for locating the test positions corresponding to the four magnets. After detecting the horizontal contact position between the indenter and the magnetic tactile sensor, a pressing test was performed, with 20 cycles at each position. The loading speed was set to 0.2 mm / s, and the sampling rate was 100 Hz.

[0109] The contact force detection method of the present invention standardizes the magnetic flux density variation components of the three coordinate axes to obtain magnetic flux data with more regular changes, thereby increasing the success rate of model training and improving the training effect, while reducing the different stiffness distribution on the sensor caused by PCB integration.

[0110] Combination Figure 6 As shown, in this embodiment of the invention, it further includes:

[0111] S51: Select the initial regression model;

[0112] S52: Obtain multiple sets of three-dimensional magnetic flux density change vectors and contact force magnitude parameters of the detection point when the magnet is pressed by the contact force, wherein each set of three-dimensional magnetic flux density change vectors corresponds to a contact force position parameter and a contact force magnitude parameter;

[0113] S53: The initial regression model is trained by using the three-dimensional magnetic flux density change vector and the contact force position parameter as inputs to the initial regression model and the contact force magnitude parameter as outputs to the initial regression model.

[0114] S54: Use the trained initial regression model as the contact force magnitude regression model.

[0115] In this embodiment, similar to the steps for training the initial classification model described above, the components of the magnetic flux density change along the three axes and the contact force position parameters in the dataset used to train the initial classification model are input into the initial regression model for training. The magnitude of the contact force corresponding to the components along the three axes is obtained. After the model outputs the contact force magnitude parameters, the regression correlation coefficient of the initial classification model is measured. When the regression correlation coefficient of the initial classification model reaches 0.99 or above, it proves that the model training is complete. If the deviation is greater than or equal to the preset result threshold, it means that the model output has not met the training standard and the model needs to be trained again based on the dataset until the deviation is less than the preset result threshold, thus completing the training.

[0116] The contact force detection method of the present invention obtains the magnitude of the contact force by training a feedforward neural network based on the location of the contact force and the change in magnetic flux density. This enables the contact force to be accurately measured even when the change in magnetic flux density is small, thereby improving the accuracy of contact force detection.

[0117] After training, in order to evaluate the static performance of the magnetic tactile sensor, Figure 1 Taking the magnetic tactile sensor as an example, the pressing test was performed again on the areas corresponding to the four magnets for 20 cycles. The standardized magnetic flux density change data was collected and input into the trained contact force position classification model (K-nearest neighbor classification model) and contact force magnitude regression model (feedforward neural network) to obtain the contact force position and force measurement value for further evaluation. In the evaluation, the full-scale output, hysteresis error, repeatability error and static error of the magnetic tactile sensor were obtained. Based on the above errors, it can be seen that the positioning accuracy of the magnetic tactile sensor for contact force has increased, which effectively improves the detection accuracy of contact force.

[0118] Combination Figure 7 As shown, the present invention also provides a magnetic tactile sensor 100, which includes a plurality of magnets. The portion of the magnetic tactile sensor excluding the magnets is made of a deformable material. All the magnets are placed on the same horizontal plane. The magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The magnetic tactile sensor 100 includes:

[0119] The contact force position detection unit 130 is used to obtain the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model; wherein, the three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system;

[0120] The contact force magnitude detection unit 140 is used to obtain the magnitude parameter of the contact force based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model.

[0121] The processing unit 150 is used to obtain the detection result of the contact force based on the position parameter and the size parameter.

[0122] Combination Figure 8 As shown, the magnetic tactile sensor 100 also includes a model training unit 110 and a magnetic flux density detection unit 120;

[0123] The contact force position detection unit 130 is also used to establish the preset three-dimensional coordinate system based on the magnet, wherein the three-dimensional coordinate system includes an XY plane, the XY plane is the horizontal plane, input the three-dimensional magnetic flux density change vector into the contact force position classification model, and obtain the position parameters of the contact force based on the output of the contact force position classification model.

[0124] The contact force magnitude detection unit 140 is further configured to obtain a four-dimensional contact force vector based on the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system and the position parameter; and input the four-dimensional contact force vector into the contact force magnitude regression model to obtain the magnitude parameter of the contact force.

[0125] The magnetic flux density detection unit 120 is used to acquire the magnetic flux density of the detection point; when the magnet is subjected to the contact force, the difference between the magnetic flux density of the detection point on each coordinate axis when the contact force is applied and the magnetic flux density of the detection point on each coordinate axis when the contact force is not applied is obtained, and the difference is used as the magnetic flux density change vector of the coordinate axis; acquiring the magnetic flux density of the detection point includes: acquiring the regional magnetic flux density of each magnet at the detection point; obtaining the magnetic flux density of the detection point based on the regional magnetic flux density of each magnet at the detection point; obtaining the magnetic flux density of the detection point based on the regional magnetic flux density of each magnet at the detection point includes: obtaining the magnetic flux density of the detection point based on the regional magnetic flux density of each magnet at the detection point and a superposition formula; the superposition formula is: Among them, B C The magnetic flux density at the detection point. The magnetic flux density of the region at the detection point for each of the magnets, where n is the number of the magnets.

[0126] The model training unit 110 is used to select an initial classification model; acquire multiple sets of three-dimensional magnetic flux density change vectors of the detection point in the preset three-dimensional coordinate system and contact force position parameters of the contact force when the magnet is pressed by the contact force; use each set of three-dimensional magnetic flux density change vectors as sample data of the initial classification model and the contact force position parameters as sample labels of the initial classification model to train the initial classification model; and use the trained initial classification model as the contact force position classification model.

[0127] The step of obtaining multiple sets of data on the three-dimensional magnetic flux density change vector of the detection point in the preset three-dimensional coordinate system and the contact force position parameters of the contact force when the magnet is pressed by the contact force includes: obtaining the initial magnetic flux density change of the detection point on each coordinate axis based on the corresponding area of ​​each magnet being pressed by the contact force a preset number of times; obtaining the three-dimensional magnetic flux density change vector by eliminating a preset initial offset based on the initial magnetic flux density change on each coordinate axis; and obtaining the contact force position parameters of the contact force based on the corresponding area of ​​each magnet.

[0128] Select an initial regression model; obtain multiple sets of three-dimensional magnetic flux density change vectors and contact force magnitude parameters at the detection point when the magnet is pressed by the contact force, wherein each set of three-dimensional magnetic flux density change vectors corresponds to a contact force position parameter and a contact force magnitude parameter; use the three-dimensional magnetic flux density change vectors and contact force position parameters as inputs to the initial regression model, and use the contact force magnitude parameters as outputs to train the initial regression model; use the trained initial regression model as the contact force magnitude regression model.

[0129] The magnetic tactile sensor of the present invention inputs the magnetic flux density change vector detected by the detection point in the magnetic tactile sensor into the contact force position classification model, thereby obtaining the position parameters of the contact force according to the trained contact force classification model; then, according to the position parameters and the magnetic flux density change vector, the magnitude parameters of the contact force are obtained through the trained contact force magnitude regression model. This enables the detection of contact force with the same accuracy according to the two models for any size of contact force, thereby improving the accuracy of detecting smaller contact forces.

[0130] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0131] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for detecting contact force, characterized in that, The contact force detection method is used for a magnetic tactile sensor, which includes multiple magnets. The portion of the magnetic tactile sensor excluding the magnets is made of a deformable material. All magnets are placed on the same horizontal plane. The magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The contact force detection method includes: Based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model, the position parameters of the contact force are obtained. The preset three-dimensional coordinate system is established based on the magnet, wherein the three-dimensional coordinate system includes an XY plane, and the XY plane is the horizontal plane; The three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point along the three coordinate axes of the preset three-dimensional coordinate system. The magnitude parameter of the contact force is obtained based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model; specifically, this includes: obtaining a four-dimensional contact force vector based on the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system and the position parameter; and inputting the four-dimensional contact force vector into the contact force magnitude regression model to obtain the magnitude parameter of the contact force. The detection result of the contact force is obtained based on the position parameter and the magnitude parameter.

2. The contact force detection method according to claim 1, characterized in that, The step of obtaining the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet at a detection point in a preset three-dimensional coordinate system and the contact force position classification model includes: The three-dimensional magnetic flux density change vector is input into the contact force location classification model; The position parameters of the contact force are obtained based on the output of the contact force position classification model.

3. The contact force detection method according to claim 1, characterized in that, Before obtaining the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet's detection point in a preset three-dimensional coordinate system and the contact force position classification model, the method further includes: Obtain the magnetic flux density at the detection point; When the magnet is subjected to the contact force, the difference between the magnetic flux density of the detection point on each coordinate axis when the contact force is applied and the magnetic flux density of the detection point on each coordinate axis when the contact force is not applied is obtained, and the difference is used as the magnetic flux density change vector of the coordinate axis.

4. The contact force detection method according to claim 3, characterized in that, The process of obtaining the magnetic flux density at the detection point includes: Obtain the regional magnetic flux density of each of the magnets at the detection point; The magnetic flux density of the detection point is obtained based on the magnetic flux density of the region of each magnet at the detection point.

5. The contact force detection method according to claim 4, characterized in that, The step of obtaining the magnetic flux density at the detection point based on the magnetic flux density of each magnet in the region at the detection point includes: The magnetic flux density at the detection point is obtained based on the magnetic flux density of each magnet in the region at the detection point and the superposition formula; The superposition formula is: ; in, The magnetic flux density at the detection point. The magnetic flux density of each magnet in the region at the detection point. The number of magnets.

6. The contact force detection method according to any one of claims 1 to 5, characterized in that, Also includes: Select an initial classification model; Multiple sets of data are obtained when the magnet is pressed by the contact force, to obtain the three-dimensional magnetic flux density change vector of the detection point in the preset three-dimensional coordinate system, and the contact force position parameters of the contact force; Each set of three-dimensional magnetic flux density change vectors is used as sample data for the initial classification model, and the position parameters of the contact force are used as sample labels for the initial classification model to train the initial classification model. The trained initial classification model is used as the contact force location classification model.

7. The contact force detection method according to claim 6, characterized in that, The acquisition of multiple sets of three-dimensional magnetic flux density change vectors of the detection point in the preset three-dimensional coordinate system, and contact force position parameters of the contact force when the magnet is pressed by the contact force, includes: Based on the contact force, each magnet's corresponding area is pressed a preset number of times to obtain the initial magnetic flux density change of the detection point on each coordinate axis; Based on the initial magnetic flux density change on each of the coordinate axes, the three-dimensional magnetic flux density change vector is obtained by eliminating a preset initial offset; The contact force position parameters are obtained based on the corresponding region of each magnet.

8. The contact force detection method according to claim 1, characterized in that, Also includes: Select an initial regression model; Multiple sets of three-dimensional magnetic flux density change vectors and contact force magnitude parameters of the detection point are obtained when the magnet is pressed by the contact force, wherein each set of three-dimensional magnetic flux density change vectors corresponds to a contact force position parameter and a contact force magnitude parameter; The three-dimensional magnetic flux density change vector and the contact force position parameter are used as inputs to the initial regression model, and the contact force magnitude parameter is used as the output of the initial regression model to train the initial regression model. The trained initial regression model is used as the regression model for the magnitude of the contact force.

9. A magnetic tactile sensor, characterized in that, The magnetic tactile sensor includes multiple magnets. The portion of the magnetic tactile sensor excluding the magnets is made of a deformable material. All magnets are placed on the same horizontal plane. The magnetization direction of each magnet is opposite to that of its adjacent magnets, and the magnetization direction of each magnet is perpendicular to the horizontal plane. The magnetic tactile sensor includes: The contact force position detection unit is used to obtain the position parameters of the contact force based on the three-dimensional magnetic flux density change vector of each magnet at the detection point in the preset three-dimensional coordinate system and the contact force position classification model; and to establish the preset three-dimensional coordinate system based on the magnet, wherein the three-dimensional coordinate system includes the XY plane, and the XY plane is the horizontal plane; The three-dimensional magnetic flux density change vector includes the magnetic flux density change vector of the detection point along the three coordinate axes of the preset three-dimensional coordinate system. The contact force magnitude detection unit is used to obtain the magnitude parameter of the contact force based on the three-dimensional magnetic flux density change vector, the position parameter, and the contact force magnitude regression model; specifically, it includes: obtaining a four-dimensional contact force vector based on the magnetic flux density change vector of the detection point on the three coordinate axes of the preset three-dimensional coordinate system and the position parameter; and inputting the four-dimensional contact force vector into the contact force magnitude regression model to obtain the magnitude parameter of the contact force. The processing unit is used to obtain the detection result of the contact force based on the position parameter and the size parameter.

Images