Estimation device, estimation method, estimation program, and learning model generation device

Abstract

To estimate a striking state including at least one of a state of a striking tool having a conductive flexible material and a state of an object to be struck that is struck by the striking tool without using a special detection device.SOLUTION: An estimation device detects electric characteristics between a plurality of detection points predetermined in a flexible material of a striking tool having the conductive flexible material whose electric characteristics change according to a change in an imparted stimulus, inputs detected time-series electric characteristics to a learning model that is learned so as to output striking state information with the time-series electric characteristics as input, using the time-series electric characteristics when a stimulus is given to the flexible material and the striking state information indicating a striking state including at least one of a state of the striking tool for giving a stimulus to the flexible material and a state of an object to be struck that is struck by the striking tool as learning data, and estimates the striking state information indicating the striking state corresponding to the input time-series electric characteristics.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to an estimation device, an estimation method, an estimation program, and a learning model generation device.

Background Art

[0002] Conventionally, it has been practiced to detect a shape change occurring in an object and estimate the state of a person or object that deforms the object using the detection result. In terms of detecting the shape change occurring in the object, it is difficult to detect the deformation without inhibiting the deformation of the object. In addition, since strain sensors used for detecting rigid bodies such as metal deformation are difficult to use for articles, a special detection device is required to detect the deformation of the object. For example, a technique is known in which displacement and vibration of an object are measured by a camera to obtain a deformed image and the amount of deformation is extracted (see, for example, Patent Document 1). In addition, a technique related to a flexible tactile sensor that estimates the amount of deformation from the amount of light transmission is also known (see, for example, Patent Document 2).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in terms of detecting a shape change occurring in an object or the like, when detecting the amount of deformation such as displacement of the object using a camera and an image analysis method, a system including a camera and image analysis becomes large-scale and causes the device to become large, which is not preferable. In addition, in the optical method using a camera, measurement of hidden parts that are not imaged by the camera cannot be performed. Therefore, there is room for improvement in detecting the deformation of an object.

[0005] In particular, when detecting a hitting state such as the state of a shoe and the state of a ball being kicked when kicking a ball with, for example, a sports shoe, there was no choice but to provide a dedicated detector to detect the hitting state.

[0006] The present disclosure aims to provide an estimation device, an estimation method, an estimation program, and a learning model generation device that can estimate a hitting state including at least one of the state of a hitting tool and the state of a hitting object hit by the hitting tool, by utilizing the electrical characteristics of a hitting tool provided with a flexible material having conductivity, without using a special detection device.

Means for Solving the Problems

[0007] In order to achieve the above object, a first aspect includes a detection unit that detects the electrical characteristics between a plurality of predetermined detection points on the flexible material of a hitting tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in a given stimulus; Using the time-series electrical characteristics when a stimulus is applied to the flexible material and hitting state information indicating a hitting state including at least one of the state of the hitting tool when the stimulus is applied to the flexible material and the state of the hitting object hit by the hitting tool as learning data, for a learning model trained to take the time-series electrical characteristics as input and output the hitting state information, the time-series electrical characteristics detected by the detection unit are input, and hitting state information indicating a hitting state including at least one of the state of the hitting tool corresponding to the input time-series electrical characteristics and the state of the hitting object hit by the hitting tool is estimated, and an estimation device including an estimation unit.

[0008] A second aspect is the estimation device according to the first aspect, wherein the hitting object is a sports ball, the hitting tool is the sports shoe, and the hitting state includes at least one of the state of the shoe and the state of the ball kicked by the shoe when the ball is kicked with the shoe.

[0009] A third aspect is the estimation device according to the second aspect, wherein the state of the shoe includes a state related to the way of kicking the ball with the shoe.

[0010] Aspect 4 is that in the estimation device of Aspect 2, the state of the ball includes a state related to at least one of the rotation, trajectory, and flight distance of the ball.

[0011] Aspect 5 is that in the estimation device of any one of Aspects 1 to 4, the striking tool includes a material in which conductivity is imparted to at least a part of a urethane material having a structure with at least one of a fibrous and a mesh-like skeleton, or a structure in which a plurality of minute air bubbles are scattered inside.

[0012] Aspect 6 is that in the estimation device of any one of Aspects 1 to 5, the learning model includes a model generated by learning using a network by reservoir computing using the flexible material as a reservoir.

[0013] Aspect 7 is an estimation method in which a computer detects electrical characteristics between a plurality of predetermined detection points in the flexible material of a striking tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in a given stimulus, and uses, as learning data, the time-series electrical characteristics when the flexible material is stimulated and striking state information indicating a striking state including at least one of the state of the striking tool when the flexible material is stimulated and the state of a striking object struck by the striking tool, inputs the detected time-series electrical characteristics to a learning model learned to output the striking state information with the time-series electrical characteristics as an input and the striking state including at least one of the state of the striking tool and the state of the striking object struck by the striking tool corresponding to the input time-series electrical characteristics, and estimates the striking state information.

[0014] The eighth aspect is to cause a computer to detect the electrical characteristics between a plurality of predetermined detection points on the flexible material of a striking tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in a given stimulus, and use, as learning data, the time-series electrical characteristics when the flexible material is stimulated and striking state information indicating a striking state including at least one of the state of the striking tool when the flexible material is stimulated and the state of the striking object struck by the striking tool, and input the detected time-series electrical characteristics into a learning model trained to output the striking state information with the time-series electrical characteristics as the input, and execute a process of estimating the striking state information indicating the striking state including at least one of the state of the striking tool corresponding to the input time-series electrical characteristics and the state of the striking object struck by the striking tool. This is an estimation program for this purpose.

[0015] The ninth aspect includes an acquisition unit that acquires the electrical characteristics from a detection unit that detects the electrical characteristics between a plurality of predetermined detection points on the flexible material of a striking tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in a given stimulus, and striking state information indicating a striking state including at least one of the state of the striking tool that stimulates the flexible material and the state of the striking object struck by the striking tool, and a learning model generation unit that generates a learning model that takes, as input, the time-series electrical characteristics when pressure is applied to the flexible material, and outputs the striking state information indicating the striking state including at least one of the state of the striking tool that stimulates the flexible material and the state of the striking object struck by the striking tool, based on the acquisition result of the acquisition unit. This is a learning model generation device.

Advantages of the Invention

[0016] According to the present disclosure, without using a special detection device, it is possible to estimate a striking state including at least one of the state of the striking tool and the state of the striking object struck by the striking tool by utilizing the electrical characteristics of a striking tool provided with a flexible material having conductivity.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Embodiments for Carrying Out the Invention

[0018] Hereinafter, embodiments for realizing the technology of the present disclosure will be described in detail with reference to the drawings. In addition, for components and processes having the same functions, the same reference numerals are given throughout the drawings, and redundant explanations may be omitted as appropriate. Further, the present disclosure is not limited to the following embodiments, and can be implemented with appropriate modifications within the scope of the object of the present disclosure.

[0019] In the present disclosure, a person is a concept including at least one of a human body and an object capable of applying a stimulus to an object by a physical quantity. In the following description, without distinguishing between at least one of the human body and the object, a person will be described generically as a concept including a human and an object. That is, each single body of the human body and the object, and a combined body of the human body and the object are generically referred to as a person.

[0020] First, with reference to FIGS. 1 to 7, a flexible material to which conductivity is imparted for applying the technology of the present disclosure, and a state estimation process for estimating the state of the imparting side with respect to the flexible material using the flexible material will be described.

[0021] <Flexible material>

[0022] In the present disclosure, the "flexible material" is a concept including a material that can be deformed at least partially, such as being bent. It includes soft elastomers such as rubber materials, structures having at least one of fibrous and network-like skeletons, and structures in which a plurality of minute air bubbles are scattered inside. Examples of these structures include polymer materials such as urethane materials. Further, in the present disclosure, a flexible material to which conductivity is imparted is used. The "flexible material to which conductivity is imparted" is a concept including a material having conductivity, and includes a material obtained by imparting a conductive material to a flexible material to impart conductivity, and a material in which the flexible material has conductivity. A polymer material such as a urethane material is suitable as the flexible material to which conductivity is imparted. In the following description, as an example of the flexible material to which conductivity is imparted, a member formed by blending and infiltrating (also referred to as impregnating) a conductive material into all or part of a urethane material will be described as a "conductive urethane". The conductive urethane can be formed by either blending or infiltrating (impregnating) the conductive material, and can also be formed by combining blending and infiltrating (impregnating) the conductive material. For example, when the conductive urethane formed by infiltration (impregnation) has higher conductivity than the conductive urethane 22 formed by blending, it is preferable to form the conductive urethane by infiltration (impregnation).

[0023] Conductive urethane has a function in which its electrical properties change according to a given physical quantity. Examples of the physical quantity that causes the function of changing electrical properties include a stimulus value based on a pressure value that indicates a stimulus by pressure (hereinafter referred to as pressure stimulus) that deforms the structure such as bending. Note that the pressure stimulus includes pressure application by pressure at a predetermined site and pressure distribution within a predetermined range. Further, other examples of the physical quantity include stimulus values such as the amount of moisture that indicate a stimulus (hereinafter referred to as material stimulus) that changes (alters) the properties of the material by moisture content and moisture application. Conductive urethane changes its electrical properties according to a given physical quantity. An example of the physical quantity representing this electrical property is the electrical resistance value. Further, other examples include voltage value or current value.

[0024] Conductive urethane gives conductivity to a flexible material having a predetermined volume, and thus electrical properties (i.e., change in electrical resistance value) according to a given physical quantity appear, and the electrical resistance value can be regarded as the volume resistance value of the conductive urethane. In conductive urethane, the electrical paths are intricately coordinated. For example, the electrical paths expand and contract or expand and contract according to deformation. Further, there may be cases where the electrical paths are temporarily disconnected and cases where connections different from before occur. Therefore, conductive urethane exhibits behavior in which it has different electrical properties depending on deformation and alteration according to the magnitude and distribution of a stimulus (pressure stimulus and material stimulus) by a given physical quantity between positions separated by a predetermined distance (for example, the position of a detection point where an electrode is arranged). For this reason, the electrical properties of conductive urethane change according to the magnitude and distribution of the stimulus by the given physical quantity.

[0025] Note that by using conductive urethane, it is not necessary to provide a detection point such as an electrode at the target location for deformation and alteration. It is sufficient to provide detection points such as electrodes at at least any two arbitrary locations sandwiching the location where the conductive urethane is stimulated by a physical quantity (for example, Fig. 1).

[0026] In addition, in order to improve the detection accuracy of the electrical characteristics of the conductive urethane, more detection points than two detection points may be used. Further, the conductive urethane of the present disclosure may be formed of a conductive urethane group formed by arranging a plurality of conductive urethane pieces, with the conductive urethane 22 shown in FIG. 1 as one conductive urethane piece. In this case, the electrical characteristics may be detected for each of the plurality of conductive urethane pieces, or the electrical characteristics of the plurality of conductive urethane pieces may be synthesized and detected. When the electrical characteristics are detected for each of the plurality of conductive urethane pieces, the electrical characteristics such as the electrical resistance value can be detected for each arrangement site (for example, detection sets #1 to #n). Further, as another example, the detection range on the conductive urethane 22 may be divided, detection points may be provided for each divided detection range, and the electrical characteristics may be detected for each detection range.

[0027] <Estimation device>

[0028] Next, an example of an estimation device that estimates the state of the application side with respect to the conductive urethane using the conductive urethane will be described.

[0029] FIG. 1 shows an example of the configuration of an estimation device 1 capable of executing an estimation process for estimating the state of the application side. The estimation device 1 includes an estimation unit 5 and is connected to an object 2 so that the electrical characteristics of the conductive urethane 22 are input. In the estimation device 1, the state of the application side with respect to the conductive urethane 22 included in the object 2 is estimated. The estimation device 1 can be realized by a computer including a CPU as an execution device that executes the processes described later.

[0030] The deformation and alteration of the conductive urethane described above occur due to physical quantities applied to the conductive urethane in a time series. The physical quantities applied in this time series depend on the state of the application side. Therefore, the electrical characteristics of the conductive urethane that change over time correspond to the state of the application side of the physical quantities applied to the conductive urethane. For example, when a pressure stimulus that deforms the conductive urethane or a material stimulus that alters the conductive urethane is applied, the electrical characteristics of the conductive urethane that change over time correspond to the state of the application side indicating the position, distribution, and magnitude of the pressure stimulus. Thus, it is possible to estimate the state of the application side with respect to the conductive urethane from the electrical characteristics of the conductive urethane that change over time.

[0031] In the estimation device 1, an unknown state of the application side is estimated and output using the learned learning model 51 by an estimation process described later. As a result, it becomes possible to identify the state of the application side with respect to the object 2 without directly measuring the deformation and alteration of the conductive urethane 22 included in the object using a special device or a large device. The learning model 51 is learned with the state of the application side with respect to the object 2 and the electrical characteristics of the object 2 (that is, the electrical characteristics such as the electrical resistance value of the conductive urethane 22 disposed on the object 2) as inputs. The learning of the learning model 51 will be described later.

[0032] Note that the conductive urethane 22 can be arranged on a flexible member 21 to constitute the object 2 (Fig. 2). The object 2 constituted by the member 21 on which the conductive urethane 22 is arranged includes an electrical characteristic detection unit 76. The conductive urethane 22 may be arranged on at least a part of the member 21, and may be arranged inside or outside. Also, the conductive urethane may be arranged so that the state of the application side to the conductive urethane can be estimated. For example, it may be arranged so as to be directly or indirectly, or both, contactable with a person.

[0033] Fig. 2 shows an example of the arrangement of the conductive urethane 22 in the object 2. As shown by taking the A-A cross-section of the object 2 as the object cross-section 2-1, the conductive urethane 22 may be formed to fill the entire interior of the member 21. Also, as shown in the object cross-section 2-2, the conductive urethane 22 may be formed on one side (the surface side) inside the member 21, and as shown in the object cross-section 2-3, the conductive urethane 22 may be formed on the other side (the back side) inside the member 21. Further, as shown in the object cross-section 2-4, the conductive urethane 22 may be formed in a part of the interior of the member 21. Also, as shown in the object cross-section 2-5, the conductive urethane 22 may be separately arranged outside the surface side of the member 21, and as shown in the object cross-section 2-6, it may be arranged outside the other side (the back side). When the conductive urethane 22 is arranged outside the member 21, it may be only laminated with the member 21, or the conductive urethane 22 and the member 21 may be integrated by adhesion or the like. Note that even when the conductive urethane 22 is arranged outside the member 21, since the conductive urethane 22 is a urethane material having conductivity, the flexibility of the member 21 is not inhibited.

[0034] As shown in Fig. 1, the conductive urethane 22 detects the electrical characteristics (i.e., the volume resistance value which is the electrical resistance value) of the conductive urethane 22 based on signals from at least two detection points 75 arranged at a distance. In the example of Fig. 1, a detection set #1 is shown which detects the electrical characteristics (the electrical resistance values in time series) based on the signals from two detection points 75 arranged at diagonal positions on the conductive urethane 22. Note that the number and arrangement of the detection points 75 are not limited to the positions shown in Fig. 1, and may be three or more in number and at any position as long as the electrical characteristics of the conductive urethane 22 can be detected. Note that for the electrical characteristics of the conductive urethane 22, an electrical characteristics detection unit 76 that detects the electrical characteristics (for example, the volume resistance value which is the electrical resistance value) may be connected to the detection point 75 and its output may be used.

[0035] In this embodiment, since the conductive urethane 22 is used as the sensor, for example, when a person is intervening, the discomfort given to the person is extremely less than that of the conventional sensor. Therefore, it is possible to simultaneously perform measurement and estimation of the state of the application side without harming the state of the application side regarding the person during measurement. This is an advantage compared to the conventional sensor that separately performs measurement and estimation of the state of the application side, and especially in the estimation by long-term measurement evaluation that follows time-series changes, the merit is great.

[0036] The estimation unit 5 is a functional unit that is connected to the object 2 and estimates the state of the application side using the learning model 51 based on the electrical characteristics that change according to at least one of the deformation and alteration of the conductive urethane 22. Specifically, time-series input data 4 representing the magnitude of the electrical resistance (such as the electrical resistance value) in the conductive urethane 22 is input to the estimation unit 5. The input data 4 corresponds to state data 3 indicating the state of the application side with respect to the object 2, for example, the state regarding the behavior of a person such as the posture and movement of a person who has come into contact with the object 2. For example, when a person comes into contact with the object 2, the person comes into contact in a predetermined state such as a posture, and in correspondence with this state, a stimulus (at least one of a pressure stimulus and a material stimulus) is given as a physical quantity to the conductive urethane 22 that constitutes the object 2, and the electrical characteristics of the conductive urethane 22 change. Therefore, the electrical characteristics of the conductive urethane 22 that change in time series indicated by the input data 4 correspond to the object 2, that is, the state of the application side with respect to the conductive urethane 22. Further, the estimation unit 5 outputs output data 6 representing the state of the application side corresponding to the electrical characteristics of the conductive urethane 22 that change in time series as an estimation result using the learned learning model 51.

[0037] The learning model 51 is a model that has completed learning to derive output data 6 representing the state of the application side from the electrical resistance (input data 4) of the conductive urethane 22 that changes due to stimuli (pressure stimulus and material stimulus) given as physical quantities. The learning model 51 is, for example, a model that defines a learned neural network, and is expressed as a set of information on the weights (intensities) of the connections between the nodes (neurons) that constitute the neural network.

[0038] <Learning process>

[0039] Next, the learning process for generating the learning model 51 will be described.

[0040] Fig. 3 shows the conceptual configuration of a learning model generation device that generates the learning model 51. The learning model generation device includes a learning processing unit 52. The learning model generation device can be configured to include a computer equipped with a CPU (not shown), and the learning processing unit 52 is executed by the learning data collection process and the learning model generation process executed by the CPU to generate the learning model 51.

[0041] <Learning data collection process>

[0042] In the learning data collection process, the learning processing unit 52 collects a large amount of input data 4 obtained by measuring the electrical characteristics (e.g., electrical resistance value) of the conductive urethane 22 in time series using the state data 3 representing the state of the application side as a label. Therefore, the learning data includes a large number of sets of the input data 4 indicating the electrical characteristics and the state data 3 indicating the state of the application side corresponding to the input data 4.

[0043] Specifically, in the learning data collection process, the electrical characteristics (e.g., electrical resistance value) that change due to the stimuli (pressure stimulus and material stimulus) corresponding to the state of the application side when the state in the object 2 (i.e., the state of the application side with respect to the conductive urethane 22) is formed are acquired in time series. Next, the state data 3 is assigned as a label to the acquired time-series electrical characteristics (input data 4), and the process is repeated until the set of the state data 3 and the input data 4 reaches a predetermined number or a predetermined time. The set of the state data 3 indicating these states of the application side and the time-series electrical characteristics (input data 4) of the conductive urethane 22 acquired for each state of the application side becomes the learning data. Note that the state data 3 in the learning data is stored in a memory (not shown) so as to be treated as output data 6 indicating the state of the application side where the estimation result is correct in the learning process described later.

[0044] Note that, for the learning data, time-series information may be associated by adding information indicating the measurement time to each of the electrical resistance values (input data 4) of the conductive urethane 22. In this case, for the period determined as the state on the application side, information indicating the measurement time may be added to the set of time-series electrical resistance values in the conductive urethane 22 to associate the time-series information.

[0045] An example of the above-described learning data is shown in the following table. Table 1 is an example of a data set in which time-series electrical resistance value data (r) and state data (R) indicating the state on the application side are associated as learning data regarding the state on the application side with respect to the conductive urethane 22.

[0046]

Table 1

[0047] Note that the electrical characteristics detected by the conductive urethane 22 (time characteristics based on time-series electrical resistance value data) can be regarded as a characteristic pattern regarding the state on the application side with respect to the conductive urethane 22. That is, different stimuli are applied to the conductive urethane 22 in time series depending on the state on the application side with respect to the conductive urethane 22. Therefore, it is considered that the time-series electrical characteristics within a predetermined time appear as electrical characteristics characteristic of the state on the application side. Thus, the pattern shown by the electrical characteristics detected by the conductive urethane 22 (time characteristics based on time-series electrical resistance value data) (for example, the distribution shape of the time-series electrical resistance values in the electrical characteristics) corresponds to the state on the application side and functions effectively in the learning process described later.

[0048] <Learning model generation process>

[0049] Next, the learning model generation process will be described. The learning model generation device shown in FIG. 3 generates a learning model 51 using the above-described learning data by the learning model generation process in the learning processing unit 52.

[0050] FIG. 4 is a diagram showing the functional configuration of the learning processing unit 52, that is, the functional configuration of a CPU (not shown) regarding the learning model generation process executed by the learning processing unit 52. The CPU (not shown) of the learning processing unit 52 operates as functional units of a generator 54 and an arithmetic unit 56. The generator 54 has a function of generating an output in consideration of the context of the electrical resistance values acquired in a time series as an input.

[0051] The learning processing unit 52 holds a large number of sets of the above-described input data 4 (for example, electrical resistance values) and output data 6 which is state data 3 indicating the state on the application side where the conductive urethane 22 is stimulated as learning data in a memory (not shown).

[0052] The generator 54 includes an input layer 540, an intermediate layer 542, and an output layer 544 to constitute a known neural network (NN). Since the neural network itself is a known technique, detailed description thereof is omitted. The intermediate layer 542 includes a large number of node groups (neuron groups) having inter-node connections and feedback connections. Data from the input layer 540 is input to the intermediate layer 542, and the data of the calculation result of the intermediate layer 542 is output to the output layer 544.

[0053] The generator 54 is a neural network that generates generated output data 6A as data representing the state on the application side or data close to the state on the application side from the input input data 4 (for example, electrical resistance values). The generated output data 6A is data obtained by estimating the state on the application side where the conductive urethane 22 is stimulated from the input data 4. The generator 54 generates generated output data indicating a state close to the state on the application side from the input data 4 input in a time series. By learning using a large number of input data 4, the generator 54 can generate generated output data 6A close to the state on the application side of the object 2, that is, a person or the like to whom the conductive urethane 22 is stimulated. In another aspect, by capturing the electrical characteristics which are the input data 4 input in a time series as a pattern and learning the pattern, the generator 54 can generate generated output data 6A close to the state on the application side of the object 2, that is, a person or the like to whom the conductive urethane 22 is stimulated.

[0054] The calculator 56 is a calculator that compares the generated output data 6A with the output data 6 of the learning data and calculates the error of the comparison result. The learning processing unit 52 inputs the generated output data 6A and the output data 6 of the learning data to the calculator 56. In response to this, the calculator 56 calculates the error between the generated output data 6A and the output data 6 of the learning data, and outputs a signal indicating the calculation result.

[0055] The learning processing unit 52 performs learning of the generator 54 to tune the weight parameters of the connections between the nodes based on the error calculated by the calculator 56. Specifically, the weight parameters of the connections between the nodes of the input layer 540 and the intermediate layer 542, the weight parameters of the connections between the nodes within the intermediate layer 542, and the weight parameters of the connections between the intermediate layer 542 and the output layer 544 in the generator 54 are each fed back to the generator 54 using a method such as the gradient descent method or the error backpropagation method. That is, with the output data 6 of the learning data as the target, all the connections between the nodes are optimized so as to minimize the error between the generated output data 6A and the output data 6 of the learning data.

[0056] Note that the generator 54 may use a recurrent neural network having a function of generating an output in consideration of the temporal input relationship, or other methods may be used.

[0057] The learning processing unit 52 generates a learning model 51 using the above-described learning data by the learning model generation process. The learning model 51 is expressed as a set of information on the weight parameters (weights or strengths) of the connections between the nodes of the learning result, and is stored in a memory (not shown).

[0058] Specifically, the learning processing unit 52 executes learning model generation processing according to the following procedure. In the first learning process, input data 4 (electrical characteristics) with information indicating the state of the imparting side, which is learning data of the results measured in time series, is acquired as a label. In the second learning process, a learning model 51 is generated using the learning data of the results measured in time series. That is, a set of information on the connection weight parameters (weights or strengths) between the nodes of the learning result learned using a large number of learning data as described above is obtained. Then, in the third learning process, the data expressed as a set of information on the connection weight parameters (weights or strengths) between the nodes of the learning result is stored as the learning model 51.

[0059] Then, in the above-described estimation device 1, the learned generator 54 (that is, the data expressed as a set of information on the connection weight parameters between the nodes of the learning result) is used as the learning model 51. By using the sufficiently learned learning model 51, it is not impossible to identify the state on the imparting side from the time-series electrical characteristics of the object 2, that is, the conductive urethane 22 (for example, the characteristics of the electrical resistance value that changes in time series).

[0060] <prc>

[0061] Incidentally, as described above, the conductive urethane 22 exhibits behavior according to changes (deformations) such as the complex cooperation of electrical paths, the expansion and contraction, temporary disconnection, and new connection of electrical paths, as well as changes (alterations) in the properties of the material. As a result, the conductive urethane 22 exhibits behavior with different electrical characteristics according to a given stimulus (e.g., a pressure stimulus). This makes it possible to treat the conductive urethane 22 as a reservoir for storing data related to the deformation of the conductive urethane 22. That is, the estimation device 1 can apply the conductive urethane 22 to a network model called physical reservoir computing (PRC) (hereinafter referred to as PRCN). Since PRC and PRCN themselves are known technologies, detailed descriptions are omitted, but PRC and PRCN are suitable for estimating information related to the deformation and alteration of the conductive urethane 22.

[0062] FIG. 5 shows an example of the functional configuration of the learning processing unit 52 to which PRCN is applied. The learning processing unit 52 to which PRCN is applied includes an input reservoir layer 541 and an estimation layer 545. The input reservoir layer 541 corresponds to the conductive urethane 22 included in the object 2. That is, in the learning processing unit 52 to which PRCN is applied, the object 2 including the conductive urethane 22 is treated as a reservoir for storing data related to the deformation and alteration of the object 2 including the conductive urethane 22 and is learned. The conductive urethane 22 becomes an electrical characteristic (electrical resistance value) corresponding to each of various stimuli, functions as an input layer for inputting the electrical resistance value, and also functions as a reservoir layer for storing data related to the deformation and alteration of the conductive urethane 22. Since the conductive urethane 22 outputs different electrical characteristics (input data 4) according to the stimulus given by the state of the applying side such as a person, it is possible to estimate the unknown state of the applying side from the electrical resistance value of the given conductive urethane 22 in the estimation layer 545. Therefore, in the learning processing in the learning processing unit 52 to which PRCN is applied, it is only necessary to learn the estimation layer 545.

[0063] <Configuration of the Estimation Device>

[0064] Next, an example of the specific configuration of the above-described estimation device 1 will be further described. FIG. 6 shows an example of the electrical configuration of the estimation device 1. The estimation device 1 shown in FIG. 6 is configured to include a computer as an execution device that executes processes for realizing the various functions described above. The above-described estimation device 1 can be realized by causing a computer to execute a program representing each of the above functions.

[0065] The computer that functions as the estimation device 1 includes a computer main body 100. The computer main body 100 includes a CPU 102, a RAM 104 such as a volatile memory, a ROM 106, an auxiliary storage device 108 such as a hard disk drive (HDD), and an input / output interface (I / O) 110. These CPU 102, RAM 104, ROM 106, auxiliary storage device 108, and input / output I / O 110 are configured to be connected via a bus 112 so as to be able to exchange data and commands with each other. Further, connected to the input / output I / O 110 are a communication unit 114 for communicating with an external device, an operation display unit 116 such as a display and a keyboard, and a detection unit 118. The detection unit 118 acquires input data 4 (electrical characteristics such as a time-series electrical resistance value) from the object 2 including the conductive urethane 22. That is, the detection unit 118 can acquire the input data 4 from the electrical characteristic detection unit 76 connected to the detection point 75 in the conductive urethane 22, including the object 2 in which the conductive urethane 22 is disposed. Note that the detection unit 118 may be connected via the communication unit 114.

[0066] Stored in the auxiliary storage device 108 is a control program 108P for causing the computer main body 100 to function as the estimation device 1 as an example of the estimation device of the present disclosure. The CPU 102 reads the control program 108P from the auxiliary storage device 108, expands it in the RAM 104, and executes the process. Thereby, the computer main body 100 that has executed the control program 108P operates as the estimation device 1.

[0067] Note that the auxiliary storage device 108 stores a learning model 108M including the learning model 51 and data 108D including various types of data. The control program 108P may be provided by a recording medium such as a CD-ROM.

[0068] <Estimation process>

[0069] Next, the estimation process in the estimation device 1 implemented by a computer will be further described. FIG. 7 shows an example of the flow of the estimation process by the control program 108P executed by the computer main body 100. The estimation process shown in FIG. 7 is executed by the CPU 102 when the power of the computer main body 100 is turned on. The CPU 102 reads the control program 108P from the auxiliary storage device 108, expands it in the RAM 104, and executes the process.

[0070] First, the CPU 102 reads the learning model 51 from the learning model 108M of the auxiliary storage device 108 and expands it in the RAM 104 to obtain the learning model 51 (step S200). Specifically, a network model (see FIGS. 4 and 5) that is the connection between nodes by the weight parameters expressed as the learning model 51 is expanded in the RAM 104, thereby constructing the learning model 51 in which the connection between nodes by the weight parameters is realized.

[0071] Next, the CPU 102 acquires, in time series, unknown input data 4 (electrical characteristics), which is the target for estimating the state on the application side due to the stimulus applied to the conductive urethane 22, via the detection unit 118 (step S202). Next, the CPU 102 estimates output data 6 (unknown state on the application side), which corresponds to the acquired input data 4, using the learning model 51 (step S204). Then, the CPU 102 outputs the output data 6 (state on the application side) of the estimation result via the communication unit 114 (step S206) and ends this processing routine.

[0072] Thus, according to the estimation device 1, it is possible to estimate the state of the unknown application side from the electrical resistance value of the conductive urethane 22. Specifically, in the estimation device 1, it is possible to estimate the state of the application side such as a person from the input data 4 (electrical characteristics) that changes according to the stimulus applied to the conductive urethane 22 depending on the state of the application side. That is, it is possible to estimate the state of the application side such as a person without using a special device or a large device or directly measuring the deformation of the flexible member.

[0073] <Estimation of the hitting state>

[0074] The case of applying such a conductive urethane 22 to a hitting tool will be described. The "hitting tool" according to the present embodiment is a wearable item worn by a wearer for hitting a hitting target, and there is no limitation on the type as long as it is a flexible wearable item. For example, sports shoes and gloves used in martial arts are included in the hitting tools. The "flexible wearable item" refers to a wearable item that can be deformed as the wearer moves. The "hitting target" is an object that the wearer hits with the hitting tool. When the hitting tool is sports shoes, the hitting target is a ball, and various sports balls such as a soccer ball and a rugby ball are included. When the hitting tool is a glove for martial arts, the hitting target is the human opponent.

[0075] When the conductive urethane 22 is applied to the hitting tool, a pressure stimulus due to deformation such as expansion and contraction of the hitting tool occurs according to the movement of the wearing part of the hitting tool of the wearer of the hitting tool. Therefore, the estimation device 1 can estimate hitting state information indicating a hitting state including at least one of the state of the hitting tool and the state of the hitting target hit by the hitting tool as the state of the application side described above from the time-series electrical resistance value of the conductive urethane 22. For the sake of convenience of explanation, hereinafter, the wearer of the hitting tool will simply be referred to as the "wearer".

[0076] Here, the hitting state includes at least one of, for example, the state of the shoe and the state of the ball kicked by the shoe when kicking a ball with the shoe. Specifically, the state of the shoe includes the state regarding the way of kicking the ball by the shoe. Further, the state of the ball includes the state regarding at least one of the rotation, trajectory, and flight distance of the ball.

[0077] FIG. 8 is a diagram showing an example in which the conductive urethane 22 is applied to a soccer shoe S as an object 2 for a ball game. The shoe S is an example of a hitting tool. Further, in the present embodiment, as shown in FIG. 8, a case where the shoe S is a soccer shoe for kicking a soccer ball B will be described. The ball B is an example of an object to be hit.

[0078] In the example shown in FIG. 8, the wearer H is wearing a shoe S including the conductive urethane 22 on the foot. That the conductive urethane 22 is included in the shoe S means that the arrangement example of the conductive urethane 22 and the member 21 (in this case, cotton or chemical fiber) constituting the shoe S satisfies any of the arrangement examples of the conductive urethane 22 and the member 21 as shown in FIG. 2.

[0079] As shown in FIG. 8, the conductive urethane 22 is provided over the entire inside of the shoe S. Thereby, when the wearer H puts on the shoe S, the foot of the wearer H is covered with the conductive urethane 22. Further, the shoe S includes a detection unit 78. The detection unit 78 is provided, for example, at the heel portion of the shoe S, but is not limited thereto.

[0080] As shown in FIG. 9, the detection unit 78 includes an electrical property detection unit 76 to which a plurality of detection points 75 provided on the conductive urethane 22 are connected, a communication unit 80, a power storage unit 82, and a power generation unit 84. Note that the number of the detection points 75 is appropriately set according to the size and shape of the shoe S.

[0081] The electrical property detection unit 76 detects the electrical properties input from the detection points 75 and outputs the detection results to the communication unit 80.

[0082] The communication unit 80 communicates with a mobile terminal device 30 such as a smartphone, and transmits to the mobile terminal device 30 the detection result of a physical quantity representing the electrical characteristics obtained from the detection point 75 of the electrical characteristics detection unit 76. The communication unit 80 communicates with the mobile terminal device 30 by short-range wireless communication such as Wi-Fi (registered trademark) or Bluetooth (registered trademark), for example.

[0083] The power storage unit 82 supplies the power for the electrical characteristics detection unit 76 to detect electrical characteristics and the power for the communication unit 80 to communicate with the mobile terminal device 30. The power storage unit 82 is applied with various rechargeable batteries, capacitors, etc., for example, and is charged by the generated power from the power generation unit 84.

[0084] The power generation unit 84 generates power by various well-known methods, and charges the power storage unit 82 by supplying the generated power to the power storage unit 82. For example, using a coil and a magnet, power may be generated by the relative movement of the magnet in the coil in response to the swinging of the shoe S. Alternatively, power may be generated using a power generation element or the like that converts energy such as light, heat, pressure, and vibration into electric power.

[0085] On the other hand, the mobile terminal device 30 includes the above-described configurations of the computer main body 100, the communication unit 114, and the operation display unit 116.

[0086] The mobile terminal device 30 functions as an estimation device 1, and the computer main body 100 estimates a hitting state including at least one of the state of the shoe S and the state of the ball B from the electrical characteristics in the conductive urethane 22 provided on the shoe S using the learned learning model 51.

[0087] The communication unit 114 communicates with the communication unit 80 of the shoe S, and acquires from the shoe S the detection result of a physical quantity representing the electrical characteristics obtained from the detection point 75 of the electrical characteristics detection unit 76. The communication unit 114 communicates with the mobile terminal device 30 by short-range wireless communication such as Wi-Fi (registered trademark) or Bluetooth (registered trademark), for example.

[0088] The operation display unit 116 corresponds to an example of the output unit, and displays a hitting state including at least one of the state of the shoe S and the state of the ball B estimated by the computer main body 100.

[0089] In the estimation process in the portable terminal device 30, using the learned learning model 51, as an unknown hitting state, it estimates and outputs a hitting state including at least one of the state of the shoe S and the state of the ball B kicked by the shoe S when kicking the ball B with the shoe S. Note that hereinafter, a hitting state including at least one of the state of the shoe S and the state of the ball B kicked by the shoe S when kicking the ball B with the shoe S is simply referred to as a hitting state. Thereby, it becomes possible to identify the hitting state without using a special device or a large device or directly measuring the deformation of the conductive urethane 22 included in the shoe S. Therefore, in order to estimate the hitting state from the electrical characteristics of the conductive urethane 22, the learning model 51 for estimating the hitting state from the electrical characteristics of the conductive urethane 22 is stored in the auxiliary storage device 108 in the portable terminal device 30.

[0090] Next, the learning process for generating the learning model 51 for estimating the hitting state information of the wearer will be described.

[0091] The learning processing unit 52 of the learning model generation device shown in FIG. 3 collects a large amount of input data 4 obtained by measuring the electrical resistance value of the conductive urethane 22 in time series with the state data 3 representing the movement of the wearer as learning data in the learning data collection process.

[0092] Specifically, in the learning data collection process, the electrical characteristics (for example, electrical resistance value) of the conductive urethane 22 included in the shoe S, which is subjected to a pressure stimulus due to deformation such as expansion and contraction of the shoe S according to the movement of the foot of the wearer H of the shoe S, are acquired in time series from the electrical characteristic detection unit 76 attached to the shoe S. Next, the state data 3 is assigned to the input data 4, which is the acquired electrical characteristics in time series, as a label indicating the hitting state, and a plurality of learning data combining the state data 3 and the input data 4 are prepared.

[0093] Hereinafter, the electrical resistance value will be used as an example of the electrical characteristics of the conductive urethane 22 included in the shoe S for explanation, but as described above, the current value or the voltage value may be used as the electrical characteristics of the conductive urethane 22.

[0094] As the learning data used for estimating the hitting state, for example, the data set shown in Table 1 described above is used. The data set of Table 1 when the conductive urethane 22 is applied to the shoe S is a data set in which the time-series electrical resistance value data (r) obtained from the shoe S is associated with the state data (R) as the hitting state information indicating the hitting state.

[0095] In this case, the states R1 to Rk ··· in Table 1 are the hitting states R1 to Rk ···. Examples of the hitting state include, as described above, the state of the shoe S when kicking the ball B with the shoe S and the state of the ball B kicked by the shoe S.

[0096] Specific examples of the state of the shoe S when kicking the ball B with the shoe S include the state related to the way of kicking the ball B by the shoe S. The "state related to the way of kicking" includes, for example, at least one of the position of the shoe S in contact with the ball B and the strength of the kick. Thereby, for example, it is possible to estimate the type of the way of kicking, such as whether it is a front kick or an inside kick, and the strength of the kick.

[0097] In addition, specific examples of the state of the ball B kicked by the shoe S include the state related to at least one of the rotation, trajectory, and flight distance of the ball B.

[0098] Here, the state regarding the rotation of the ball B includes, for example, at least one of the rotation speed and rotation direction of the ball B. Thereby, in addition to the rotation speed of the kicked ball B, it is possible to estimate whether the rotation direction is top spin, under spin, or no rotation, etc. Also, by being able to estimate the rotation direction of the ball B, it is possible to estimate the direction in which the ball B flies. For this reason, it becomes possible to grasp how the ball B rotates and in which direction it flies when the ball B is kicked in any kicking manner.

[0099] Also, the state regarding the trajectory of the ball B refers to, for example, the trajectory of the ball B from when it is kicked until it lands on the ground. Thereby, it becomes possible to grasp what kind of trajectory the ball B will draw when the ball B is kicked in any kicking manner.

[0100] Also, the state regarding the flying distance of the ball B refers to, for example, the straight-line distance or the length of the trajectory of the ball B from the position where the ball B is kicked to the landing point of the ball B. Thereby, it becomes possible to grasp how far the ball B will fly when the ball B is kicked in any kicking manner.

[0101] The learning processing unit 52 generates the learning model 51 by the above-described learning model generation processing using the learning data in which such time-series electrical resistance value data (r) and the state data (R) indicating the hitting state are associated with each other.

[0102] The estimation device 1 estimates the hitting state from the unknown time-series electrical resistance value data in the conductive urethane 22 by executing the estimation process shown in FIG. 7 using the learning model 51 regarding the hitting state that has machine-learned the correlation between the characteristic pattern represented by the time-series electrical resistance value data of the conductive urethane 22 and the hitting state.

[0103] Specifically, in step S200 of FIG. 7, the CPU 102 of the estimation device 1 acquires the learning model 51 related to the hitting state. In step S202, the CPU 102 acquires, as input data 4, the time-series electrical resistance value data transmitted from the electrical property detection unit 76 provided in the shoe S to be the estimation target of the hitting state.

[0104] Note that the input data 4 acquired in step S202 may be the input data 4 obtained in real time from the shoe S during the execution period of the estimation process, or the input data 4 obtained in advance before the execution of the estimation process.

[0105] In step S204, the CPU 102 inputs the input data 4 acquired in step S202 into the learning model 51 acquired in step S200, and acquires the output data 6 output from the learning model 51. The CPU 102 estimates, as the state of the shoe S when kicking the ball B with the shoe S or the state of the ball B kicked by the shoe S, the hitting state associated with the state data 3 closest to the output data 6 among the state data 3 shown in Table 2.

[0106] In step S206, the CPU 102 outputs the hitting state information indicating the hitting state estimated in step S204. For example, the hitting state information is displayed on the operation display unit 116.

[0107] In this way, according to the estimation device 1 according to the present embodiment, the conductive urethane 22 is applied to the shoe S, and the time-series electrical characteristics of the conductive urethane 22 that change due to the pressure stimulus caused by the deformation such as the expansion and contraction and inflation of the shoe S according to the movement of the foot of the wearer H of the shoe S, and the hitting state is estimated by the output data 6 obtained by inputting the unknown time-series electrical characteristics corresponding to the operation of the wearer H kicking the ball B with the shoe S into the learning model 51 in which the relevance with the hitting state information indicating the hitting state is machine-learned in advance.

[0108] Therefore, with the estimation device 1, for example, it becomes possible to monitor various hitting states described above or to advise the wearer on how to kick the ball B based on the monitoring results. Thus, the estimation device 1 can be applied, for example, in the field of sports instruction.

[0109] In this embodiment, the case where the hitting tool is a soccer shoe and the object to be hit is a soccer ball has been described. However, the hitting tool may be a glove for martial arts, and the object to be hit may be a human opponent. For example, when the martial art is boxing, the conductive urethane 22 is provided inside the boxing glove. In this case, the hitting state includes at least one of the state of the glove and the state of the opponent hit by the glove. The state of the glove includes, for example, at least one of the position where the glove contacts the opponent when punching the opponent and the strength of the punch. In addition, the state of the opponent hit by the glove includes, for example, the magnitude of the damage to the opponent. Thereby, it is possible to estimate the type and strength of punches such as straight punches and hooks. In addition, it is possible to estimate the magnitude of the damage to the opponent.

[0110] As described above, in the present disclosure, the case where the conductive urethane 22 is applied as an example of the flexible member has been described. However, it goes without saying that the flexible member is not limited to the conductive urethane 22.

[0111] In addition, the technical scope of the present disclosure is not limited to the scope described in the above embodiment. Various changes or improvements can be made to the above embodiment without departing from the gist, and the forms to which such changes or improvements are made are also included in the technical scope of the present disclosure.

[0112] In the above embodiment, the case where the estimation process and the learning process are realized by a software configuration using a flowchart has been described. However, the present disclosure is not limited to this, and for example, each process may be realized by a hardware configuration.

[0113] Further, a part of the estimation device 1, for example, a neural network such as the learning model 51, may be configured as a hardware circuit.

Description of Reference Numerals

[0114] 1 Estimation device 2 Object 3 State data 4 Input data 5 Estimation unit 6 Output data 6A Generated output data 22 Conductive urethane 51 Learning model 52 Learning processing unit 54 Generator 56 Arithmetic unit 75 Detection point 76 Electrical property detection unit< / prc>

Claims

1. A detection unit that detects the electrical characteristics between a plurality of predetermined detection points on the flexible material of a striking tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in the applied stimulus; Using, as learning data, the time-series electrical characteristics when a stimulus is applied to the flexible material and a striking state information indicating a striking state including at least one of the state of the striking tool when the stimulus is applied to the flexible material and the state of the object to be struck by the striking tool, for a learning model that is trained to take the time-series electrical characteristics as input and output the striking state information, input the time-series electrical characteristics detected by the detection unit, and estimate the striking state information indicating a striking state including at least one of the state of the striking tool corresponding to the input time-series electrical characteristics and the state of the object to be struck by the striking tool; comprising The object to be struck is a ball for a ball game, and the striking tool is the shoe for the ball game; The striking state includes at least one of the state of the shoe and the state of the ball kicked by the shoe when the ball is kicked with the shoe; The state of the ball includes a state related to at least one of the rotation, trajectory, and flight distance of the ball estimation device.

2. The state of the shoe includes a state related to the way of kicking the ball with the shoe The estimation device according to claim 1.

3. The flexible material includes a material in which conductivity is imparted to at least a part of a urethane material having a structure with at least one of a fibrous and a mesh-like skeleton, or a structure in which a plurality of minute air bubbles are scattered inside The estimation device according to claim 1 or claim 2.

4. The learning model includes a model generated by training using a network by reservoir computing using the flexible material as a reservoir The estimation device according to any one of claims 1 to 3.

5. A computer detects the electrical characteristics between a plurality of predetermined detection points on the flexible material of a striking tool provided with a flexible material having conductivity and whose electrical characteristics change in response to a change in the applied stimulus; Using, as learning data, time-series electrical characteristics when the flexible material is stimulated and impact state information indicating an impact state including at least one of the state of the striking tool that stimulates the flexible material and the state of the object to be struck by the striking tool, for a learning model that is trained to take the time-series electrical characteristics as input and output the impact state information, input the detected time-series electrical characteristics, and estimate impact state information indicating an impact state including at least one of the state of the striking tool corresponding to the input time-series electrical characteristics and the state of the object to be struck by the striking tool. The object to be struck is a ball for a ball game, and the striking tool is the shoe for the ball game. The impact state includes at least one of the state of the shoe and the state of the ball kicked by the shoe when the ball is kicked with the shoe. The state of the ball includes a state related to at least one of the rotation, trajectory, and flight distance of the ball. Estimation method.

6. On a computer, A flexible material having conductivity and whose electrical characteristics change in response to a change in the applied stimulus is provided Detect the electrical characteristics between a plurality of predetermined detection points on the flexible material of the striking tool, Using, as learning data, time-series electrical characteristics when the flexible material is stimulated and impact state information indicating an impact state including at least one of the state of the striking tool that stimulates the flexible material and the state of the object to be struck by the striking tool, for a learning model that is trained to take the time-series electrical characteristics as input and output the impact state information, input the detected time-series electrical characteristics, and estimate impact state information indicating an impact state including at least one of the state of the striking tool corresponding to the input time-series electrical characteristics and the state of the object to be struck by the striking tool. The object to be struck is a ball for a ball game, and the striking tool is the shoe for the ball game. The impact state includes at least one of the state of the shoe and the state of the ball kicked by the shoe when the ball is kicked with the shoe. The state of the ball includes a state related to at least one of the rotation, trajectory, and flight distance of the ball. An estimation program for causing the above processing to be executed.

7. An acquisition unit that acquires the electrical characteristics from a detection unit that detects the electrical characteristics between a plurality of predetermined detection points on the flexible material of a striking tool including a flexible material having conductivity and whose electrical characteristics change in response to a change in the applied stimulus, and striking state information indicating a striking state including at least one of the state of the striking tool that applies a stimulus to the flexible material and the state of a striking object struck by the striking tool. A learning model generation unit that generates a learning model that takes, as input, the time-series electrical characteristics when pressure is applied to the flexible material, based on the acquisition result of the acquisition unit, and outputs striking state information indicating a striking state including at least one of the state of the striking tool that applies a stimulus to the flexible material and the state of a striking object struck by the striking tool. Including The striking object is a ball for a ball game, and the striking tool is the shoe for the ball game. The striking state includes at least one of the state of the shoe and the state of the ball kicked by the shoe when the ball is kicked with the shoe. The state of the ball includes a state related to at least one of the rotation, trajectory, and flight distance of the ball. Learning model generation device.

Images