Motion action analysis system and motion action analysis method

Abstract

This invention provides a motion analysis system, a motion analysis method, and a computer program product. In the motion analysis system (100), the input unit (62) inputs motion data corresponding to various motion parameters representing the motion state of a running user. The selection unit (72) selects a type of clothing or accessories corresponding to a given runner type from a variety of clothing or accessories, based on which runner type the user's motion data is pre-classified into through cluster analysis of multiple runner types. The output unit (74) outputs the selected type of clothing or accessories as recommended clothing or accessories for the user.

Description

Technical Field

[0001] This invention relates to a technique for analyzing motion, and more particularly to a motion analysis system, a motion analysis method, and a computer program product. Background Technology

[0002] In recent years, with the increasing health awareness of people and the advancement of various measurement technologies using information terminals and wearable devices, runners have been able to easily acquire and record running information (such as time, distance, altitude, temperature, heart rate, cadence, or stride) and use it to analyze performance (for example, see Patent Document 1).

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2018-126360 Summary of the Invention

[0006] The problem that the invention aims to solve

[0007] To improve runners' performance (records) and reduce the risk of accidents (running injuries), ideally, runners should choose products such as shoes or training methods such as strength training / exercise programs that best suit their individual needs. However, when purchasing shoes or deciding on training methods, runners often rely on recommendations from acquaintances or information published in magazines, books, or online. It's difficult to objectively evaluate whether the chosen products truly suit their running style. Even if the shoes don't feel particularly awkward when actually worn, runners often struggle to determine if they contribute to improved performance. Especially given the current market availability of numerous product models, there may be models more suitable for the individual runner.

[0008] The present invention was made in view of this problem, and its object is to provide a technology for appropriately analyzing motion movements to bring useful information to the user.

[0009] Technical means to solve the problem

[0010] To address the aforementioned problem, a motion analysis system according to an embodiment of the present invention includes: an input unit that inputs motion data corresponding to various motion parameters representing the motion state of a running user; a selection unit that selects a type of clothing from various clothing items corresponding to a runner type based on a pre-classified cluster analysis of multiple runner types based on the user's motion data; and an output unit that outputs the type of clothing selected by the selection unit as recommended clothing items for the user to wear.

[0011] "Apparel" can refer to running shoes or compression pants worn during exercise, especially running. "Motion parameters" can include parameters describing the running motion, such as cadence, ground contact time, ground contact rate, vertical stiffness, vertical movement, landing impact, kickoff acceleration, sinking, kickoff time, braking force, lateral impact, pelvic rotation, and braking time. "Indicator parameters" can include parameters describing running performance, such as heart rate, oxygen uptake, running time, running distance, speed, pace, presence of pain, and presence of fatigue.

[0012] Through this model, the user's runner type can be objectively determined by action data corresponding to various action parameters representing the user's action state, and more appropriate shoes can be recommended based on the objectively classified runner type. Furthermore, by using cluster analysis to classify runner types, a non-human and non-uniform classification method can be used to objectify runner types, allowing for more appropriate shoe selection based on this objectively classified runner type.

[0013] A model storage unit may also be provided, which stores a predictive model of the runner type generated by machine learning using motion data of multiple people who will be classified into multiple runner types as teacher data. An inference unit may also be provided to infer the runner type by inputting the user's motion data into the predictive model.

[0014] It can also be set up with: an analysis unit, which performs cluster analysis on the motion data of multiple people and classifies them into multiple runner types; and a model generation unit, which, for each of the multiple runner types, uses the motion data of multiple people classified into that runner type as teacher data to perform machine learning to generate a predictive model.

[0015] A project storage unit may also be provided, which stores types of apparel corresponding to runner types. These types of apparel possess the characteristic that, by altering various motion parameters in motion data categorized into multiple runner types, they contribute to improving athletic performance. The selection unit can then select, based on the information stored in the project storage unit, the apparel corresponding to the runner type to which the user's motion data belongs from a variety of apparel.

[0016] The input unit also inputs indicator data corresponding to the indicator parameters representing the user's athletic performance during running. The estimation unit, based on the following multiple regression analysis model, uses multiple action parameters from the action data of multiple people classified as a runner type as explanatory variables and indicator parameters from the indicator data of multiple people classified as runner types as target variables to calculate the contribution of each action parameter to improving athletic performance. The estimation unit estimates which action parameter contributes more to improving athletic performance for users of the runner type. The selection unit selects from a variety of clothing and equipment that have the characteristic of contributing to improving athletic performance by changing the estimated action parameters.

[0017] Another embodiment of the present invention is a motion analysis system. The motion analysis system includes: an input unit for inputting motion data corresponding to various motion parameters representing the motion state of a running user; an analysis unit for performing cluster analysis on the motion data of multiple users and classifying it into multiple runner types; a model generation unit for generating a predictive model by using the motion data of multiple users classified into the multiple runner types as teacher data for machine learning; an estimation unit for estimating the runner type by inputting the user's motion data into the predictive model; and an output unit for outputting information corresponding to the tendency of the values ​​of the motion parameters in the estimated runner type. "Information corresponding to the tendency of the values ​​of the motion parameters in the estimated runner type" refers to information beneficial to users seeking to improve athletic performance. For example, it may first include information for reference in the selection of clothing and equipment such as shoes, and may also include training methods or competition information. As training methods, it may include a menu of running exercises with specified pace, distance, and route for speed training, hill training, etc., or it may include intensive exercises other than running such as muscle strength training or stretching, or repetitive practice for mastering movements.

[0018] Another embodiment of the present invention is a motion analysis method. The method includes: inputting motion data corresponding to various motion parameters representing the motion state of a running user using a predetermined information acquisition unit; selecting, through predetermined computer processing, a type of clothing corresponding to the runner type from a variety of clothing items, based on which runner type the user's motion data is pre-classified among multiple runner types according to cluster analysis based on multiple users' motion data; and outputting, using a predetermined information output unit, the selected type of clothing item as recommended clothing items for the user to wear.

[0019] Another embodiment of the present invention is a computer program product, comprising a computer program that, when executed by a computer, enables the computer to perform the following functions: inputting motion data corresponding to various motion parameters representing the motion state of a running user; selecting, from various types of clothing and accessories, a type of clothing and accessories corresponding to the runner type based on which runner type the user's motion data is pre-classified among multiple runner types based on cluster analysis of multiple people's motion data; and outputting the selected type of clothing and accessories as recommended clothing and accessories for the user to wear.

[0020] Furthermore, any combination of the above-mentioned constituent elements, or any substitution of the constituent elements or statements of the present invention among methods, apparatus, programs, temporary or non-temporary storage media storing programs, systems, etc., are also valid embodiments of the present invention.

[0021] The effects of the invention

[0022] This invention allows for the appropriate analysis of movement actions, providing users with useful information. Attached Figure Description

[0023] Figure 1 This is a diagram representing the basic structure of a motion analysis system.

[0024] Figure 2 This is a graph illustrating action data and indicator data.

[0025] Figure 3 It is a graph showing the relationship between the product model of shoes, etc. and the changing characteristics of the motion parameters of each product model.

[0026] Figure 4 This is a functional block diagram representing the basic structure of the user terminal and motion analysis server.

[0027] Figure 5 It is a tree diagram representing the clustering analysis of action data using the sum of squared deviations method.

[0028] Figure 6 This is a graph illustrating the regression formula for each runner type.

[0029] Figure 7 This is an example image representing the first screen used to recommend shoes to a user.

[0030] Figure 8 This is an example image of a second screen that recommends shoes to a user.

[0031] Figure 9 This is an example image of a third screen that recommends shoes to a user.

[0032] Figure 10 This is an example image of the fourth screen that recommends shoes to the user.

[0033] Figure 11 This is an example of a screen displaying a suggestion to the user for improvement based on stretching or training.

[0034] Figure 12 This is an example image showing a screen prompting a user to participate in a marathon.

[0035] Figure 13 This is an example image showing a screen displaying a practice menu to the user.

[0036] Figure 14 It is a flowchart illustrating the process of selecting clothing and accessories.

[0037] [Explanation of Symbols]

[0038] 10: User Terminal

[0039] 12: Running Watch

[0040] 14: Motion sensor

[0041] 16: Wearable devices

[0042] 18: Internet

[0043] 20: Motion Analysis Server

[0044] 30, 40, 80, 90, 92, 94, 96, 110, 112, 114: First column

[0045] 32, 42, 82, 91, 93, 95, 97, 111, 113, 115: Second column

[0046] 34, 44: Third column

[0047] 46: Column Four

[0048] 48: Fifth column

[0049] 50: Information Acquisition Department

[0050] 52: Information Processing Department

[0051] 54: Data Storage Department

[0052] 56: Information Display Department

[0053] 58, 60: Ministry of Communications

[0054] 62: Input Section

[0055] 64: Storage Department

[0056] 65: Analysis Department

[0057] 66: Model Generation Department

[0058] 68: Model Storage Department

[0059] 70: Presumption Department

[0060] 72: Selection Department

[0061] 74: Output Department

[0062] 76: Project Storage Department

[0063] 98: First Menu

[0064] 99: Second Menu

[0065] 100: Motion Analysis System

[0066] 101: First Cluster

[0067] 102: Second cluster

[0068] 103: Third cluster

[0069] 104: Fourth cluster

[0070] A001, A002, A003: Shoes

[0071] N001, N002: Attire

[0072] S10, S12, S14, S16, S18: Steps Detailed Implementation

[0073] In this embodiment, based on various information obtained by the wearable device worn by the user while running, the user's runner type is analyzed, and according to the runner type, shoes or functional clothing that can be expected to improve athletic performance (hereinafter, "shoes" and "functional clothing" are also appropriately referred to as "shoes, etc.") are selected from multiple options and recommended to the user.

[0074] Figure 1This describes the basic structure of a motion analysis system. A user runs while wearing a wearable device 16, such as a running watch 12 or motion sensor 14, on their wrist or waist, acquiring various detection data using the running watch 12 or motion sensor 14. The running watch 12 includes a positioning module or sensors such as an accelerometer, heart rate sensor, and barometer, acquiring information such as date and time, location coordinates, altitude, heart rate, temperature, and cadence as detection data. The motion sensor 14 acquires detection data, for example, using a nine-axis sensor. As for the wearable device 16, motion sensors built into portable terminals such as smartphones can also be used, as well as chest strap-type heart rate sensors or motion sensors installed inside or outside shoes.

[0075] The user synchronizes the detection data acquired by the wearable device 16 to the user terminal 10 via near-field communication or other means. Based on the detection data from the wearable device 16, the user terminal 10 acquires running log data such as running date and time, time, distance, altitude, speed, pace, and temperature. Furthermore, it acquires values ​​for various motion parameters representing movement status, such as stride width, cadence, vertical movement ratio, ground contact time, ground contact time balance, landing impact, landing pattern, and pronation, or values ​​for one or more indicators representing exercise performance, such as heart rate and oxygen uptake. The user terminal 10 sends the acquired running log data, motion data corresponding to the motion parameters, or indicator data corresponding to the indicator parameters to the motion analysis server 20 via network 18. The motion analysis server 20 determines the user's runner type based on the motion data received from multiple user terminals 10, and recommends shoes and other suitable items to the user based on the runner type. Additionally, it provides the user with suggestions corresponding to their runner type.

[0076] User terminal 10 can be a personal computer, or an information terminal such as a smartphone or tablet. Motion analysis server 20 can be a server computer. User terminal 10 and motion analysis server 20 can be constructed from a computer including a central processing unit (CPU), a graphics processing unit (GPU), random access memory (RAM), read-only memory (ROM), auxiliary storage devices, communication devices, etc. User terminal 10 and motion analysis server 20 can be constructed from separate computers, or they can be implemented from a single computer or information terminal that combines the functions of both. In this embodiment, an example implemented from a separate computer will be described.

[0077] Figure 2This diagram illustrates motion data and indicator data. The data shown is processed or calculated by the user terminal 10 based on information obtained from the wearable device 16, and is sent to the motion analysis server 20. In a variant, the user terminal 10 may directly transmit the information obtained from the wearable device 16 to the motion analysis server 20, where the data shown in this diagram is processed or calculated.

[0078] The first column (30) displays the date and time the detection data was acquired. The second column (32) displays the motion data. The third column (34) displays the indicator data.

[0079] The second column, 32, contains four motion data parameters: "Landing Impact," "Running Efficiency," "Stability," and "Range of Motion." Here, "Running Efficiency" can represent the kick acceleration. "Stability" can represent the lateral impact. "Range of Motion" can represent the pelvic rotation. In addition, motion data may also include parameters such as "Up-Down Movement," "Rebound" (expressed as vertical rigidity), and "Rhythm" (expressed as stride frequency).

[0080] The third column (34) contains four parameters: "Heart Rate," "Maximum Distance," "Maximum Speed," and "Subjective." Alternatively, "Oxygen Uptake" calculated based on the heart rate can be included instead of "Heart Rate." "Subjective" can be manually entered by the user, representing feelings of urgency, fatigue, or pain during or after running.

[0081] Figure 3 This diagram illustrates the relationship between the product model of shoes, etc., and the variation characteristics of the motion parameters of each product model. The variation characteristics of the motion parameters of each product model shown in this diagram are defined as the wearable equipment characteristic data described later, and used as a reference for selecting shoes, etc.

[0082] The first column, 40, shows the product model name for shoes, etc. The second column, 42, shows the variation in the motion parameter "landing impact." The third column, 44, shows the variation in the motion parameter "running efficiency." The fourth column, 46, shows the variation in the motion parameter "stability." The fifth column, 48, shows the variation in the motion parameter "range of motion."

[0083] As for the product models of shoes and the like in the first column 40, product model names such as "A001", "A002", and "A003" are shown for shoes, and product model names such as "N001" and "N002" are shown for functional clothing such as compression tights or joint braces. The different motion parameters that can be expected to change when wearing these shoes and the like are different. For example, "Shoe A001" shows characteristics of a 10% reduction in "landing impact", a 10% increase in "running efficiency", a 20% increase in "stability", and no change in "range of motion" compared to the benchmark clothing. "Shoe A002" shows characteristics of a 20% reduction in "landing impact", a 20% increase in "running efficiency", a 10% decrease in "stability", and no change in "range of motion" compared to the benchmark clothing. "Shoe A003" shows characteristics of a 10% increase in "landing impact", a 30% increase in "running efficiency", a 10% decrease in "stability", and no change in "range of motion" compared to the benchmark clothing.

[0084] "Outfit N001" shows that compared to the benchmark suit, "landing impact" remains unchanged, "running efficiency" increases by 20%, "stability" decreases by 20%, and "range of motion" increases by 10%. "Outfit N002" shows that compared to the benchmark suit, "landing impact" remains unchanged, "running efficiency" increases by 10%, "stability" also increases by 10%, and "range of motion" decreases by 10%.

[0085] Thus, since the motion parameters that can be expected to change and their degree of change are different in each product model of shoes, for example, it is possible to recommend "Shoe A001" to users who expect to improve their athletic performance when "stability" is improved compared to "landing impact", or to recommend "Outfit N001" to users who expect to improve their athletic performance when "range of motion" is improved compared to "stability".

[0086] Figure 4 This is a functional block diagram representing the basic structure of the user terminal and motion analysis server. This diagram depicts the functional blocks, which can be implemented in various forms by hardware, software, or a combination thereof.

[0087] User terminal 10 includes an information acquisition unit 50, an information processing unit 52, a data storage unit 54, an information display unit 56, and a communication unit 58. The information acquisition unit 50 acquires various detection data from the wearable device 16 via the communication unit 58. Additionally, the information acquisition unit 50 accepts user input and acquires subjective information based on the user, such as the degree of fatigue during or after running, or the presence or absence of pain. Based on the detection data and subjective information acquired by the information acquisition unit 50, the information processing unit 52 generates running log data, motion data, and indicator data, and stores them in the data storage unit 54. The running log data, motion data, indicator data, and user identification information are sent to the motion analysis server 20 via the communication unit 58 and the network 18.

[0088] The motion analysis server 20 includes a communication unit 60, an input unit 62, a storage unit 64, an analysis unit 65, a model generation unit 66, a model storage unit 68, an estimation unit 70, a selection unit 72, an output unit 74, and an item storage unit 76.

[0089] The input unit 62 receives running log data, action data, indicator data, and user identification information sent from the user terminal 10 via the communication unit 60. The input unit 62 establishes a correspondence between the data received from the user terminal 10 and the user identification information of the user terminal 10 and stores it in the storage unit 64.

[0090] The analysis unit 65 performs cluster analysis on the motion data of multiple people stored in the storage unit 64 and classifies them into multiple runner types.

[0091] Figure 5 This is a tree diagram representing the clustering analysis of motion data using the Ward method. In this diagram, an example is shown where users with approximate values ​​for multiple motion parameters are grouped together and stratified, and then divided into four clusters.

[0092] The first cluster, 101, is a "first runner type" consisting of 30 leaves. The movement data classified into this cluster shows movement tendencies such as "cadence type", "long ground contact time", and "strong kicking power".

[0093] The second cluster, 102, is a "second runner type" consisting of 39 leaves. Based on the motion data classified into this cluster, it shows motion tendencies such as "stride length type", "long ground contact time", and "high landing impact".

[0094] The third cluster, 103, is a "third runner type" consisting of 17 leaves. Based on the motion data classified into this cluster, it shows motion tendencies such as "cadence type", "short ground contact time", and "high landing impact".

[0095] The fourth cluster, 104, is a "fourth runner type" consisting of 34 leaves. Based on the motion data classified into this cluster, it shows motion tendencies such as "stride length type", "low landing impact", and "high propulsion".

[0096] Regarding the movement tendencies of each runner type as described above, movement parameters with similar values ​​are extracted from multiple sets of movement data categorized into each runner type, and these particularly numerous similar movement parameters are identified as the movement tendencies of that runner type.

[0097] return Figure 4 The model generation unit 66 generates a predictive model for each of the multiple runner types by using the motion data of multiple people classified as that runner type as teacher data and employing random forest machine learning. The generated predictive model is stored in the model storage unit 68. The estimation unit 70 estimates which runner type the user is classified as by inputting the user's motion data into the predictive model. Furthermore, in a variation, machine learning algorithms based on decision trees or deep forests, other than random forests, may also be used.

[0098] The model generation unit 66 also generates a multiple regression analysis model. For each runner type, this model uses multiple motion parameters from the motion data of multiple runners categorized into that type as explanatory variables, and indicator parameters from the indicator data of multiple runners categorized into the same type as target variables, to determine the contribution of each motion parameter in that runner type to improving athletic performance. The storage unit 64 stores multiple indicator data, each representing athletic performance for different purposes. For each athletic purpose, i.e., each indicator data, the model generation unit 66 generates a separate multiple regression analysis model, using the indicator parameters as target variables and multiple motion parameters as explanatory variables. The generated multiple regression analysis model for each runner type is saved in the model storage unit 68.

[0099] In the multiple regression analysis model for each runner type stored in the model storage unit 68, the estimation unit 70 estimates, for each type of indicator parameter, which of the various action parameters contributes more to improving athletic performance for the user.

[0100] Here, the question of which of the various motion parameters contributes to improving the athletic performance represented by the indicator parameter varies depending on the user or runner type. This is because each user or runner type has different body structure (skeletal or muscular strength) or physical abilities, running posture, and different running purposes, goals, or intentions. Therefore, for each user or runner type, the improvement in which indicator parameter leads to performance improvement also varies. Thus, in this embodiment, motion data and indicator data are stored in storage unit 64 according to different runner types. A stepwise multiple regression analysis is performed on the stored data to identify which motion parameter is most likely to influence athletic performance improvement for each runner type. Furthermore, the types of motion parameters that help improve indicator parameters that align with the user's intentions are identified.

[0101] Figure 6 The regression formulas for each runner type are illustrated. The runner types are shown in column 80 of the first column. The regression formulas for each runner type, derived through multiple regression analysis, are shown in column 82 of the second column. In the example in this figure, a stepwise multiple regression analysis was performed with the indicator parameter "oxygen uptake" as the target variable. The lower the "oxygen uptake" due to changes in any movement parameter, the greater the contribution of that change to improving athletic performance. Based on experiments, the explained rate (adjusted R²) of the indicator parameter estimated using the regression formulas shown in the figure is... 2 The accuracy is approximately 59% to 96%. Therefore, the estimation of the index parameters by the multiple regression analysis of this embodiment shows sufficient precision.

[0102] For example, a regression formula "stability × a + rebound × b + constant c" is obtained based on the motion data and indicator data of the first runner type. Furthermore, as a multiple regression analysis model actually generated by the model generation unit 66 for each runner type, a regression formula is also generated that includes motion parameters other than "stability" or "rebound". In this regression formula, the estimation unit 70 reduces the motion parameters that contribute particularly highly to improving the indicator parameters of the runner type from a variety of motion parameters to one or more equally limited numbers of motion parameters. In this figure, for convenience, a regression formula using only the reduced motion parameters is shown.

[0103] In addition, the coefficient "a" for "stability" is negative. The higher the value of the "stability" movement parameter, such as "left-right impact," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance. The coefficient "b" for "rebound" is positive. The lower the value of the "rebound" movement parameter, such as "vertical rigidity," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance.

[0104] For example, a regression formula “running efficiency × d + vertical movement × e + constant f” is obtained based on the motion data and index data of the second runner type. That is, the estimation unit 70 reduces the motion parameters from a variety of motion parameters to the motion parameter “running efficiency” and the motion parameter “vertical movement”, which are the motion parameters that contribute a particularly high proportion to improving the index parameters of the runner type. The regression formula using only the reduced motion parameters is shown in this figure.

[0105] Furthermore, the coefficient "d" for "running efficiency" is negative. The higher the value of the motion parameters for "running efficiency," such as "kick acceleration," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance. The coefficient "e" for "vertical movement" is positive. The lower the value of the motion parameter for "vertical movement," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance.

[0106] For example, a regression formula “running efficiency × g + lightweight × h + constant i” is obtained based on the motion data and index data of the third runner type. That is, the estimation unit 70 reduces the motion parameters from a variety of motion parameters to the motion parameters “running efficiency” and “lightweight”, which are the motion parameters that contribute particularly highly to improving the index parameters of the runner type. The regression formula using only the reduced motion parameters is shown in this figure.

[0107] Furthermore, the coefficient "g" for "running efficiency" is negative. The higher the value of the movement parameters for "running efficiency," such as "kick-off acceleration," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance. The coefficient "h" for "lightweight" is also negative. The higher the value of the movement parameters for "lightweight," such as "step frequency," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance.

[0108] For example, a regression formula “mobility range × j + landing impact × k + constant m” is obtained based on the motion data and index data of the fourth runner type. That is, the estimation unit 70 reduces the motion parameters from a variety of motion parameters to the motion parameter “mobility range” and the motion parameter “landing impact”, which are the motion parameters that contribute a particularly high proportion to improving the index parameters of the runner type. The regression formula using only the reduced motion parameters is shown in this figure.

[0109] Furthermore, the coefficient "j" for "range of motion" is negative. The larger the value of the motion parameter of "range of motion," such as "pelvic rotation," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance. The coefficient "k" for "landing impact" is positive. The smaller the value of the motion parameter of "landing impact," the more it can reduce the indicator parameter "oxygen uptake," and the more it contributes to improving athletic performance.

[0110] Furthermore, this figure illustrates an example of using multiple regression analysis with "oxygen uptake" as the target variable to estimate the types of movement parameters that contribute to improving athletic performance by reducing "oxygen uptake". The model generation unit 66 further estimates the types of movement parameters that contribute to improving each indicator parameter by using multiple regression analysis with other indicator parameters such as "heart rate", "maximum distance", "maximum speed", "pace reduction rate", and "subjective" as target variables. In the case where the indicator parameter "heart rate" is the target variable, the types of movement parameters that contribute to reducing "heart rate" are estimated.

[0111] When the target variable is the indicator parameter "maximum distance," we can estimate the types of motion parameters that contribute to extending "maximum distance." For example, for users aiming to complete long-distance races such as marathons, being able to run longer distances is beneficial, and this implies improved athletic performance. Conversely, the motion parameter "rebound" might have a negative effect on extending "maximum distance," while a small amount of "rebound" might contribute to improved athletic performance.

[0112] When the target variable is the metric "maximum speed," we can infer the types of motion parameters that contribute to increasing "maximum speed." For example, for a user aiming to speed up, increasing instantaneous speed is beneficial, which translates to improved athletic performance. For instance, the motion parameter "rebound" might have a positive effect on increasing "maximum speed," and is highly likely to contribute to improved athletic performance. This differs from the case where the target is "maximum distance," where the same "rebound" has a negative effect.

[0113] With the target variable being the indicator parameter "pace reduction rate," we can estimate the types of motion parameters that contribute to suppressing "pace reduction rate." For example, for users whose goal is to run long distances consistently, it is beneficial to maintain a consistent pace, which implies improved athletic performance.

[0114] When the "subjective" nature of the indicator parameters is used as the target variable, we can presume the types of movement parameters that contribute to reducing fatigue or physical pain during or after running, or the types of movement parameters that contribute to improving the feeling of speed during or after running. For example, for users whose goal is to maintain training consistency, it is beneficial to be able to run without fatigue or injury, or to run with a pleasant mood, which implies improved athletic performance.

[0115] In this embodiment, a method using multiple regression analysis to estimate motion parameters that contribute to improved athletic performance is employed. However, machine learning can also be used instead of multiple regression analysis. In this case, the teacher data for machine learning can include not only motion parameters and indicator parameters, but also a combination of user attributes such as gender, age, height, weight, and athletic experience, or information such as the type of shoes worn.

[0116] Refer again Figure 4 The following explanation is provided. The project storage unit 76 stores information for recommending various shoes and functional apparel for running to users. Additionally, the project storage unit 76 stores apparel characteristic data, which, for each of the various shoes, pre-determines which characteristic among various motion parameters contributes to improving athletic performance. In the apparel characteristic data, for each product model of the various shoes, the types of motion parameters expected to change by wearing those shoes and the degree of change in those motion parameters are pre-determined as motion parameter change characteristics. Examples of motion parameter change characteristics shown in the apparel characteristic data are as follows: Figure 3 As shown.

[0117] Based on the wearable equipment characteristic data, the selection unit 72 selects shoes, etc., from a variety of shoes, etc., that have characteristics that contribute to improving athletic performance by changing the type of motion parameters estimated by the estimation unit 70. The selection unit 72 also selects shoes, etc., with higher variability in the type of motion parameters estimated by the estimation unit 70, based on the wearable equipment characteristic data. For each indicator data point, the selection unit 72 individually selects shoes, etc., that have characteristics that contribute to improving athletic performance by changing the type of motion parameters estimated by the estimation unit 70 for each user's purpose. Multiple shoes, etc., can be selected, and their recommendation priority can be determined according to the variability from high to low. For example, for all candidate shoes, etc. stored in the model storage unit 68, the selection unit 72 substitutes the variability of each motion parameter into a regression formula to calculate the improvement rate of the indicator parameters, and selects recommended candidate shoes, etc., in order of the improvement rate from high to low.

[0118] For example, in the case of the first runner type, the selection unit 72 selects shoes from a variety of shoes that have characteristics that improve the "stability" of motion parameters and characteristics that reduce the "rebound" of motion parameters. If there are multiple candidates for shoes that possess both characteristics that improve "stability" and reduce "rebound," the selection unit 72 considers the degree of variation of each and comprehensively selects shoes with a high degree of contribution to improving the indicator parameters (i.e., a high rate of improvement in the indicator parameters) as the first candidate, and shoes with a relatively low degree of contribution or rate of improvement as the second or subsequent candidates. For each indicator parameter such as "oxygen uptake," "heart rate," "distance," "speed," and "subjective," the selection unit 72 selects one or more candidates for shoes or similar shoes that have characteristics that contribute to improving the indicator parameters for the user.

[0119] The project storage unit 76 also stores the content of various improvement proposals suggested to the user. Regarding each improvement proposal, text relating to the effect of changing the value of each motion parameter is stored in the project storage unit 76. For example, when the motion parameter "landing impact" (with a positive coefficient) is presumed to contribute highly to the improvement index parameter in the regression formula for each runner type, text such as "If the landing impact during running is small, speed can be achieved with less energy" is stored as the content to be proposed at this time. Similarly, when the motion parameter "stability" (with a negative coefficient) is presumed to contribute highly to the improvement index parameter, text such as "If the stability during running is high, speed can be achieved with less energy" is stored as the content to be proposed at this time. Furthermore, the project storage unit 76 stores the content of improvement proposals for each motion parameter according to the user's different purposes, that is, according to different index parameters.

[0120] The project storage unit 76 also stores various training proposals that are presented to the user. Regarding each training proposal, text related to the training content proposed for each motion parameter and the effect of changes in the value of the motion parameter through the training is stored in the project storage unit 76. Furthermore, the project storage unit 76 stores the training proposal content for each motion parameter according to the user's different purposes, i.e., according to different indicator parameters.

[0121] The project storage unit 76 also stores information on multiple marathons proposed to users with the goal of participation, or various training proposals suggested to users for each marathon as preparation for participation. Within the project storage unit 76, for each marathon, information is also stored indicating which marathon is suitable for which of the various running goals (e.g., "run faster," "run longer," "run more comfortably," etc.), information on the characteristics of the marathon route (e.g., the timing of the event, the slope of the route, seasonal environmental characteristics, etc.), and information on training content corresponding to the characteristics of each marathon.

[0122] Furthermore, in a modified example, the project storage unit 76 may also store types of clothing and accessories that contribute to improved athletic performance by altering the motion parameters in a manner corresponding to the tendency of various motion parameters in motion data classified into multiple runner types, as types of clothing and accessories corresponding to runner types. In this case, recommended shoes, etc., can be selected simply by determining which runner type the user is classified as. That is, the selection unit 72 determines the runner type by inputting the user's motion data into the prediction model, and selects clothing and accessories corresponding to the runner type to which the user's motion data belongs from a variety of clothing and accessories based on the information stored in the project storage unit 76. In addition, the project storage unit 76 may also store types of improvement proposals, types of training proposals, and information about marathon races in a manner corresponding to runner types.

[0123] The output unit 74 sends one or more types of shoes, etc., selected by the selection unit 72 to the user terminal 10 via the communication unit 60 as recommended shoes, etc., for the user. The output unit 74 may, for example, display introductions to the recommended shoes, etc., or suggestions for improving exercise, according to the user's different goals. That is, it displays the types of shoes, etc., selected by the selection unit 72, and the types of motion parameters that are desired to be changed, according to different indicator parameters that serve as indicators of athletic performance.

[0124] Figure 7 This is an example of the first screen showing shoe recommendations to the user. The first column (90) displays the improvement proposals. The second column (91) displays the recommended shoe types in the order they were suggested.

[0125] In the first column (90), which displays the user's tendency in terms of motion parameters, strings such as "Your Tendency," "If the landing impact during running is small, you can achieve speed with less energy," and "If the stability during running is high, you can achieve speed with less energy" are displayed. In other words, the main idea is that reducing the motion parameter "landing impact" and improving the motion parameter "stability" contribute to improving athletic performance by increasing the indicator parameter "maximum speed."

[0126] In the second column 91, as content recommending the types of shoes selected by the user, it displays a string such as "Shoes recommended for you", two types of shoes recommended, the number of stars indicating the degree of recommendation, and strings such as "reduce landing impact" and "improve stability" indicating which action parameters can be improved for each pair of shoes.

[0127] Figure 8 This is an example of a second screen that recommends shoes to the user. The first column (92) displays the improvement proposals. The second column (93) displays the recommended shoe types, etc., in the order they were suggested. In this example, [the text abruptly ends here, likely due to an incomplete sentence or missing information]. Figure 7 Unlike the first screen example, this design allows users to easily choose shoes based on their goals, budget, running ability, and other factors.

[0128] In column 92, as improvement proposals for users, improvement proposals for motion parameters are displayed according to the user's different preferences. For example, the indicators "speed" (for "maximum speed"), "length" (for "maximum distance"), and "pleasure" (for "subjective comfort") are displayed together, along with the improvement proposals for the motion parameters. For users with the goal of "speed" (for "maximum speed"), a string is displayed indicating that improving the motion parameter "running efficiency" contributes to improving athletic performance. For users with the goal of "length" (for "maximum distance"), a string is displayed indicating that improving the motion parameter "stability" contributes to improving athletic performance. For users with the goal of "pleasure" (for "subjective comfort"), a string is displayed indicating that reducing the motion parameter "landing impact" contributes to improving athletic performance.

[0129] In column 93, as part of the recommendation of shoes and other items to the user, a string such as "Shoes Recommended for You" is displayed, along with the two recommended shoe types. The number of stars indicates the level of recommendation, and strings such as "Speed," "Length," and "Enjoy" indicate which performance parameter can be expected to improve for each pair of shoes. For example, for the first candidate shoe, three stars are shown, along with the characters "Length," indicating extended maximum distance, and "Enjoy," indicating improved speed and reduced fatigue or pain. Strings such as "Reduced landing impact" and "Improved stability" indicate which performance parameter can be expected to improve. For the second candidate shoe, two stars are shown, along with the characters "Speed," indicating increased maximum speed, and "Enjoy," indicating improved speed and reduced fatigue or pain. Strings such as "Reduced landing impact" indicate which performance parameter can be expected to improve.

[0130] The first menu 98 is a drop-down menu for specifying the user's budget, and the second menu 99 is a drop-down menu for specifying the user's running ability. When the user specifies their budget in the first menu 98 and their running ability in the second menu 99, the selection unit 72 narrows down the candidate shoes to be displayed to shoes that are suitable for the user's specified budget and running ability level, and then displays them.

[0131] Figure 9 This is an example of a third screen showing shoe recommendations to a user. In this example, the estimated marathon time for the user's current state is calculated and displayed, along with the estimated marathon time after wearing the recommended shoes.

[0132] In the first column 94, the current user's estimated marathon time is displayed, for example, as "3 hours 52 minutes 30 seconds". In the second column 95, the estimated marathon time for each recommended shoe is displayed in the order of shoe recommendations. For example, the first candidate shoe is displayed along with "3 hours 50 minutes 30 seconds", representing the estimated marathon time when wearing this shoe, which is 2 minutes shorter than the current time. The second candidate shoe is also displayed along with "3 hours 51 minutes 30 seconds", representing the estimated marathon time when wearing this shoe, which is 1 minute shorter than the current time. Additionally, a link to the practice menu page, such as "If you further implement this practice menu," is displayed along with the string "3 hours 35 minutes 30 seconds", representing the estimated marathon time after the practice menu, which is 17 minutes shorter.

[0133] In this case, to predict the marathon time, the model generation unit 66 generates a multiple regression analysis model as follows: In this model, running time is input as the target variable, and in addition to the various motion parameters described in this embodiment, parameters representing running status such as distance, speed, pace, temperature, and altitude are further input as explanatory variables. Based on this multiple regression analysis model, the estimation unit 70 estimates the marathon time.

[0134] Figure 10 This is an example of the fourth screen that recommends shoes to the user. In this example, [the text abruptly ends here]. Figure 9 The recommended running volume from the training menu, along with the load caused by that volume and the probability of pain due to the current running posture, are displayed in the first column 96. For example, strings such as "In this training menu, it is planned to run 320 km over 3 months at an average pace of 5:30 / km" and "50% probability of knee pain" indicate the body parts that may experience pain and their probability. Additionally, the probability of pain when wearing recommended shoes is displayed in the second column 97. For example, "30% probability of pain when wearing these shoes" indicates the probability of pain when wearing the recommended shoes.

[0135] Figure 11 This is an example of a screen that presents the user with improvement suggestions through stretching or training. The content of the improvement suggestion is displayed in the first column (110). The stretching and training suggestions are displayed in the second column (111).

[0136] In the first column 110, the user's tendency as a parameter of action is displayed with strings such as "Your tendency", "If the impact of landing when running is small, you can run a long distance continuously", and "If the stability when running is high, you can exert speed with less energy".

[0137] In the second column (111), a marker for "Length" indicating the target parameter "Limited Distance" (aimed at running longer), a marker for "Enjoyment" indicating the target parameter "Subjective" (aimed at running more comfortably), and a stretching menu are provided. This indicates that by engaging in the stretching menu, improvements in both the "Limited Distance" and "Subjective" parameters can be expected. Additionally, a marker for "Speed" indicating the target parameter "Maximum Speed" (aimed at running faster), a marker for "Enjoyment" indicating the target parameter "Subjective" (aimed at running more comfortably), and a muscle strength training menu are provided. This indicates that by engaging in the muscle strength training menu, improvements in both the "Maximum Speed" and "Subjective" parameters can be expected.

[0138] Figure 12 This is an example of a screen that suggests a user participate in a marathon. The first column (112) displays the proposed improvement. The second column (113) displays a list of the proposed marathons.

[0139] In the first column 112, the user's tendency as a parameter of action is displayed with strings such as "Your tendency", "If the temperature is low (15°C), you can run a long distance continuously", and "If there are few slopes, you can exert speed with less energy".

[0140] Column 2, number 113, displays a list of the proposed marathon races, including their names, types, and months. In this list, each marathon is characterized by the labels "Speed," "Length," and "Enjoyment," indicating whether it's suitable for users aiming to run faster, longer, or more comfortably.

[0141] Figure 13 This is an example of a screen displaying a practice menu to the user. The first column (114) shows the content of the improvement proposal. The second column (115) displays the practice menu proposal.

[0142] In the first column 114, the user's tendency as a parameter of action is displayed with strings such as "Your tendency" and "If the temperature is low (15℃), you can run a long distance continuously".

[0143] In the second column 115, the name of the marathon is displayed as the target race, and strings such as "slope ○○m (high slope)" and "temperature 15℃ (high)" are displayed as characteristics of the marathon. In addition, a training menu and a suggested training route are displayed as "training plan".

[0144] Figure 14This is a flowchart illustrating the selection process of clothing and equipment. Input unit 62 inputs the user's motion data and stores it in storage unit 64 (S10). Estimation unit 70 inputs the motion data into a prediction model to estimate which of several runner types the user belongs to (S12). Based on a multiple regression analysis model, estimation unit 70 estimates the types of motion parameters that contribute to improving athletic performance for the given runner type (S14). Selection unit 72 selects shoes, etc., from various types of shoes that can contribute to improving athletic performance by changing the motion parameters estimated by estimation unit 70 (S16). Output unit 74 displays the selected shoes, etc., recommended to the user on user terminal 10 via network 18.

[0145] The present invention has been described above based on embodiments. Those skilled in the art will understand that the embodiments are illustrative, and various modifications can exist in the combination of their constituent elements or processing techniques; furthermore, such modifications are also within the scope of the present invention.

[0146] Industrial availability

[0147] This invention relates to a technique for analyzing motion.

Claims

1. A motion action analysis system, characterized by, include: The input unit inputs motion data corresponding to various motion parameters representing the motion state of a running user, and inputs indicator data corresponding to various indicator parameters representing multiple motion performances of the user during the running, wherein the various indicator data represent motion performances for different purposes. The model generation unit generates a multiple regression analysis model for each of the multiple runner types pre-classified by cluster analysis based on the motion data of multiple people. The multiple regression analysis model uses the multiple motion parameters in the motion data of multiple people classified into the runner type as explanatory variables and the indicator parameters in the indicator data of multiple people classified into the same runner type as target variables to calculate the contribution of each motion parameter in the runner type to improving athletic performance. The estimation unit, based on the user's action data, estimates which of the multiple runner types the user's action data is classified into, and based on the multiple regression analysis model corresponding to the estimated runner type, estimates the action parameter that contributes more to improving the athletic performance represented by each of the multiple indicator parameters. The selection unit selects the type of clothing and equipment corresponding to the estimated runner type from a variety of clothing and equipment, and individually selects clothing and equipment with characteristics that change the estimated motion parameters for each of the various index parameters. as well as The output unit outputs the types of clothing selected by the selection unit as recommended clothing for the user.

2. The motion analysis system of claim 1, wherein Also includes: The model storage unit stores a predictive model for each of the multiple runner types, generated by using the motion data of multiple people classified as the runner type as teacher data for machine learning. as well as The estimation unit estimates the runner type by inputting the user's action data into the prediction model.

3. The motion analysis system of claim 2, wherein Also includes: The analysis department performs cluster analysis on the motion data of the multiple people and classifies them into the multiple runner types; Specifically, the model generation unit uses the motion data of multiple runners classified into the runner type as teacher data to perform machine learning and generate the prediction model for each of the multiple runner types.

4. The motion analysis system according to any one of claims 1 to 3, characterized in that, Also includes: The project storage department stores types of apparel with the following characteristics as apparel types corresponding to the runner types: characteristics that contribute to improving athletic performance by altering motion parameters in a manner corresponding to the tendency of various motion parameters in motion data classified into the multiple runner types for each runner type. The selection unit selects clothing items from the various clothing items that correspond to the runner type to which the user's motion data belongs, based on the information stored in the project storage unit.

5. The motion analysis system according to any one of claims 1 to 3, characterized in that, The selection unit selects from a variety of clothing items which have the characteristic of contributing to improved athletic performance by changing the estimated motion parameters, based on the results of the multiple regression analysis model that presumes which motion parameters contribute more to improved athletic performance for the runner type.

6. A motion analysis system, characterized in that, include: The input unit inputs motion data corresponding to various motion parameters representing the motion state of a running user, and inputs indicator data corresponding to various indicator parameters representing multiple motion performances of the user during the running, wherein the various indicator data represent motion performances for different purposes. The analysis department performs cluster analysis on the motion data of multiple people and classifies them into multiple runner types; The model generation unit generates a predictive model by using the motion data of multiple runners classified into the same runner type as teacher data for machine learning for each of the multiple runner types. It also generates a multiple regression analysis model for each of the multiple runner types. The multiple regression analysis model uses the various motion parameters in the motion data of multiple runners classified into the same runner type as explanatory variables and the indicator parameters in the indicator data of multiple runners classified into the same runner type as target variables to calculate the contribution of each motion parameter in the runner type to improving athletic performance. The estimation unit estimates the runner type by inputting the user's action data into the prediction model, and based on the multiple regression analysis model corresponding to the estimated runner type, estimates the action parameter that contributes more to improving the athletic performance represented by each of the multiple indicator parameters. The selection unit selects the type of clothing and equipment corresponding to the estimated runner type from a variety of clothing and equipment, and individually selects clothing and equipment with characteristics that change the estimated motion parameters for each of the various index parameters. as well as The output unit outputs information that corresponds to the tendency of the values ​​of the motion parameters in the estimated runner type, and outputs the type of clothing selected by the selection unit as recommended clothing for the user.

7. A method for analyzing motion, characterized in that, include: The process of using a specified information acquisition unit to input motion data corresponding to multiple motion parameters representing the motion state of a running user, and inputting indicator data corresponding to multiple indicator parameters representing multiple motion performances of the user during the running, wherein the multiple indicator data represent motion performances for different purposes; For each of the multiple runner types pre-classified by cluster analysis based on the motion data of multiple people, a multiple regression analysis model is generated. The multiple regression analysis model uses the multiple motion parameters in the motion data of multiple people classified into the runner type as explanatory variables and the indicator parameters in the indicator data of multiple people classified into the same runner type as target variables to determine the contribution of each motion parameter in the runner type to improving athletic performance. Through computer-defined processing, based on the user's action data, it is inferred which of the multiple runner types the user's action data is classified into, and based on the multiple regression analysis model corresponding to the inferred runner type, for each of the multiple indicator parameters, the process of individually inferring the action parameter that contributes more to improving the athletic performance represented by each indicator parameter. The process of selecting the type of clothing and equipment corresponding to the presumed runner type from a variety of clothing and equipment, and individually selecting clothing and equipment with characteristics that change the presumed motion parameters for each of the various index parameters. as well as The process of using a specified information output unit to output the types of clothing and accessories selected as recommended clothing and accessories for the user.

Images