JP7714949B2 - Golf swing analysis system and analysis method - Google Patents

Description

This disclosure relates to a golf swing analysis system and analysis method.

To analyze golf swings, a system has been proposed that uses a high-precision camera to directly capture images of the behavior of the golf club head during the swing (see, for example, Patent Document 1 below).

Japanese Patent Application Laid-Open No. 2007-167549

However, the analytical device in Patent Document 1 had the problem of requiring large-scale equipment.

This disclosure was devised in consideration of the above-mentioned problems, and its primary purpose is to provide a golf swing analysis system and analysis method that is simple in configuration and capable of analyzing golf swings.

The present disclosure relates to a system for analyzing a golf swing using a golf club having a grip, a shaft, and a head, and includes: a swing sensor that measures first data, which is time-series data of at least one of the angular velocity and acceleration of the golf club during a golf swing; a trajectory sensor that measures second data, which is data related to the trajectory of a ball hit by the golf swing; and a processing device that receives the first data and the second data and executes predetermined processing, wherein the processing device includes a feature calculation unit that calculates predetermined feature values from the first data, and an estimation unit that uses the feature values and the second data to estimate the initial conditions of the head during the golf swing.

By adopting the above-described configuration, the golf swing analysis system disclosed herein enables analysis of golf swings with a simple configuration.

1 is an overall configuration diagram of a golf swing analysis system according to an embodiment of the present invention; FIG. 1 is a block diagram of an analysis system according to an embodiment of the present invention. 10 is a graph showing first data of angular velocity obtained by a swing sensor. 10 is a graph showing first data of acceleration obtained by a swing sensor. 1 is a graph showing a spectrogram of angular velocity. 1 is a graph showing a spectrogram of acceleration. FIG. 1 is a plan view illustrating a head speed, a face angle, and an approach angle of a golf club during a swing. 1 is a side view illustrating a golf club during a swing, illustrating a dynamic loft angle and a blow angle. FIG. FIG. 1 is a front view of a golf club illustrating left and right impact points and upper and lower impact points. FIG. 10 is a block diagram illustrating machine learning in an estimation unit. 1 is a flowchart of a golf swing analysis method according to an embodiment of the present invention. 10 is a flowchart showing a modified example of the estimation step. 10 is a graph showing the correlation between a spectrogram and left and right impact points. FIG. 1 is a block diagram illustrating an example of machine learning.

Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings.
The specific configurations shown in the embodiments and drawings are for the purpose of understanding the contents of the present disclosure, and the present disclosure is not limited to the specific configurations shown. Furthermore, in multiple embodiments, the same or common elements are designated by the same reference numerals throughout the specification, and redundant explanations will be omitted.

Figure 1 shows an overall configuration diagram of a golf swing analysis system 1 according to this embodiment. Figure 2 shows a block configuration diagram of the analysis system 1 according to this embodiment. By analyzing the swing of a golfer 3, the analysis system 1 according to this embodiment is useful for, for example, effectively fitting a golf club 2 or giving lessons to the golfer 3.

As shown in Figures 1 and 2, the analysis system 1 of this embodiment is for analyzing a golf swing by a golfer 3 using a golf club 2, and includes, for example, a swing sensor 100, a trajectory sensor 200, and a processing device 300.

[Golf Clubs]
The golf club 2 has, for example, a grip 21, a shaft 22, and a head 23. In this embodiment, a wood-type golf club is shown as the golf club 2. In other embodiments, the golf club 2 may be an iron-type or a putter-type club.

[Swing sensor]
The swing sensor 100 can measure first data, which is time-series data of at least one of the angular velocity and acceleration of the golf club 2 during a golf swing. As shown in FIG. 2 , the swing sensor of this embodiment includes an angular velocity sensor 101 and an acceleration sensor 102. Therefore, the swing sensor 100 of this embodiment can measure, as the first data, time-series data of both the angular velocity and acceleration of the golf club 2 during a golf swing. In another aspect, the swing sensor 100 may measure, as the first data, only one of the angular velocity and acceleration of the golf club 2 during a golf swing.

The swing sensor 100 is preferably one that can be attached to the golf club 2, as shown in FIG. 1, for example. In a preferred embodiment, the swing sensor 100 of this embodiment is detachably attached to the grip 21 or shaft 22 of the golf club 2. The swing sensor 100 is preferably small and lightweight so as not to interfere with the swing motion of the golfer 3.

The swing sensor 100 of this embodiment measures first data relating to a local coordinate system defined for the golf club 2. This coordinate system has orthogonal x-, y-, and z-axes. The z-axis coincides with the axial direction of the shaft 22, with the direction from the head 23 toward the grip 21 considered positive. The y-axis is the direction along the target ball flight line when the golf club 2 is addressed, and can also be considered the front-to-back direction of the head 23 (face-to-backface direction). On the y-axis, the direction toward the front (ball flight direction) is considered positive. The x-axis is perpendicular to the y- and z-axes, with the direction from the heel side toward the toe side of the head 23 considered positive.

The swing sensor 100 of this embodiment can measure the angular velocity and acceleration around the three axes as the first data. Figures 3 and 4 are graphs showing the results of measuring the angular velocity and acceleration around the three axes using the angular velocity sensor 101 and acceleration sensor 102, respectively, for the swing of a certain golfer 3. The points where large changes occur correspond to the moment the ball is hit. The first data is acquired in time series at a predetermined sampling period (e.g., 1 millisecond).

As shown in FIG. 2, the swing sensor 100 of this embodiment may further include a communication unit 103 for providing the first data to the processing device 300. The communication unit 103 is preferably wireless so as not to interfere with the swing motion, but may also be wired using a cable. The swing sensor 100 may also include a removable storage medium (not shown) instead of or in addition to the communication unit 103. In this case, the first data may be imported into the processing device 300 via the storage medium.

[Ballistic Sensor]
In the analysis system 1 of this embodiment, a golfer 3 is scheduled to actually hit a ball 4 with a golf swing. The trajectory sensor 200 of this embodiment is capable of measuring second data, which is data related to the trajectory of the ball hit with the golf swing to be analyzed. In this specification, "data related to the trajectory of the ball" is sufficient as long as it specifies the motion state of the ball 4 hit with a golf club, and includes, for example, at least one, and preferably multiple, of ball speed, launch angle, swing angle, backspin, and side spin.

In this embodiment, the trajectory sensor 200 is installed, for example, on the opposite side of the target ball flight line from the golfer 3. The trajectory sensor 200 incorporates, for example, a camera and a control device (not shown). The camera captures an image of the struck ball 4, and the control device analyzes the captured image data to measure the second data. Various commercially available trajectory measurement devices can be used as the trajectory sensor 200, with suitable devices being small and portable, such as "SkyTrak" (a registered trademark of SkyTrak, LLC).

As shown in FIG. 2, the trajectory sensor of this embodiment includes a ball speed measurement unit 201 for measuring the speed of the ball 4, a launch angle measurement unit 202 for measuring the launch angle of the ball 4, a swing angle measurement unit 203 for measuring the left/right swing angle of the ball 4 relative to the target flight line, a backspin measurement unit 204 for measuring the backspin of the ball 4, and a side spin measurement unit 205 for measuring the side spin of the ball 4. As such, the trajectory sensor 200 of this embodiment represents the most preferable one capable of measuring all five pieces of data related to the ball trajectory exemplified above. In other aspects, the trajectory sensor 200 is only required to measure at least one of the above five pieces of data, and preferably measures two or more.

The ballistic sensor 200 of this embodiment further includes a communication unit 206 for providing second data to the processing device 300. The communication unit 206 of this embodiment is wireless, but may also be wired using a cable. Furthermore, the ballistic sensor 200 may include a removable storage medium (not shown) instead of or in addition to the communication unit 206. In this case, the second data may be imported into the processing device 300 via the storage medium.

[Processing device]
The processing device 300 of this embodiment receives the first data and the second data and executes a predetermined process. The processing device 300 may be implemented as, for example, various general-purpose personal computers, tablet computers, smartphones, or dedicated terminals. In this embodiment, a desktop computer is illustrated in FIG. 1 .

The processing device 300 of this embodiment includes a display as the display unit 304, and a keyboard and mouse as the input unit 305. The display unit 304 appropriately displays the results of operations performed by the processing device 300. The input unit 305 can also be used to input information required to operate the processing device 300 and instructions required for the displayed results.

As shown in FIG. 2, the processing device 300 of this embodiment further includes a control unit 301, a memory unit 306, a feature calculation unit 302, an estimation unit 303, and a communication unit 307.

The communication unit 307 can receive the first data from the communication unit 103 of the swing sensor 100 and the second data from the communication unit 206 of the trajectory sensor 200. In this embodiment, the communication unit 307 is wireless, but it may also be wired using a cable. Furthermore, the processing device 300 may be equipped with a removable storage medium (not shown) instead of or in addition to the communication unit 307. In this case, the feature amount and second data may be imported into the processing device 300 via the storage medium.

The memory unit 306 includes, for example, a magnetic memory unit or non-volatile memory that stores programs for causing the processing device 300 to execute predetermined processes, and volatile memory for temporarily storing work information. The control unit 301 executes predetermined processes in accordance with the programs stored in the memory unit 306. The control unit 301 can also store the processing results in the memory unit 306 or display them on the display unit 304.

[Feature Calculation Unit]
The feature calculation unit 302 can calculate predetermined feature amounts from the first data. The feature calculation unit 302 of this embodiment receives the first data as shown in FIGS. 3 and 4 , calculates feature amounts, and stores the calculated feature amounts in, for example, the storage unit 306. The feature calculation unit 302 is realized, for example, by causing the control unit 301 to function in accordance with a program. In another aspect, the feature calculation unit 302 may be implemented by various dedicated processors or the like.

The above-mentioned feature quantity may be, for example, various indices related to the swing of the golfer 3. Generally, when the swing of the golf club 2 changes, the impact position of the ball 4 by the head 23 also changes. Furthermore, the golf club 2 generates unique vibrations for each impact position by the head 23. Therefore, the vibration of the golf club 2 that occurs after hitting the ball is a feature quantity related to the golf swing. In this embodiment, taking note of this situation, a spectrogram is used as the feature quantity of the first data (angular velocity and acceleration of the golf club 2).

A spectrogram is the result of calculating the frequency spectrum of the first data, which is time-series data. The spectrogram of this embodiment includes intensity as a signal component of the frequency of the angular velocity of the golf club 2. Figure 5 shows an example of such a spectrogram. As shown in the upper part of Figure 5, the feature quantity of this embodiment includes a spectrogram (intensity) of the angular velocity of the golf club 2 around the x-axis, y-axis, and z-axis of the local coordinate system defined for the golf club 2. In the upper part of Figure 5, each spectrogram is a three-dimensional graph with the horizontal axis representing time, the vertical axis representing frequency, and the intensity of the color representing the frequency intensity. Darker colors indicate greater frequency intensity.

The range of application of the spectrogram is, for example, a predetermined time range that includes the time before and after the moment when the golf club 2 hits the ball 4. As shown in Figure 3, the time when the ball 4 is hit is when, for example, a large change (peak) appears in the angular velocity of the golf club 2. Therefore, the range for calculating the spectrogram is extracted as a time range t1 of several milliseconds before and after this peak time, based on the time of this peak.

To analyze a golf swing, at least one feature of the first data is sufficient. To further improve the accuracy of the analysis, and as a preferred aspect, this embodiment further includes a spectrogram containing phase as a signal component of the frequency of the angular velocity of the golf club 2 as a feature of the first data. An example of such a spectrogram is shown in the bottom part of Figure 5.

The lower part of Figure 5 shows a spectrogram (phase) of the angular velocity of the golf club 2 around the x-axis, y-axis, and z-axis of the local coordinate system defined for the golf club 2. Generally, as the impact position between the head 23 and the ball 4 changes, the head 23 behaves by opening and closing, for example, around the axis of the shaft 22. This behavior is expressed as a change in the phase (phase angle) of the vibration of the golf club 2. Therefore, by adding a spectrogram that shows the time change in the phase of the frequency as a feature, the accuracy of golf swing analysis can be further improved.

In the lower part of Figure 5, each spectrogram is a three-dimensional graph with the horizontal axis representing time, the vertical axis representing frequency, and the intensity of the color representing phase. Darker colors indicate larger phases. Furthermore, the time range of this spectrogram, like the upper part of Figure 5, is a predetermined time range t1 that includes both the time before and the time when the golf club 2 strikes the ball 4.

In this embodiment, since the first data also includes time-series data on the acceleration of the golf club 2 during a swing, the feature calculation unit 302 also calculates a spectrogram (intensity, phase) for the acceleration time-series data. This increases the number of explanatory variables when performing regression analysis of the features of a golf swing, and ultimately further improves the accuracy of golf swing analysis.

The upper part of Figure 6 shows a three-dimensional spectrogram graph of the acceleration of the golf club 2, with the horizontal axis representing time, the vertical axis representing frequency, and the color intensity representing intensity. The lower part of Figure 6 shows a three-dimensional spectrogram graph of the acceleration of the golf club 2, with the horizontal axis representing time, the vertical axis representing frequency, and the color intensity representing phase.

As described above, the feature calculation unit 302 of this embodiment calculates feature quantities of 12 components from the first data obtained from one golf swing. However, the present disclosure is not limited to this configuration. For example, if the first data includes time series data of only the angular velocity of the golf club 2, only the intensity spectrogram (three components around the x, y, and z axes) may be used as the feature quantity, or only the phase spectrogram (three components around the x, y, and z axes) may be used. Similarly, if the first data includes time series data of only the acceleration of the golf club 2, only the intensity spectrogram (three components around the x, y, and z axes) may be used as the feature quantity, or only the phase spectrogram (three components around the x, y, and z axes) may be used.

[Estimation part]
The estimation unit 303 of this embodiment can estimate the initial conditions of the head 23 using the feature amount calculated by the feature amount calculation unit 302 and the second data.

In this specification, "initial head conditions" refers to information that specifies the state of the head 23 during a swing or the state of impact with the ball, and includes, for example, at least one of the following: head speed, face angle, approach angle, dynamic loft angle, blow angle, left/right impact point, and up/down impact point. Details of these are as follows:

"Head speed" refers to the speed F of the head 23 during a swing just before striking the ball. This speed F is defined by the velocities Fx, Fy, and Fz in the directions of the axes of an X-Y-Z coordinate system defined in a three-dimensional swing space. The X axis is horizontal and aligns with the target ball flight direction D, the Z axis is vertical, and the Y axis is perpendicular to both the X and Z axes. "Face angle" refers to the angle α of the face 23a relative to the vertical plane VP, which is perpendicular to the target ball flight direction D, when viewed from the top of the swing. "Entry angle" refers to the angle β of the movement path K of the head 23 during a swing relative to the target ball flight direction D, when viewed from the top of the swing. As shown in Figure 8, "dynamic loft angle" refers to the angle γ of the face 23a relative to the vertical plane VP when viewed from the side of the swing. "Blow angle" refers to the angle δ of the movement path K of the head 23 during a swing relative to the horizontal plane HP when viewed from the side of the swing. Of these initial head conditions, the "face angle α" and "dynamic loft angle γ" change during the golf swing of the head 23, so they can be defined as values at a specific position of the head 23 during the swing.

Also, as shown in Figure 9, the "left and right impact points" and "up and down impact points" refer to the positions where the ball 4 is struck by the face 23a of the head 23, and respectively refer to the X and Y coordinate values in an XY coordinate system with the area center of gravity FC of the face 23a as the origin.

Conventional golf swing estimation devices rely on data measured by a sensor attached to the golf club 2, but the analysis system 1 of this embodiment estimates the initial conditions of the head by additionally taking into account second data, i.e., data related to the trajectory of the ball 4 hit by the golf swing.

A golf swing affects the bending and twisting of the shaft 22, the movement speed of the head 23, the orientation and angle of the face 23a, and the position of the face 23a, and these effects are reflected in data related to the trajectory of the struck ball 4, i.e., ball speed, launch angle, swing angle, backspin, and side spin. Therefore, by using both the feature quantities of the first data measured from the golf swing and the second data measured from the trajectory of the ball 4 affected by the first data, it is possible to accurately analyze the swing without considering the characteristics of the shaft 22 or the head 23. Furthermore, with the analysis system 1 of this embodiment, there is no need to directly observe the behavior of the head 23 of the golf club 2 during the swing, and therefore the initial conditions of the head can be accurately estimated simply and with high precision.

The estimation unit 303 is preferably configured using AI, for example. In this embodiment, the estimation unit 303 employs a trained machine learning model. An example is shown in Figure 10.

As shown in FIG. 10, for example, the feature quantities of the first data, the second data, and the initial conditions of the head are input to the estimation unit 303 as learning data. Then, by using regression or the like in which the initial conditions of the head are the objective variable and the feature quantities of the first data and the second data are the explanatory variables, a machine learning model that outputs the initial conditions of the head is obtained. For the regression, either a linear regression model or a nonlinear regression model may be adopted, taking into account fitting accuracy.

The learning data is prepared by collecting impact data from multiple golf swings. Each piece of data is created as a set for each golf swing, including the features of the first data (spectrogram), the second data, and the initial conditions of the head (e.g., left/right impact point). The initial conditions of the head can be obtained in advance, for example, by capturing an image of the golf club 2 during a golf swing with a high-resolution camera.

Furthermore, by changing the initial head conditions of the learning data to any of head speed, face angle, approach angle, dynamic loft angle, blow angle, or upper or lower impact point, an inference model for estimating each of these can be obtained. In the analysis system 1 of this embodiment, the feature quantities of the first data and the second data of the golf swing performed by golfer 3 are input to the estimation unit 303 using a machine learning model, making it possible to estimate the desired initial head conditions.

[How to analyze a golf swing]
11 is a flowchart illustrating a method for analyzing a golf swing according to this embodiment. The method for analyzing a golf swing according to this embodiment includes step S1 of measuring first data, which is time-series data of at least one of the angular velocity and acceleration of the golf club 2 during the golf swing, step S2 of measuring second data, which is data related to the trajectory of the ball 4 hit by the golf swing, and processing step S3 of executing predetermined processing using the first data and second data. Steps S1 and S2 can be performed by the swing sensor 100 and trajectory sensor 200, respectively, as already described.

The processing step S3 also includes a calculation step S31 in which a predetermined characteristic amount is calculated from the first data, and an estimation step S32 in which the initial conditions of the head 23 are estimated using the characteristic amount and the second data. These steps S31 and S32 can be performed by the processing device 300 as already described.

[Modification of the estimation step]
12 shows a modified example of the estimation step S32. In this modified example, impact point information of the head 23 and information other than the impact point information are estimated as the initial conditions of the head 23. In other words, this modified example can estimate at least two initial head conditions. Note that "impact point information" refers to the left/right impact point and/or the up/down impact point described above. Furthermore, "information other than impact point information" refers to any one of the head speed, face angle, approach angle, dynamic loft angle, and blow angle.

In this example, the estimation step S32 includes a first estimation step S321 in which impact point information is estimated using the feature amounts of the first data and the second data, and a second estimation step S322 in which information other than the impact point information is estimated using the impact point information estimated in the first estimation step S321, the feature amounts of the first data, and the second data. Therefore, this modified example further takes into account information that has already been estimated when estimating the initial conditions of the head. This further improves the accuracy of estimating information other than the impact point information. Note that the second estimation step S322 employs a machine learning model trained separately from the first estimation step S321. This machine learning model is obtained by training the model to estimate the desired head initial conditions from the impact point information and information other than the impact point information.

Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the specific disclosure above and can be implemented in various modifications within the scope of the technical concept described in the claims.

A more specific example of the present disclosure will be described below, but please note that the present disclosure is not limited to such an example. The main specifications of this example are as follows:

[Initial conditions of the head]
The initial conditions for the head were head speed, face angle, approach angle, dynamic loft angle, blow angle, left/right impact point, and up/down impact point, and machine learning models for each were constructed.

[First Data]
The angular velocity and acceleration of the golf club during the golf swing were measured around three axes of the local coordinate system.

[Feature Amount of First Data]
Spectrograms (intensity, phase) were calculated for each of the angular velocity and acceleration of the first data (12 components in total).

[Second Data]
As the second data, five components of ball speed, launch angle, deflection angle, backspin, and side spin were measured.

[Machine learning model for estimation]
As an example, a machine learning model for inferring left and right hitting points is described.
First, a correlation analysis is performed between the feature amounts of the first data, the second data, and the left/right impact points as the initial conditions of the head. This correlation analysis is performed for each of the 12 spectrogram components of the feature amounts of the first data. Figure 13 is a graph showing an example of the analysis results. The graph in Figure 13 has frequency on the vertical axis, time on the horizontal axis, and p-value on the hue, with 0 on the horizontal axis representing the moment the ball is hit.

In Figure 13, it can be seen that the two regions enclosed by ellipses (for example, the region of frequencies of approximately 10-40 Hz and 50-100 Hz immediately after hitting the ball) have low p-values. Therefore, based on the analysis results in Figure 13, several pieces of data with low p-values are extracted as primary training data, as they are highly correlated with left and right impact points. In this way, by selecting training data based on p-values, it is possible to extract high-quality training data that is effective for estimating the initial head conditions (left and right impact points), thereby enabling efficient machine learning.

Next, as shown in Figure 14, regression is performed using the training data selected by the p-value. In the regression, the feature quantities of the narrowed-down first data and the second data are used as explanatory variables, and the initial condition of the head (here, the left/right impact point) is used as the target variable. In this example, a machine learning model that estimates the initial condition of the head (left/right impact point) is defined using lasso regression, a method of regularized linear regression. This regression excludes explanatory variables with a regression coefficient of 0, and adopts explanatory variables with a regression coefficient other than 0. This helps remove explanatory variables that do not contribute to estimating the left/right impact point and prevent overlearning.

In addition, in this embodiment, after regularized linear regression, a final machine learning model for left and right impact points was set using GPR (Gaussian process regression), a nonlinear regression method. In GPR, a new machine learning model is set that estimates the initial condition of the head (here, left and right impact points), which is the objective variable, using the feature quantities of the first data further narrowed down by lasso regression and the second data as explanatory variables. Note that if the machine learning model obtained by lasso regression is evaluated to have higher estimation accuracy than the machine learning model obtained by GPR, the machine learning model obtained by lasso regression may be adopted.

The same procedure was used to set the machine learning models for the initial conditions of the other heads.

Table 1 shows the validation error (RMSE) for left and right impact points, face angle, and approach angle as representative examples of the initial conditions of the head estimated in this example. The validation error was calculated using the k-fold intersection method (k = 173). As a comparative example, the results of an analysis system (comparative example) that captures images of the golf club head during a swing with a camera and calculates the initial conditions of the head by analyzing the image data are also shown.

As is clear from Table 1, this embodiment was confirmed to be capable of maintaining the same level of analytical accuracy as conventional analysis systems using camera data, with a simple configuration and without the need for large-scale equipment.

[The present disclosure]
The present disclosure includes the following:

The present disclosure (1) provides a system for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
a swing sensor that measures first data that is time-series data of at least one of angular velocity and acceleration of the golf club during a golf swing;
a trajectory sensor for measuring second data relating to the trajectory of the ball hit by the golf swing;
a processing device that receives the first data and the second data and executes a predetermined process;
The processing device includes:
a feature amount calculation unit that calculates a predetermined feature amount from the first data;
an estimation unit that estimates an initial condition of the head during the golf swing using the feature amount and the second data,
A golf swing analysis system.

The present disclosure (2) is the golf swing analysis system described in the present disclosure (1), wherein the feature is a spectrogram of the first data.

The present disclosure (3) is a golf swing analysis system according to the present disclosure (2), wherein the spectrogram includes the intensity of at least one of the angular velocity and acceleration frequencies.

The present disclosure (4) is a golf swing analysis system according to the present disclosure (2) or (3), wherein the spectrogram includes the phase of at least one of the angular velocity and acceleration.

The present disclosure (5) is a golf swing analysis system in any combination with any of the present disclosures (1) to (4), in which the estimation unit is a trained machine learning model.

The present disclosure (6) is the golf swing analysis system described in the present disclosure (5), wherein the machine learning model is trained to output the initial conditions of the head from the feature amount and the second data.

The present disclosure (7) is a golf swing analysis system according to the present disclosure (5) or (6), in which the machine learning model is a linear regression model or a nonlinear regression model.

The present disclosure (8) is a golf swing analysis system in any combination with any of the present disclosures (1) to (7), wherein the second data includes at least one of ball speed, launch angle, swing angle, backspin, and side spin of the hit ball.

The present disclosure (9) is a golf swing analysis system that can be arbitrarily combined with any of the present disclosures (1) to (8), in which the initial conditions of the head include at least one of head speed, face angle, approach angle, dynamic loft angle, blow angle, left/right impact point, and up/down impact point.

The present disclosure (10) provides a method for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
measuring first data which is time series data of at least one of angular velocity and acceleration of the golf club during a golf swing;
measuring second data relating to the trajectory of the ball hit by the golf swing;
a processing step of executing a predetermined process using the first data and the second data,
The processing step includes:
a calculation step of calculating a predetermined feature amount from the first data;
an estimation step of estimating an initial condition of the head using the feature amount and the second data,
This is a method for analyzing a golf swing.

The present disclosure (11) is the golf swing analysis method described in the present disclosure (10), wherein the estimation step is performed by a machine learning model that has learned the initial conditions of the head from the feature amount and the second data.

The present disclosure (12) is a method for estimating information on a hitting point of the head and information other than the hitting point information as the initial conditions of the head,
the estimation step includes a first estimation step of estimating the hitting point information using the feature amount and the second data;
The golf swing analysis method according to the present disclosure (10) or (11) includes a second estimation step of estimating information other than the impact point information using the impact point information estimated in the first estimation step, the feature amount, and the second data.

The present disclosure (13) provides a processing device for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
a communication unit for receiving first data, which is time-series data of at least one of the angular velocity and acceleration of the golf club during a golf swing, and second data, which is data relating to the trajectory of a ball hit by the golf swing;
a feature amount calculation unit that calculates a predetermined feature amount from the first data;
an estimation unit that estimates an initial condition of the head during the golf swing using the feature amount and the second data,
A processing device for analyzing a golf swing.

The present disclosure (14) provides a program for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
receiving first data, which is time series data of at least one of angular velocity and acceleration of the golf club during a golf swing, and second data, which is data relating to the trajectory of a ball hit by the golf swing;
a calculation step of calculating a predetermined feature amount from the first data;
an estimation step of estimating an initial condition of the head during the golf swing using the feature amount and the second data;
This is a program for analyzing golf swings that causes a computer to execute the above.

REFERENCE SIGNS LIST 1 Analysis system 2 Golf club 4 Ball 21 Grip 22 Shaft 23 Head 23a Face 100 Swing sensor 200 Trajectory sensor 300 Processing device 302 Feature amount calculation unit 303 Estimation unit

Claims (13)

A system for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
a swing sensor that measures first data that is time-series data of at least one of angular velocity and acceleration of the golf club during a golf swing;
a trajectory sensor for measuring second data relating to the trajectory of the ball hit by the golf swing;
a processing device that receives the first data and the second data and executes a predetermined process;
The processing device includes:
a feature amount calculation unit that calculates a predetermined feature amount from the first data;
an estimation unit that estimates the initial condition by regression using an initial condition of the head during the golf swing as a response variable and the feature amount and the second data as explanatory variables ,
The initial condition includes at least a position where the face of the head hits the ball.
Golf swing analysis system. The golf swing analysis system of claim 1, wherein the feature is a spectrogram of the first data. The golf swing analysis system of claim 2, wherein the spectrogram includes the intensity of at least one of the angular velocity and acceleration frequencies. The golf swing analysis system of claim 2 or 3, wherein the spectrogram includes the phase of at least one of the angular velocity and acceleration. The golf swing analysis system of any one of claims 1 to 4, wherein the estimation unit is a trained machine learning model. The golf swing analysis system of claim 5, wherein the machine learning model is trained to output the initial conditions of the head from the feature amount and the second data. The golf swing analysis system of claim 5 or 6, wherein the machine learning model is a linear regression model or a nonlinear regression model. The golf swing analysis system according to any one of claims 1 to 7, wherein the second data includes at least one of ball speed, launch angle, swing angle, backspin, and side spin of the hit ball. 9. The golf swing analysis system according to claim 1, wherein the initial conditions of the head include at least one of a face angle, an approach angle, a dynamic loft angle, a blow angle, a left/ right impact point as the impact position , and an up /down impact point as the impact position . A method for analyzing a golf swing using a golf club having a grip, a shaft, and a head, comprising:
measuring first data which is time series data of at least one of angular velocity and acceleration of the golf club during a golf swing;
measuring second data relating to the trajectory of the ball hit by the golf swing;
a processing step of executing a predetermined process using the first data and the second data,
The processing step includes:
a calculation step of calculating a predetermined feature amount from the first data;
an estimation step of estimating an initial condition of the head using the feature amount and the second data ,
the estimation step estimates impact point information of the head and information other than the impact point information as the initial conditions of the head,
the estimation step includes a first estimation step of estimating the hitting point information using the feature amount and the second data;
a second estimating step of estimating information other than the hit point information by using the hit point information estimated in the first estimating step, the feature amount, and the second data,
How to analyze your golf swing. The golf swing analysis method of claim 10, wherein the estimation step is performed by a machine learning model that learns the initial conditions of the head from the feature amount and the second data. A processing device for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
a communication unit for receiving first data, which is time-series data of at least one of the angular velocity and acceleration of the golf club during a golf swing, and second data, which is data relating to the trajectory of a ball hit by the golf swing;
a feature amount calculation unit that calculates a predetermined feature amount from the first data;
an estimation unit that estimates the initial condition by regression using an initial condition of the head during the golf swing as a response variable and the feature amount and the second data as explanatory variables,
The initial condition includes at least a position where the face of the head hits the ball.
Processing device for analysis of golf swing. A program for analyzing a golf swing using a golf club having a grip, a shaft, and a head,
receiving first data, which is time series data of at least one of angular velocity and acceleration of the golf club during a golf swing, and second data, which is data relating to the trajectory of a ball hit by the golf swing;
a calculation step of calculating a predetermined feature amount from the first data;
an estimation step of estimating the initial condition by regression using an initial condition of the head during the golf swing as a response variable and the feature amount and the second data as explanatory variables, the initial condition including at least a position where the ball is hit with a face of the head;
A program for analyzing golf swings that causes a computer to execute the above .