A material recognition method, device, computer device, and storage medium

Abstract

An embodiment of the present application discloses a material identification method, which includes: obtaining touch information for a target material, where the touch information is data obtained after touching the target material; determining a target material feature of the target material according to the touch information, and the target material feature includes a material roughness feature and a material stiffness feature; among multiple material features included in a feature library, determining a matching material feature that matches the target material feature, and using the material type corresponding to the matching material feature as the material type of the target material. By adopting the embodiment of the present application, the uncertainty of material identification can be reduced and the identification accuracy can be improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a method and device for material recognition, a computer device, and a storage medium. Background Art

[0002] With the continuous development of technology, the recognition of the material of an object has gradually become a popular research topic. In current automated recognition, most of the methods for recognizing the material on the surface of an object apply computer vision technology. For example, an image recognition method is used to extract visual information through a camera to recognize the material on the surface of the object to be measured. However, since various materials may be visually similar but have different structures, it is difficult to distinguish similar materials through this method, and the recognition process is easily affected by external factors such as light, which may lead to inaccurate recognition of material properties, resulting in low recognition accuracy. Summary of the Invention

[0003] In view of the above problems, the embodiments of this application provide a method and device for material recognition, a computer device, and a storage medium, which can achieve higher material recognition accuracy.

[0004] One aspect of the embodiments of this application provides a method for material recognition, including:

[0005] Obtaining touch information for a target material, where the touch information is data obtained after touching the target material;

[0006] Determining a target material feature of the target material according to the touch information, where the target material feature includes a material roughness feature and a material stiffness feature;

[0007] Among multiple material features included in a feature library, determining a matching material feature that matches the target material feature, and using the material type corresponding to the matching material feature as the material type of the target material.

[0008] One aspect of the embodiments of this application provides a device for material recognition, including:

[0009] An obtaining device, configured to obtain touch information for a target material, where the touch information is data obtained after touching the target material;

[0010] A determining module, configured to determine a target material feature of the target material according to the touch information, where the target material feature includes a material roughness feature and a material stiffness feature;

[0011] A matching module, configured to determine a matching material feature that matches the target material feature among multiple material features included in a feature library, and using the material type corresponding to the matching material feature as the material type of the target material.

[0012] One aspect of the embodiments of the present application provides a computer device, including: a network interface, a processor, and a memory. The network interface, the processor, and the memory are connected to each other. The network interface is used to provide data communication functions, the memory is used to store computer programs, and the processor is used to call the computer programs to execute some or all of the steps described in one aspect of the embodiments of the present application.

[0013] One aspect of the embodiments of the present application provides a storage medium that stores a computer program. The computer program includes program instructions, which are loaded and executed by one or more of the processors to execute the material identification method described in one aspect of the embodiments of the present application.

[0014] One aspect of the embodiments of the present application provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the material identification method described in one aspect.

[0015] It can be seen that in the embodiments of the present application, the material type of the material to be identified is determined through touch information. Compared with determining the material type through vision, it is not affected by light, is less interfered by external factors, the identification process is more stable, and the material identification accuracy is improved to a certain extent. Moreover, due to the characteristic that the relationships between various forces at different contact positions during touch are different, the feedback information on the surface of the touched material is collected, and the characteristics of two dimensions, namely the material roughness characteristic and the material stiffness characteristic, are extracted. Since both roughness and stiffness are inherent properties of the material on the surface of the material, the identification accuracy of the material type can be further ensured based on these two characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic flowchart of the steps of a material identification method provided by the embodiments of the present application;

[0018] Figure 2a It is a schematic structural diagram of a sensor provided by the embodiments of the present application;

[0019] Figure 2bIt is a schematic structural diagram of a tactile feedback system provided by an embodiment of the present application;

[0020] Figure 3 It is a schematic step - flow diagram of a material recognition method provided by an embodiment of the present application;

[0021] Figure 4 It is a schematic diagram of a curved - surface model of a sensor housing provided by an embodiment of the present application;

[0022] Figure 5 It is a schematic step - flow diagram of a material recognition method provided by an embodiment of the present application;

[0023] Figure 6 It is a schematic step - flow diagram of an anthropomorphic touch method provided by an embodiment of the present application;

[0024] Figure 7 It is a schematic step - flow diagram of a touch recognition method provided by an embodiment of the present application;

[0025] Figure 8 It is a schematic structural diagram of a material recognition device provided by an embodiment of the present application;

[0026] Figure 9 It is a schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0027] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0028] Artificial Intelligence (AI) is to use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, a theory, method, technology, and application system that perceives the environment, acquires knowledge, and uses knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling the machines to have the functions of perception, reasoning, and decision - making.

[0029] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0030] The solution provided by the embodiments of this application relates to sensors in the basic technologies of artificial intelligence, which can convert the sensed information into the required form according to certain rules and meet the requirements of information transmission, processing, storage, control, etc. Touch data of materials is collected through sensors, and the material characteristics of different contact surfaces are extracted to achieve the function of material identification. Automatic detection and automatic control are realized in this process, and it can be widely applied to modern industrial production, especially in the field of automatic material classification, to improve the production efficiency of industry.

[0031] Please refer to Figure 1 , which is a schematic flow chart of the steps of a material identification method provided by the embodiments of this application. The method includes:

[0032] S101, obtain touch information for a target material, where the touch information is data obtained after touching the target material.

[0033] In a possible embodiment, the target material is an unknown material corresponding to the material to be identified. For example, the target material can be an object made of a metal material, such as copper or iron, or other small precision devices, or materials such as rubber or wood, which are not limited here. The touch information is data obtained after the touch device comes into contact with the surface of the target material. A tactile sensor is installed in the touch device here, such as a six-axis force-torque sensor, as Figure 2a shown, Figure 2a is a schematic structural diagram of the sensor. Its structure includes a hemispherical sensor housing 1, a sensor strain gauge 2, and a sensor base 3. Optionally, as Figure 2bAs shown in the figure, in addition to a tactile sensor, the touch device further includes a robotic arm, a host computer, and relevant software components to form a tactile feedback system. In summary, the overall structure of the touch device includes a sensor housing 4, a sensor 5, a robotic arm 6, and a host computer 7, which can form a tactile feedback system. The original measurement values obtained by the tactile sensor when touching the target material are fed back to the host computer or the processor. Here, the host computer or the processor can be a computer connected to the tactile sensor via wired or wireless means. Through the host computer program interface, the control of the touch device, the real-time monitoring and storage of touch information, and the functional debugging of the entire tactile feedback system can be achieved. The robotic arm maintaining the movement of the tactile sensor on the contact surface with the material can be regarded as anthropomorphic touch, and touch information can be obtained during anthropomorphic touch. Specifically, since the sensor housing and the object come into contact and cause slight deformation, and the sensor housing is fixed to the sensor by screws, the sensor can also sense the stress from the housing. When the housing undergoes slight deformation, it presses the strain gauge on the sensor, and the force is converted into a differential signal and sent to the host computer as the data detected by the sensor, such as force and torque measurement values. These measurement values are used as touch information for further processing. Of course, there can be multiple options for the specific method of obtaining sensor data or the devices used depending on different situations, which are not restricted here.

[0034] S102. Determine the target material characteristics of the target material according to the touch information. The target material characteristics include material roughness characteristics and material stiffness characteristics.

[0035] In a possible embodiment, by setting the data acquisition frequency of the touch device, it is possible to obtain multiple sets of touch information when it contacts the target material. Based on these multiple sets of touch information, the processor can obtain the data required for calculating the material roughness characteristics and material stiffness characteristics. These data are the relevant data of the force applied to the surface of the target material when contacting it. According to these data, the target material characteristics can be obtained. The target material characteristics are characteristics related to the material of the contact surface of the target material, which can represent the material to which the contact surface belongs, including stiffness and roughness. Among them, stiffness is the ability of a material, component, or structure under external force to resist deformation. From the perspective of tactile perception, when touching an object, one can feel the resistance of the object to mechanical deformations such as compression and bending. Roughness refers to the surface roughness of the material, which is an expression of the three-dimensional topography characteristics of the surface. Compared with a single attribute, using two inherent attributes of the material can better distinguish the material type and more comprehensively identify the material to which the contact surface of the target material belongs.

[0036] S103. Among the multiple material characteristics included in the feature library, determine the matching material characteristic that matches the target material characteristic, and use the material type corresponding to the matching material characteristic as the material type of the target material.

[0037] In a possible embodiment, material features corresponding to all material types are stored in a feature library. These material features are represented by two-dimensional data, namely data in two dimensions of roughness and stiffness. The specific storage method is not limited. Each material type corresponds to a two-dimensional data, which is used to match the target material features obtained after touching the target material. Optionally, this two-dimensional data can be regarded as a point in a Cartesian coordinate system. The specific matching method can be to determine the feature distance between the target material features and each material feature in the feature library, that is, the square of the Euclidean distance between the target material features and the material features in the feature library (the square of the straight-line distance between two points in the Cartesian coordinate system). Therefore, there are multiple feature distances corresponding to multiple material features in the feature library and the target material features. Then, select the material feature with the smallest feature distance as the matching material feature. Through this matching material feature, the type of the material to be identified can be known. The specific expression for calculating the Euclidean distance is as follows:

[0038]

[0039] For example, for an unknown material on the surface of a target material, there are target material features (1, 2). In the feature library, the material features corresponding to material A are (2, 3), the material features corresponding to material B are (2, 1), and the material features corresponding to material C are (1, 2). The squares of the Euclidean distances between the target material features and each of these material features are calculated to be 2, 2, and 0 respectively. The minimum value among them is 0, corresponding to material C. Therefore, it can be determined that the contact surface material of the target material belongs to material C.

[0040] In summary, the embodiments of the present application have at least the following advantages:

[0041] By touching the surface of the target material and moving on the contact surface, a preliminary process is performed on multiple groups of touch information collected, and then the material roughness feature and the material stiffness feature are extracted as the target material features. The target material features are compared with the material features in the feature library to identify the material type. Compared with using a single feature to distinguish the material type, multiple features are comprehensively compared to distinguish the material type, reducing the uncertainty of recognition and improving the accuracy of identifying the material type.

[0042] Please refer to Figure 3 , which is a schematic flowchart of the steps of a material recognition method provided by an embodiment of the present application. The method includes:

[0043] S301, obtain touch information for the target material, where the touch information is data obtained after touching the target material.

[0044] For the specific implementation manner of this step, reference can be made to S101 in the corresponding embodiment above, and details will not be elaborated here. Figure 1 ​

[0045] S302. The number of the touch information is N, where N is a positive integer. N contact point data of the target material are determined based on the N touch information.

[0046] After touching the target material, N touch information can be obtained according to the acquisition frequency. First, the contact point coordinates are solved using the touch information, and then the normal force is obtained. The tangential force is solved using the relationship between the force components. The contact point data can include the above three data. The required data can be selected during calculation to obtain the characteristics of the target material. The first part of the N touch information is used to calculate the stiffness characteristic, and the second part is used to calculate the roughness characteristic. The tangential force is not required when calculating the stiffness characteristic, and the contact point coordinates are not required when calculating the roughness characteristic. To save storage space, the contact point data corresponding to the part of the touch information used to calculate the stiffness characteristic may not include the tangential force, and the contact point data corresponding to the part of the touch information used to calculate the roughness characteristic may not include the contact point coordinates. Using the normal force of the first contact point data that meets the threshold range as the dividing line, the N contact point data are divided into n stiffness contact point data and m roughness contact point data, where n and m are both positive integers. Each stiffness contact point data includes the contact point coordinates and the normal force, and each roughness contact point data includes the normal force and the tangential force. The N contact point data are obtained in chronological order and have corresponding generation timestamps. The generation timestamp of any roughness contact point data is greater than that of any stiffness contact point data. That is to say, the time to obtain the roughness contact point data is after the time to obtain the stiffness contact point data. Among the n stiffness contact point data obtained here, according to the generation timestamp, the normal forces of the first n - 1 stiffness contact point data are not within the threshold range, and only the normal force of the nth stiffness contact point data, that is, the target stiffness contact point data, is within the threshold range. The normal forces of the m roughness contact point data are also within the threshold range.

[0047] Optionally, for all N contact point data, they can be solved under the static friction model using the tactile information (i.e., contact point data) solution equation derived from the force relationship between the housing and the sensor. The expression of the tactile information solution equation is as follows:

[0048]

[0049] where x, y, and z are the coordinates of the contact point in the coordinate system centered on the sensor strain gauge (i.e., contact point coordinates), f x , f y , f z , m x , m y , m z are the measured values of the sensor (i.e., touch information), and S(x, y, z) is the surface equation of the sensor housing.

[0050] Under the static friction model, it is known that f = [f x , f y , f z , m = [m x , m y , m z , and the following relationships exist with the unknown data:

[0051] f = p

[0052] The relationship between the force and the moment is:

[0053] m = q + c × p

[0054] where c is the position vector from the center of the strain gauge to the contact point, and q is the moment at the contact point.

[0055] To obtain the solution of this non - linear equation system, an algorithm in the non - linear least - squares method can be used. For example, the Levenberg - Marquardt algorithm is used to iterate the solution of the equation. The LM algorithm can be used to find the fitting result that matches the non - linear regression model, minimizing the sum of the squares of the residuals of the original model during the iteration to solve the solution of the non - linear equation system. The specific expression is as follows:

[0056] x k+1 = x k -(J T J + μI) -1 g k

[0057] where X k is the solution vector obtained at the k - th time, J is the Jacobian matrix of the equation, (J T J + μI) -1 is the search step size, and g k is the descent direction at the k - th time. Finally, when g(k + 1)=g(k) is less than a given threshold, the iteration ends.

[0058] According to the decomposition of the force, the force acting on the surface of the target material can be decomposed into the tangential force and the normal force. The specific formula for solving the normal force is:

[0059]

[0060] where p is the three - dimensional force vector measured by the strain gauge, and n is the normal vector corresponding to the surface equation.

[0061] Then, according to the relationship between the three - dimensional force vector and the normal force, the specific expression for solving the tangential force is:

[0062] F t = p - F n

[0063] In the test, the tactile information obtained after calculating the six - dimensional force and torque data measured by the sensor can be used to establish a surface model of the sensor housing. As Figure 4 shown, the contact point coordinates (x1, y1) and the tangential force F t1 and the normal force F n1 of this contact point can be obtained on this surface model. Visualizing these data can more intuitively understand the data of the material type of the target material.

[0064] S303. Determine the stiffness characteristics based on the contact point coordinates of the starting stiffness contact point data, the contact point coordinates of the target stiffness contact point data, and the normal force.

[0065] The starting stiffness contact point data here is the first stiffness contact point data among n stiffness contact point data, that is, the stiffness contact point data with the smallest generated timestamp. During the process of touching the material, this starting stiffness contact point data can be regarded as the contact point data obtained from the measurement value with the critical contact position feedback. As the contact state with the material surface changes continuously, when fully contacting the object, n groups of data can be obtained. Using the coordinate change between contact points and the normal force of the target stiffness contact point, the stiffness characteristics of the target material can be solved. Specifically, the coordinate change between contact points here refers to the normal displacement, that is, the change value between the contact point coordinates of the starting stiffness contact point data and the contact point coordinates of the target stiffness contact point data. Dividing the normal force of the target stiffness contact point data by the normal displacement can obtain the stiffness characteristics.

[0066] For example, if the contact point coordinates of the starting stiffness contact point data are (1, 2), the contact point coordinates of the target stiffness contact point data are (1, 4), the coordinate unit is millimeter, and the normal force is 5N, then the stiffness characteristic calculation is 5N divided by the normal displacement of 2mm, and the final result is 2.5 Newtons per millimeter.

[0067] S304. Determine the roughness characteristics based on the normal force and tangential force of m roughness contact point data.

[0068] The data of m roughness contact points is obtained after the data of target stiffness contact points. After obtaining the data of target stiffness contact points, the robotic arm drives the sensor to move in the tangential force direction, generating a tangential displacement. Moreover, the subsequent adjustment of the normal displacement cannot ensure that it starts from the critical contact position. If the stiffness characteristic data is continuously calculated, there will be a large difference, resulting in inaccurate final recognition results. Therefore, after the data of target stiffness contact points appears, the normal force in the subsequent contact point data is also collected as the force feedback for control and does not participate in the calculation of the normal displacement. The contact point data with the normal force within the threshold range in the contact point data is selected as the roughness contact point data. Since the actual object is not completely smooth, the actual roughness can be represented by the coefficient of kinetic friction, that is, the ratio of the tangential force to the normal force, to obtain the unit roughness characteristic. The m unit roughness characteristics are averaged, and the average value is taken as the roughness characteristic of the target material.

[0069] S305. Combine the stiffness characteristic and the roughness characteristic into the target material characteristic of the target material.

[0070] In a possible embodiment, performing a single touch operation on the target material (a single touch operation can be considered as the contact sensor starts to contact the target material until the contact sensor leaves the target material) will generate a corresponding target material characteristic. This target material characteristic is obtained by combining the data of the stiffness characteristic and the roughness characteristic belonging to the contact surface of the target material. The roughness characteristic is the average value of m unit roughness characteristics. Compared with a single characteristic, the combination of the two characteristics can enrich the dimension of the material characteristics and reduce the misjudgment rate in the recognition process.

[0071] S306. Among the multiple material characteristics included in the feature library, determine the matching material characteristic that matches the target material characteristic, and use the material type corresponding to the matching material characteristic as the material type of the target material.

[0072] For the specific implementation manner of this step, reference can be made to S103 in the corresponding embodiment above, which will not be elaborated here. Figure 1 Corresponding to S103 in the embodiment, it will not be repeated here.

[0073] In summary, the embodiments of the present application have at least the following advantages:

[0074] By touching the surface of the target material and moving on the contact surface, multiple groups of touch information are collected and preliminarily processed to obtain contact point data. The least squares method is used to solve the unknown contact point data, making the error between the contact point data and the actual data the smallest. Using these contact point data to extract the stiffness characteristic and roughness characteristic of the target material is more accurate, and then matching the minimum characteristic distance with the material characteristics in the feature library to identify the material type, improving the accuracy of identifying the material type, and further developing the significance of tactile detection for the recognition of the contact surface material.

[0075] Please refer to Figure 5 , which is a schematic flowchart of steps of another material recognition method provided by an embodiment of this application. The method includes:

[0076] S501. Obtain K sets of sample contact point data.

[0077] Before recognizing the material type of the contact surface of the target material, it is necessary to collect all the sample data corresponding to the existing material types for further processing for subsequent recognition, that is, it is necessary to construct a feature library. Therefore, it is necessary to obtain the sample contact point data corresponding to the material types of all materials to form multiple sets of sample contact point data. Optionally, if there are K types of materials, then there will be K sets of sample contact point data. Here, K is a positive integer. Each set of sample contact point data corresponds to a material type, and each set of sample contact point data includes multiple unit sample contact point data sets. Each unit sample contact point data set includes multiple sample contact point data. Each sample contact point data includes contact point coordinates, normal force, and tangential force. Among them, a unit sample contact point data set is obtained by performing a touch operation on a sample material once. Each touch operation can be controlled within a short time, and this short time is at the second level. A large amount of sample contact point data can be obtained within this short time.

[0078] In some possible embodiments, for any one of the multiple unit sample contact point data sets, the process of obtaining any one of the unit sample contact point data sets includes:

[0079] The initial touch information is obtained, and the sample initial contact point data corresponding to the initial touch information is determined. The sample initial contact point data includes the initial normal force. The initial touch information is the data obtained after the touch device touches the sample material, which can be regarded as the touch information at the critical contact position with the material. Optionally, the sensor in the touch device can be a six-axis force-torque sensor, such as the force / torque sensor ATI nano 17 force / torque sensor (Calibration SI-25-025 resolution: 1 / 160 N for Fx Fy Fz 1 / 32 Nmm for MxMy Mz Range FxFy=25 N Fz=35N Mx My Mz=250 Nmm) suitable for narrow research application space. The six-axis force-torque sensor contacts the target material with the movement of the robot arm. The Dobot Magician robot arm can be used to drive the tactile sensor, which has 4 axes, a valid load of 500g, a maximum extension distance of 320mm, and a repeatability accuracy of 0.2mm. The host computer or processor collects the raw data measured by the sensor, that is, the initial touch information, and uses the nonlinear least squares method to solve to obtain the starting coordinates, starting normal force and starting tangential force of the starting contact point to form the sample starting contact point data. The data here does not include the noise data generated during the acquisition process.

[0080] Determine the first movement instruction according to the starting normal force, and obtain the first movement touch information. The first movement touch information here is the data obtained after the touch device moves and touches according to the first movement instruction. Optionally, the first movement instruction is the control instruction issued by the host computer or processor to the robot arm after the aforementioned sample starting contact point data is input into the robot arm control algorithm. The control instruction may include information such as the displacement, speed, and direction of the robot arm to move. Specifically, the robot arm control algorithm will determine whether the starting normal force and the preset normal force deviation are within the range. If not within the range, a first movement instruction will be issued to adjust the contact position of the touch device touching the sample material. Generally speaking, in order to protect the touch device, the force of the touch device contacting the sample material at the beginning will not exceed its maximum bearing range, but a smaller normal force is set to contact the sample material, and then gradually adjusted to the corresponding normal force range. For example, if the initial normal force is smaller than the minimum value of the normal force threshold range, the control instruction will instruct the touch device to appropriately increase the normal force and move the touch device closer to the contact surface of the sample material along the direction of the initial normal force at a certain speed, and then obtain the measurement value fed back by the touch device, that is, the first mobile touch information.

[0081] Determine the first moving contact point data corresponding to the first moving touch information. According to the first moving touch information, the first moving contact point data can be determined in the same way as solving the target contact point data using the touch information solution equation in the foregoing embodiments, which will not be elaborated here. The obtained first moving contact point data includes a moving normal force and a moving tangential force, and the moving normal force therein is used to determine the next first moving instruction. The moving direction included in the next first moving instruction is determined by the moving normal force of the current first moving contact point data. If the moving normal force is not within the threshold range, the moving instruction instructs the touch device to move along the direction of the moving normal force. In this way, the moving normal force in each newly input first moving contact point data is judged in a loop to obtain the latest control instruction, so that the touch device moves continuously, collects more touch information, and obtains multiple first moving contact point data suitable for stiffness analysis.

[0082] When the moving normal force of the first moving contact point data is within the threshold range, both the sample starting contact point data and all the first moving contact point data during the movement are used as sample stiffness contact point data. So far, the first moving contact point data here previously included at least one first moving contact point data and a starting contact point data where the moving normal force was not within the threshold range. Because when the moving normal force of the first moving contact point data is not within the threshold range, a new first moving instruction will be determined according to the moving normal force, determining the new first moving contact point data between the touch device and the sample material, so there will be multiple first moving instructions to adjust the contact position, and finally the touch device and the sample material reach a fully contact state.

[0083] After the normal force of the obtained first moving contact point data is within the threshold range, a second movement instruction is determined according to the moving tangential force of the last first moving contact point data, and then the sample roughness contact point data is determined according to the second movement instruction. After the first first moving contact point data that meets the normal force threshold range is obtained, the touch device moves along the tangential direction of the last first moving contact point data under the control of the second movement instruction, generating a tangential displacement. After the movement, second movement touch information can be obtained, that is, the second movement touch information is the data obtained after the touch device moves and touches according to the second movement instruction. There is corresponding second to-be-determined moving contact point data according to the second movement touch information. This second to-be-determined moving contact point data includes coordinates, moving tangential force, and moving normal force. If the moving normal force of the second to-be-determined moving contact point data is within the threshold range, this second to-be-determined moving contact point data is used as the second moving contact point data, and the next second movement instruction is determined according to the moving tangential force of this second moving contact point data. Based on the next second movement instruction, the next second to-be-determined moving contact point data is determined, and then it is further determined whether the next second to-be-determined moving contact point data is to be used as the second moving contact point data.

[0084] If the moving normal force of the second to-be-determined moving contact point data is not within the threshold range, the second to-be-determined moving contact point data is not used as the second moving contact point data. The next second movement instruction is determined according to the moving normal force of this second to-be-determined moving contact point data. Based on the next second movement instruction, the next second to-be-determined moving contact point data is determined, and then it is further determined whether the next second to-be-determined moving contact point data is to be used as the second moving contact point data.

[0085] When the touch device meets the stop condition, the obtained second moving contact point data is used as the sample roughness contact point data. Optionally, the stop condition can be the displacement distance limit of the end of the robotic arm. When the coordinates of the end of the robotic arm exceed the threshold, the operation stops. It can also set the length of the data acquisition time, which is not limited here.

[0086] Combine the sample roughness contact point data and the sample stiffness contact point data into any unit sample contact point data set, which has multiple sample roughness contact point data and multiple sample stiffness contact point data. Thus, a unit sample contact point data set is determined, and multiple unit sample contact point data sets can be determined in the same way. Because in the calculation of the stiffness feature, it is only related to the normal force and the normal displacement. Except for adjusting the normal force from the critical contact position to the full contact position at the start of touching, there may also be a small adjustment of the normal force to the threshold range during the operation process, but the points adjusted during this process cannot be guaranteed to start from the critical contact position, and there will be a large difference in the calculation of the stiffness. Therefore, during the subsequent touching process, even if there is a normal displacement, this normal displacement will no longer be used to calculate the stiffness. Instead, the moving normal force and the moving tangential force that meet the threshold range are used as the sample roughness contact point data. The above data acquisition process can be understood as measuring the physical properties of the material of the object contact surface by simulating the tactile mechanical model of the human finger through the sensor, and adopting different control strategies at different stages.

[0087] S502. Determine the sample material features corresponding to each unit sample contact point data set respectively, and cluster all the sample material features into K feature clusters, and the K feature clusters correspond to K feature cluster centers.

[0088] Each unit sample contact point data set contains multiple sample contact point data. Using the coordinates, normal force, and tangential force in the multiple sample contact point data of a unit sample contact point data set, a sample material feature can be obtained, that is, the roughness feature and the stiffness feature of the sample material. The specific method is similar to that in the foregoing embodiment for determining the target material feature using the N contact point data of the target material, and will not be elaborated here. Correspondingly, each sample contact point data set has multiple sample material features. For example, each sample contact point data set corresponds to b sample material features, and the b sample material features are obtained by the touch device touching the sample material anthropomorphically b times. If the contact surfaces of K types of materials in the sample material are each touched anthropomorphically b times, then all the sample material features are K×b sample material features included in the K sample contact point data sets. Cluster these sample material features into K feature clusters corresponding to K types of materials. Optionally, the clustering method can be the K-means algorithm. Cluster all the sample contact point data through the K-means algorithm to obtain the clustering center points of the K feature clusters, and these clustering center points will be used in the steps of matching the target material features mentioned above.

[0089] S503. Establish a feature library according to the K feature cluster centers and the material types corresponding to each feature cluster center.

[0090] The material corresponding to each feature cluster center is the material corresponding to most of the feature data in the feature cluster. K feature cluster centers and their corresponding material types constitute the data in the feature library and are unique. For example, if the feature cluster center corresponding to material A is (k1, l1), the feature cluster center corresponding to material B is (k2, l2)..., and so on, feature libraries of different materials can be constructed, and each feature cluster center is different and uniquely corresponds to one material. On the premise that the material of the surface of the sample material is known, constructing a feature library can better assist in the identification of unknown materials. The construction method can be selected according to the actual situation. Software methods (such as database technology design) or electronic hardware implementation, there is no restriction here.

[0091] In summary, the embodiments of the present application have at least the following advantages:

[0092] By touching the sample material multiple times with a touch device, sufficient sample contact point data is collected, and the roughness and stiffness characteristics of the sample material are extracted. Sufficient data preparation is made for establishing a feature library of all materials. Among them, the selection and calculation of data are tried to avoid data with large errors, thereby improving the accuracy of the data in the material type feature library and the credibility of the material identification results.

[0093] See also Figure 6 , is a schematic diagram of the steps of an anthropomorphic touch method provided in an embodiment of the present application. The anthropomorphic touch method in this embodiment can summarize the content of obtaining target contact point data or sample contact point data in the aforementioned embodiment. The implementation process of the anthropomorphic touch method is specifically explained below in conjunction with the aforementioned embodiment. The specific steps can be divided into:

[0094] S601, the sensor part contacts the contact surface of different materials with the movement of the robot arm. The sensor part has the same structure as that mentioned in the above embodiment, see Figure 2a Optionally, the sensor can be a six-axis force-torque sensor. Driven by the robotic arm, the six-axis force-torque sensor contacts the material. Different contact surfaces may be made of different materials. The purpose of contacting contact surfaces of different materials here can be to collect sample touch information of known materials or to collect touch information of unknown materials.

[0095] S602. The host computer or the processor collects sensor data. When the sensing part moves with the robotic arm and comes into contact with the outside world, the six-axis force-torque sensor sends the measured values of force and torque, that is, the sensor data, corresponding to the aforementioned touch information, to the host computer. Specifically, when the sensor housing comes into contact with an object, a slight deformation occurs during the contact. Since the housing is fixed to the sensor by screws, the sensor can also sense the stress from the housing. When the housing undergoes a slight deformation and presses the strain gauge on the sensor, the force is converted into a differential signal, that is, the sensor data. This differential signal is a signal that can be recognized and processed by the host computer or the processor, corresponding to the touch information mentioned in the foregoing embodiments.

[0096] S603. Use the non-linear least squares method to solve the current contact information. Here, the contact information corresponds to the aforementioned sensor data, that is, the measured values of force and torque, and is also the touch information. Using the host computer software to solve the touch information obtained by the six-axis force-torque sensor with the non-linear least squares method, the contact point coordinates, the normal force and shear force (i.e., tangential force) at the contact position and other tactile information (i.e., contact point data) can be obtained. The specific solution process is the same as that for solving the contact point data in the above embodiments.

[0097] S604. Input the solved tactile information into the robotic arm control algorithm. The robotic arm control algorithm can be a PID algorithm that combines the three links of proportion, integral, and differential, or other control algorithms, which are not limited herein. Among them, the control algorithm mainly processes the normal force in the tactile information to control the robotic arm.

[0098] S605. The host computer or the processor sends a control instruction to the robotic arm. By using the control algorithm to detect whether the normal force in the tactile information is within the threshold range, corresponding control instructions can be obtained, including content such as direction, displacement, and speed. The host computer sends this control instruction to the robotic arm to adjust the position of the sensor housing and the contact surface of the material.

[0099] S606. The robotic arm drives the sensing part to maintain contact with the contact surface and move on the surface. The robotic arm drives the sensor to come into contact with the surface of the material to be measured. The touch information returned by the tactile sensor is input into the host computer. The host computer controls the robotic arm to drive the sensor to move on the contact surface to obtain touch information. In addition, the robotic arm can also drive the sensor to come into contact with and move on the contact surface of a known material to obtain touch information to determine the sample contact point data. By performing multiple anthropomorphic touches on the known material, sufficient training data can be obtained.

[0100] In summary, the embodiments of the present application have at least the following advantages:

[0101] By touching the surfaces of different materials with a touch device, sensor information of different materials is fed back to obtain test data of unknown materials or a sufficient amount of sample data, which makes necessary data preparations for material identification. Anthropomorphic touch simulates the perception of human fingers to sense more visual or information that humans cannot feel, and can more accurately perform tactile detection on the materials of different contact surfaces.

[0102] Please refer to Figure 7 , which is a schematic flow chart of the steps of a touch recognition method provided by an embodiment of the present application. The touch recognition method in this embodiment can summarize the content of identifying the material type in the foregoing embodiment. The following specifically explains the implementation process of the touch recognition method in combination with the foregoing embodiment. The specific steps can be divided into:

[0103] S701, the tactile sensing part collects touch information of anthropomorphic touch on contact surfaces of different materials. Using the tactile sensor feedback system as shown in Figure 2b , anthropomorphic touch can be performed on the contact surface of the same material multiple times. A large amount of touch information can be formed by different material contact surfaces, and the host computer stores and processes the large amount of touch information.

[0104] S702, use the K-means algorithm to perform clustering analysis on the touch data to extract features. Touch data can be calculated based on the touch information, which is also the sample contact point data mentioned in the foregoing embodiment. Based on multiple anthropomorphic touches, a large amount of sample contact point data is obtained as sample data. The host computer performs clustering analysis on the sample data and extracts various data features of touches on materials, including roughness features and stiffness features. Here, the clustering algorithm, that is, the training program uses the K-means algorithm to analyze the relationship between the data obtained in different tests, and depicts the similarity between the data through the minimum squared error of the Euclidean distance between the data to achieve the classification effect.

[0105] S703, establish a feature library of tactile data corresponding to different material contact surfaces. According to the results of the clustering analysis, a feature library is established that includes different materials and data features obtained from tactile data (including roughness features and stiffness features). The construction and storage of the feature library can be carried out in various ways, which are not limited here. The surface of an object may be composed of different material combinations. Constructing a feature library for the materials of the contact surface, compared with constructing a feature library for the object, extends the significance of tactile detection to a certain extent.

[0106] S704, send the data into the feature library for matching to complete the recognition during the anthropomorphic touch test. Perform anthropomorphic touch on the material of the contact surface to be tested, compare the test data (that is, the stiffness feature and roughness feature obtained from the target contact point data) with the data in the feature library, and identify the material of the contact surface in the test through the minimum squared error of the Euclidean distance between the data. This step can also be used by the host computer to restore information such as the material and contour of the object through an algorithm.

[0107] In summary, the embodiments of the present application have at least the following advantages:

[0108] By using a clustering algorithm to perform clustering analysis on all sample data to construct a feature library corresponding to different materials, and according to the data detected by anthropomorphic touch, sending it into the feature library for matching can realize the identification of materials. Fusing the roughness feature and the stiffness feature for data screening, classification, sorting and compression can achieve real-time processing of data, enhance the recognition ability, and make the results more accurate and reliable.

[0109] Please refer to Figure 8 , which is a schematic structural diagram of a material identification device provided by the embodiments of the present application. The device includes:

[0110] An acquisition module 801, configured to acquire touch information for a target material, where the touch information is data obtained after touching the target material;

[0111] A determination module 802, configured to determine target material features of the target material according to the touch information, where the target material features include a material roughness feature and a material stiffness feature;

[0112] A matching module 803, configured to determine a matching material feature that matches the target material feature among multiple material features included in the feature library, and use the material type corresponding to the matching material feature as the material type of the target material.

[0113] Optionally, the determination module 802 is specifically configured to:

[0114] Determine N contact point data of the target material based on N touch information, where the N contact point data includes n stiffness contact point data and m roughness contact point data. Each stiffness contact point data includes a contact point coordinate and a normal force, and each roughness contact point data includes a normal force and a tangential force. The generation timestamp of any roughness contact point data is greater than the generation timestamp of any stiffness contact point data. The normal forces of the stiffness contact point data other than the target stiffness contact point data among the n stiffness contact point data do not belong to the threshold range, and the normal force of the target stiffness contact point data belongs to the threshold range. The target stiffness contact point data is the stiffness contact point data with the largest generation timestamp among the n stiffness contact point data, and both n and m are positive integers;

[0115] Determine a stiffness feature according to the contact point coordinate of the starting stiffness contact point data and the contact point coordinate and normal force of the target stiffness contact point data, where the starting stiffness contact point data is the stiffness contact point data with the smallest generation timestamp among the n stiffness contact point data;

[0116] Determine the roughness feature based on the normal force and tangential force of the m roughness contact point data;

[0117] Combine the stiffness feature and the roughness feature into the target material feature of the target material.

[0118] Optionally, the determining module 802 is specifically configured to:

[0119] Determine the normal displacement according to the contact point coordinates of the starting stiffness contact point data and the contact point coordinates of the target stiffness contact point data;

[0120] Take the ratio of the normal force of the target stiffness contact point data to the normal displacement as the stiffness feature.

[0121] Optionally, the determining module 802 is specifically configured to:

[0122] Take the ratio of the tangential force to the normal force of each roughness contact point data as the unit roughness feature;

[0123] Perform an averaging process on the m unit roughness features to obtain the roughness feature.

[0124] In a possible embodiment, the device further includes: a building module 804, where:

[0125] The obtaining module 801 is further configured to obtain K sets of sample contact point data, each set of sample contact point data in the K sets of sample contact point data corresponds to a material type, each set of sample contact point data includes a plurality of unit sample contact point data sets, each unit sample contact point data set includes a plurality of sample contact point data, and each sample contact point data includes contact point coordinates, a normal force, and a tangential force, and K is a positive integer;

[0126] The determining module 802 is further configured to determine the sample material feature corresponding to each unit sample contact point data set, and cluster all the sample material features into K feature clusters, and the K feature clusters correspond to K feature cluster centers;

[0127] The building module 804 is configured to build a feature library according to the K feature cluster centers and the material type corresponding to each feature cluster center.

[0128] Optionally, the obtaining module 801 is specifically configured to:

[0129] Obtain starting touch information, and determine the sample starting contact point data corresponding to the starting touch information, where the sample starting contact point data includes a starting normal force, and the starting touch information is data obtained after a touch device touches a sample material;

[0130] Determine a first movement instruction according to the starting normal force, and obtain first movement touch information, where the first movement touch information is data obtained after the touch device moves and touches according to the first movement instruction;

[0131] Determine first movement contact point data corresponding to the first movement touch information. The first movement contact point data includes a movement normal force and a movement tangential force, and the movement normal force of the first movement contact point data is used to determine the next first movement instruction;

[0132] When the movement normal force of the first movement contact point data is within a threshold range, both the sample starting contact point data and the first movement contact point data are used as sample stiffness contact point data;

[0133] Determine a second movement instruction according to the movement tangential force of the first movement contact point data, and determine sample roughness contact point data according to the second movement instruction;

[0134] Combine the sample roughness contact point data and the sample stiffness contact point data into the any unit sample contact point data set.

[0135] Optionally, the obtaining module 801 is specifically configured to:

[0136] Obtain second movement touch information, where the second movement touch information is data obtained after the touch device moves and touches according to the second movement instruction;

[0137] Determine second movement contact point data corresponding to the second movement touch information. The second movement contact point data includes a movement tangential force and a movement normal force, the movement normal force of the second movement contact point data is within a threshold range, and the movement tangential force of the second movement contact point data is used to determine the next second movement instruction;

[0138] When the touch device meets a stop condition, the obtained second movement contact point data is used as the sample roughness contact point data.

[0139] Optionally, the matching module 803 is specifically configured to:

[0140] Obtain multiple material features included in the feature library;

[0141] Determine a feature distance between the target material feature and each material feature included in the feature library, and select the material feature with the smallest feature distance among the multiple feature distances corresponding to the multiple material features as the matching material feature.

[0142] For the device embodiment, since it is basically similar to the method embodiment, refer to the partial description of the method embodiment for the related parts.

[0143] Please refer to Figure 9 , which is a schematic structural diagram of a computer device provided by an embodiment of the present application. As Figure 9 shown, the computer device may include a processor 901, a memory 902, a network interface 903, and at least one communication bus 904. Among them, the processor 901 is used to schedule computer programs, and may include a central processing unit, a controller, and a microprocessor; the memory 902 is used to store computer programs, and may include a high-speed random access memory and a non-volatile memory, such as a disk storage device and a flash memory device; the network interface 903 provides data communication functions, and the communication bus 904 is responsible for connecting each communication component.

[0144] Among them, the processor 901 may be used to call the computer program in the memory to perform the following operations:

[0145] Obtain touch information for a target material, where the touch information is data obtained after touching the target material;

[0146] Determine the target material characteristics of the target material according to the touch information, where the target material characteristics include material roughness characteristics and material stiffness characteristics;

[0147] Among the multiple material characteristics included in the feature library, determine the matching material characteristics that match the target material characteristics, and use the material type corresponding to the matching material characteristics as the material type of the target material.

[0148] In a possible embodiment, the processor 901 is used to:

[0149] Determine N contact point data of the target material based on N touch information, where the N contact point data includes n stiffness contact point data and m roughness contact point data. Each stiffness contact point data includes a contact point coordinate and a normal force, and each roughness contact point data includes a normal force and a tangential force. The generation timestamp of any roughness contact point data is greater than the generation timestamp of any stiffness contact point data. The normal forces of the stiffness contact point data other than the target stiffness contact point data among the n stiffness contact point data do not belong to the threshold range, and the normal force of the target stiffness contact point data belongs to the threshold range. The target stiffness contact point data is the stiffness contact point data with the largest generation timestamp among the n stiffness contact point data, and both n and m are positive integers;

[0150] Determine the stiffness characteristics according to the contact point coordinates of the starting stiffness contact point data and the contact point coordinates and normal force of the target stiffness contact point data, where the starting stiffness contact point data is the stiffness contact point data with the smallest generation timestamp among the n stiffness contact point data;

[0151] Determine the roughness feature based on the normal force and tangential force of the m roughness contact point data;

[0152] Combine the stiffness feature and the roughness feature into the target material feature of the target material.

[0153] In a possible embodiment, the processor 901 is configured to:

[0154] Determine the normal displacement according to the contact point coordinates of the starting stiffness contact point data and the contact point coordinates of the target stiffness contact point data;

[0155] Take the ratio of the normal force of the target stiffness contact point data to the normal displacement as the stiffness feature.

[0156] In a possible embodiment, the processor 901 is configured to:

[0157] Take the ratio of the tangential force to the normal force of each roughness contact point data as the unit roughness feature;

[0158] Perform an averaging process on the m unit roughness features to obtain the roughness feature.

[0159] In a possible embodiment, the processor 901 is configured to:

[0160] Obtain K sets of sample contact point data, each set of sample contact point data in the K sets of sample contact point data corresponds to a material type, each set of sample contact point data includes multiple unit sample contact point data sets, each unit sample contact point data set includes multiple sample contact point data, and each sample contact point data includes contact point coordinates, normal force, and tangential force, where K is a positive integer;

[0161] Determine the sample material features corresponding to each unit sample contact point data set respectively, and cluster all the sample material features into K feature clusters, and the K feature clusters correspond to K feature cluster centers;

[0162] Establish a feature library according to the K feature cluster centers and the material type corresponding to each feature cluster center.

[0163] In a possible embodiment, the processor 901 is configured to:

[0164] Obtain starting touch information, and determine the sample starting contact point data corresponding to the starting touch information, where the sample starting contact point data includes the starting normal force, and the starting touch information is data obtained after the touch device touches the sample material;

[0165] Determine a first movement instruction based on the starting normal force, and obtain first movement touch information, where the first movement touch information is data obtained after the touch device moves according to the first movement instruction and touches.

[0166] Determine first movement contact point data corresponding to the first movement touch information. The first movement contact point data includes a movement normal force and a movement tangential force. The movement normal force of the first movement contact point data is used to determine the next first movement instruction.

[0167] When the movement normal force of the first movement contact point data is within a threshold range, both the sample starting contact point data and the first movement contact point data are used as sample stiffness contact point data.

[0168] Determine a second movement instruction according to the movement tangential force of the first movement contact point data, and determine sample roughness contact point data according to the second movement instruction.

[0169] Combine the sample roughness contact point data and the sample stiffness contact point data into the any unit sample contact point data set.

[0170] In a possible embodiment, the processor 901 is configured to:

[0171] Obtain second movement touch information, where the second movement touch information is data obtained after the touch device moves according to the second movement instruction and touches.

[0172] Determine second movement contact point data corresponding to the second movement touch information. The second movement contact point data includes a movement tangential force and a movement normal force. The movement normal force of the second movement contact point data is within a threshold range. The movement tangential force of the second movement contact point data is used to determine the next second movement instruction.

[0173] When the touch device meets the stop condition, use the obtained second movement contact point data as the sample roughness contact point data.

[0174] In a possible embodiment, the processor 901 is configured to:

[0175] Obtain multiple material features included in the feature library.

[0176] Determine the feature distance between the target material feature and each material feature included in the feature library, and select the material feature with the smallest feature distance among the multiple feature distances corresponding to the multiple material features as the matching material feature.

[0177] It should be understood that the computer device described in the embodiments of the present application can implement the description of the material recognition method in the embodiments, and can also execute the description of the material recognition device in the corresponding embodiments, which will not be elaborated here. In addition, the description of the beneficial effects of using the same method will not be elaborated here either.

[0178] In addition, it should be pointed out that the embodiments of the present application also provide a storage medium, in which a computer program for the foregoing material recognition method is stored. The computer program includes program instructions. When one or more processors load and execute the program instructions, the description of the material recognition method in the embodiments can be implemented, which will not be elaborated here. The description of the beneficial effects of using the same method will not be elaborated here either. It can be understood that the program instructions can be deployed on one or multiple computer devices that can communicate with each other for execution.

[0179] The embodiments of the present application also provide a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps performed in the embodiments of the foregoing various methods.

[0180] The foregoing disclosures are only the preferred embodiments of the present application. Of course, the scope of rights of the present application cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present application still fall within the scope covered by the present application.

Claims

1. A method for material identification, characterized in that, Including: Obtaining N touch information for a target material, where each touch information is data obtained after touching the target material; N is a positive integer; Determining N contact point data of the target material based on the N touch information, where the N contact point data includes n stiffness contact point data and m roughness contact point data. Each stiffness contact point data includes a contact point coordinate and a normal force, and each roughness contact point data includes a normal force and a tangential force. The generation timestamp of any roughness contact point data is greater than that of any stiffness contact point data. The normal forces of the stiffness contact point data other than the target stiffness contact point data among the n stiffness contact point data do not belong to the threshold range, and the normal force of the target stiffness contact point data belongs to the threshold range. The target stiffness contact point data is the stiffness contact point data with the maximum generation timestamp among the n stiffness contact point data. Both n and m are positive integers; Determining a normal displacement based on the contact point coordinate of the starting stiffness contact point data and the contact point coordinate of the target stiffness contact point data, and taking the ratio of the normal force of the target stiffness contact point data to the normal displacement as a stiffness feature; where the starting stiffness contact point data is the stiffness contact point data with the minimum generation timestamp among the n stiffness contact point data; Taking the ratio of the tangential force to the normal force of each roughness contact point data as a unit roughness feature, and performing an averaging process on the m unit roughness features to obtain a roughness feature; Combining the stiffness feature and the roughness feature into the target material feature of the target material; Among the multiple material features included in the feature library, determining a matching material feature that matches the target material feature, and taking the material type corresponding to the matching material feature as the material type of the target material.

2. The method according to claim 1, characterized in that, The method further includes: Obtaining K sets of sample contact point data, where each set of sample contact point data in the K sets of sample contact point data corresponds to a material type. Each set of sample contact point data includes multiple unit sample contact point data sets, each unit sample contact point data set includes multiple sample contact point data, and each sample contact point data includes a contact point coordinate, a normal force, and a tangential force. K is a positive integer; Determining the sample material features corresponding to each unit sample contact point data set respectively, and clustering all the sample material features into K feature clusters, where the K feature clusters correspond to K feature cluster centers; Establishing a feature library based on the K feature cluster centers and the material types corresponding to each feature cluster center.

3. The method according to claim 2, characterized in that, For any one of the multiple unit sample contact point data sets, the process of obtaining the any one of the unit sample contact point data sets includes: Obtaining starting touch information, and determining the sample starting contact point data corresponding to the starting touch information. The sample starting contact point data includes a starting normal force. The starting touch information is data obtained after a touch device touches a sample material; Determine a first movement instruction according to the starting normal force, and obtain first movement touch information, where the first movement touch information is data obtained after the touch device moves and touches according to the first movement instruction; Determine first movement contact point data corresponding to the first movement touch information. The first movement contact point data includes a movement normal force and a movement tangential force, and the movement normal force of the first movement contact point data is used to determine the next first movement instruction; When the movement normal force of the first movement contact point data is within a threshold range, use both the sample starting contact point data and the first movement contact point data as sample stiffness contact point data; Determine a second movement instruction according to the movement tangential force of the first movement contact point data, and determine sample roughness contact point data according to the second movement instruction; Combine the sample roughness contact point data and the sample stiffness contact point data into the any unit sample contact point data set.

4. The method according to claim 3, characterized in that, The determining the sample roughness contact point data according to the second movement instruction includes: Obtain second movement touch information, where the second movement touch information is data obtained after the touch device moves and touches according to the second movement instruction; Determine second movement contact point data corresponding to the second movement touch information. The second movement contact point data includes a movement tangential force and a movement normal force. The movement normal force of the second movement contact point data is within a threshold range, and the movement tangential force of the second movement contact point data is used to determine the next second movement instruction; When the touch device meets the stop condition, use the obtained second movement contact point data as the sample roughness contact point data.

5. The method according to claim 1, characterized in that, The determining, among multiple material features included in the feature library, a matching material feature that matches the target material feature includes: Obtain the multiple material features included in the feature library; Determine the feature distance between the target material feature and each material feature included in the feature library, and select, from the multiple feature distances corresponding to the multiple material features, the material feature with the smallest feature distance as the matching material feature.

6. A material identification device, characterized in that, including: An obtaining module, configured to obtain N touch information for a target material, where each touch information is data obtained after touching the target material; N is a positive integer; A determination module, configured to determine N contact point data of the target material based on the N touch information, where the N contact point data includes n stiffness contact point data and m roughness contact point data, each stiffness contact point data includes a contact point coordinate and a normal force, each roughness contact point data includes a normal force and a tangential force, the generation timestamp of any roughness contact point data is greater than the generation timestamp of any stiffness contact point data, the normal forces of the stiffness contact point data other than the target stiffness contact point data among the n stiffness contact point data do not belong to the threshold range, and the normal force of the target stiffness contact point data belongs to the threshold range, the target stiffness contact point data is the stiffness contact point data with the largest generation timestamp among the n stiffness contact point data, and both n and m are positive integers; determine a normal displacement according to the contact point coordinate of the starting stiffness contact point data and the contact point coordinate of the target stiffness contact point data, and use the ratio of the normal force of the target stiffness contact point data to the normal displacement as a stiffness feature; where the starting stiffness contact point data is the stiffness contact point data with the smallest generation timestamp among the n stiffness contact point data; use the ratio of the tangential force to the normal force of each roughness contact point data as a unit roughness feature, and perform an averaging process on the m unit roughness features to obtain a roughness feature; combine the stiffness feature and the roughness feature into the target material feature of the target material; A matching module, configured to determine a matching material feature that matches the target material feature among multiple material features included in the feature library, and use the material type corresponding to the matching material feature as the material type of the target material.

7. A storage medium, characterized in that, The storage medium stores a computer program, the computer program includes program instructions, and one or more processors load and execute the program instructions to execute the method according to any one of claims 1-5.

8. A computer program product, characterized in that, The computer program product includes computer instructions, the computer instructions are stored in a computer-readable storage medium, and when the computer instructions are read and executed by a processor from the computer-readable storage medium, the method according to any one of claims 1-5 is executed.

Images