KR102379992B1 - An Automatic Coaching System And Method For Coaching A User's Exercise - Google Patents

Abstract

An automatic coaching system and an automatic coaching method related to an exercise state of a user are provided. The automatic coaching method for the user's exercise state is an automatic coaching method for the user's exercise state applied in an automatic coaching system, and uses data collected from an acceleration sensor or a position sensor worn on the user to at least one metric a metric calculation step of calculating a new coaching target metric selection step of calculating an error rate between the calculated metric and a previously stored metric reference value, selecting a new coaching target metric based on the error rate, and outputting a coaching message related to the coaching target metric; a new metric improvement checking step of confirming the occurrence of new metric improvement when the new coaching target metric is not selected and outputting an alarm when the new metric improvement occurs; and an exercise state output step of outputting an alarm for an exercise state when the metric improvement and the new coaching target metric do not exist.

Description

An Automatic Coaching System And Method For Coaching A User's Exercise

The present invention relates to an automatic coaching system and method for coaching a user's exercise, and more specifically, automatically generates a coaching value for a metric related to the user's exercise information, posture information, or injury risk information, and It relates to a system and method for outputting.

It has been steadily pointed out that the amount of exercise in modern people's daily life is quite insufficient to maintain proper physical health, and accordingly, interest in systematic exercise methods that can effectively promote health is increasing. As one of the exercise methods that meet these needs, there is a walking or running exercise that anyone can easily do.

However, in the case of overweight people and the elderly with weak joints, such walking or running exercise impacts the joints on which the weight is carried or the ground reaction force, such as the knees and ankles, and there is a risk of joint damage caused by the exercise. point has been pointed out. On the other hand, in general, people use a treadmill such as a health center for walking or driving exercise, but there are also many cases where they use an outdoor walking path, a park, or the like.

However, if the floor of such an outdoor walkway is hard such as asphalt, or if you exercise in shoes that do not sufficiently absorb shock, there is a risk of joint damage due to impact even for people with general health conditions.

In order to prevent this problem, various research efforts are being made, such as developing running shoes with shock absorption function or developing various designs to minimize the risk of injury on treadmills.

In the case of a group of researchers or engineers who specialize in research on shock absorption during driving, they are equipped with various experimental equipment for research and development. can do However, it is almost impossible to estimate by yourself how much the general public is actually being impacted while walking or driving, and the risk of injury due to it, other than the group that does this research professionally. In addition, it is also practically impossible for the general public to go to a specialized research facility as described above and measure the posture and risk of injury while driving.

Korean Patent Registration No. 1430135 (“Shoe Outsole”, 2014.08.07) Korean Patent Publication No. 2011-0107420 ("Fall prevention and gait training system", 2011.10.04)

The technical problem to be solved by the present invention has been devised in order to solve the problems of the prior art as described above, and an object of the present invention is to use portable equipment that can be easily worn and exercised by ordinary people to exercise posture, injury The goal is to provide a system and method to inform users of the level of risk and to appropriately perform coaching.

However, the technical problems to be solved by the present invention are not limited to the above problems, and may be variously expanded without departing from the technical spirit and scope of the present invention.

An automatic coaching method for a user's exercise state according to an embodiment of the present invention for solving the above problem is an automatic coaching method for a user's exercise state applied in an automatic coaching system, wherein an acceleration sensor worn on the user or a metric calculation step of calculating at least one metric using data collected from a location sensor; a new coaching target metric selection step of calculating an error rate between the calculated metric and a previously stored metric reference value, selecting a new coaching target metric based on the error rate, and outputting a coaching message related to the coaching target metric; a new metric improvement checking step of confirming occurrence of new metric improvement when the new coaching target metric is not selected and outputting an alarm when the new metric improvement occurs; and an exercise state output step of outputting an alarm for the exercise state when the metric improvement and the new coaching target metric do not exist.

In this case, in the metric calculation step, the instantaneous vertical load factor (IVLR) estimate metric is calculated based on the high-pass filter having a cutoff frequency region of 5 Hz or more after fast FFT-transformation (FFT) of the data collected from the acceleration sensor. and calculating an injury risk related metric comprising:

In addition, the step of calculating the injury risk-related metric including the instantaneous vertical load factor (IVLR) estimate metric includes converting the vertical acceleration data collected by the acceleration sensor for a certain period into a high-speed free, and then passing through the high-pass filter. The method may include calculating an estimate based on a power value calculated based on data.

On the other hand, the metric calculation step calculates an estimate of the pressure center path based on the location of the user's center of mass, and calculating an exercise posture related metric including a stride length metric and an interpolation metric calculated based on the pressure path estimate. may include

Also, in the step of selecting a new coaching target metric, after selecting a new coaching candidate metric based on the error rate, at least one metric having the highest priority among the selected new coaching candidate metrics may be selected as the new coaching target metric.

In addition, the priority is calculated as a product of the error rate and a weight corresponding to the new coaching candidate metric, and when selecting the new coaching target metric, the number of alarm frequencies of the new coaching candidate metric may be considered.

In addition, the step of selecting a new coaching target metric may include storing the selected metric as the new coaching target metric in a coaching target list.

In addition, the step of confirming the improvement of the new metric may include checking whether at least one of the metrics stored in the coaching target list is improved by an improvement reference error rate or more.

Other specific details of the invention are included in the detailed description and drawings.

According to the present invention, by using equipment that is portable and can be easily worn on the body such as the head and waist, the general public can receive coaching on their own exercise posture or measure the risk of injury very easily while driving. There is a big effect that In particular, in a situation where the majority of ordinary people need self-diagnosis while exercising for health, such as in modern times, it is possible to measure the risk of such injuries on their own without using a professional management institution, so it is a leap in convenience and improvement in public health. It has the effect of improving economic efficiency.

In addition, in terms of device configuration, according to the present invention, there is a great advantage that only a sensor that measures a user's dynamic physical quantity, such as an acceleration sensor, can be used. That is, in the past, there were several problems such as a decrease in device durability and lifespan by using a pressure sensor that recognizes walking by being pressed by a user's foot, and a problem in producing and using a separate device according to the user's body size. However, in the case of the present invention, since the technical configuration itself of disposing the pressure sensor on the foot part, which is the cause of this problem, is completely excluded, various problems as described above are fundamentally eliminated. Of course, it is natural that effects such as improvement of user convenience and improvement of economic efficiency for each user or producer can also be obtained from this.

In addition, from the point of view of user experience (UX, User eXperience), if the alarm is given too often for coaching, the user may feel uncomfortable and may significantly interfere with efficient exercise, so the automatic coaching system of the present invention is It can first correct the exercise posture that is most important and urgently needed to be improved.

However, the effects of the present invention are not limited to the above effects, and may be variously expanded without departing from the spirit and scope of the present invention.

1 is a block diagram illustrating an embodiment of an automatic coaching system to which the technical idea of the present invention can be applied.
2 is a diagram schematically illustrating a wearing position of a sensor signal collecting unit to which the technical idea of the present invention can be applied.
3 to 8 are diagrams for explaining an embodiment of a method for calculating an exercise posture related metric that can be applied in the present invention.
9 to 11 are diagrams for explaining an embodiment of a method for calculating an injury risk related metric that can be applied to the present invention.
13 is a flowchart illustrating an automatic coaching method to which the technical idea of the present invention can be applied.
14 is a flowchart illustrating a method for confirming occurrence of a new coaching target metric and alarming a metric with high priority that can be applied to the present invention.
15 is a flowchart illustrating a method for metric improvement occurrence confirmation and metric improvement alarming method applicable to the present invention.
16 is a flowchart illustrating a current exercise state alarm method applicable to the present invention.
17 is a flowchart illustrating in detail a method for confirming generation of a new coaching target metric that can be applied to the present invention.
18 is a flowchart for explaining in detail a method for confirming occurrence of metric improvement that can be applied to the present invention.
19 is a graph for explaining calculation of an instantaneous vertical load factor (IVLR) estimation coefficient to which the technical idea of the present invention can be applied.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be embodied in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the art to which the present invention pertains It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims.

The terminology used herein is for the purpose of describing the embodiments, and is not intended to limit the present invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, "comprises" and/or "comprising" refers to the presence of one or more other components, steps, operations and/or elements mentioned. or addition is not excluded.

Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used with the meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless specifically defined explicitly.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. The same reference numerals are used for the same components in the drawings, and repeated descriptions of the same components are omitted.

automatic coaching system

1 is a block diagram illustrating an embodiment of an automatic coaching system to which the technical idea of the present invention can be applied.

The automatic coaching system 100 to which the technical spirit of the present invention can be applied may include a sensor signal collection unit 110 , a control unit 120 , and a coaching guide output unit 130 .

The sensor signal collection unit 110 measures a user's dynamic physical quantities such as acceleration, speed, and position, and may include a sensor unit 111 and a communication unit 112 . In this case, the hardware of the sensor unit 111 basically includes an acceleration sensor in a three-axis direction including up and down, left and right, and front and rear. In this case, the sensor unit 111 may include a sensor capable of measuring acceleration in the 6-axis direction by additionally embedding a gyroscope in the acceleration sensor. Alternatively, the sensor signal collecting unit 110 may be a sensor for measuring a 9-axis posture including, for example, an acceleration sensor, a gyro sensor, and a geomagnetic sensor. Alternatively, the sensor signal collection unit 110 may measure the user's position by having a position sensor such as a GPS signal, for example, and a suitable sensor to which ultra-precise satellite navigation technology or indoor positioning technology, which is more accurate than recent GPS, is applied. may include In addition, those skilled in the art may be equipped with an appropriate sensor in consideration of appropriate sensing and power use as the sensor signal collecting unit 110 .

The communication unit 112 may transmit the original signal collected by the sensor unit 111 to the control unit 120, and may be serial communication such as SPI, I2C, or UART depending on the physical location of the control unit 120 of the sensor unit, It may be wireless communication such as WiFi or Bluetooth.

Meanwhile, the sensor signal collecting unit 110 may be worn on various parts of the user's body, such as the head, waist, and chest, as shown in FIGS. 2(a) to 2(e) in the wearing position. When worn on the head, the sensor signal collection unit 110 may be included in a hair band, for example, as shown in FIG. 2(a), or may be included in a hat or earphone as shown in FIG. 2(b), or FIG. ) may be included in wearable glasses. Alternatively, when the sensor signal collection unit 110 is worn on the chest, the sensor signal collection unit 110 may be included in the chest band as shown in FIG. 2( d ).

In this case, only one sensor signal collecting unit 110 may be worn as shown in FIG. 2( a ), or a plurality of sensor signal collecting units 110 may be worn as shown in FIG. 2( e ). For example, the sensor signal collecting unit 110 may be worn on the head and/or waist. 1 is a block diagram for a case in which the sensor unit 111 is installed on the head side 111H and the waist side 111W at the same time. However, the sensor unit 111 may be installed either on the head side 111H or the waist side 111W.

Existing devices for measuring the amount of exercise used pressure sensors provided in shoes and footrests, which are parts that are directly pressed by the foot, for monitoring the user's gait. Accordingly, there is a problem in that the damage of the sensor occurs quickly, and the durability and lifespan of the device are shortened. In addition, there were problems such as a decrease in gait recognition and analysis accuracy due to damage to the device during use, and a decrease in convenience and economy due to frequent device replacement. In addition, when such a device is provided in the shoe, a separate device is required for each user according to the size of the user's foot, which increases the user's convenience and economic feasibility, and causes an economic burden for the producer to separately produce each size. There were problems such as

However, the present invention completely breaks away from the concept of using the pressure of the foot in gait recognition, measures the user's dynamic physical quantities such as acceleration, speed, and position measured in the user's head, etc., and a characteristic analysis of the present invention to be described below The recognition, detection and analysis of gait can be realized by applying the algorithm.

As such, the present invention is different from the prior art in measuring positions and measuring physical quantities. At this time, the root cause of the various problems pointed out in the prior art comes from the technical configuration of 'place the pressure sensor on the foot'. can be removed with

Meanwhile, the controller 120 may include a metric calculator 121 and a coaching guide generator 122 . In this case, the metric calculator 121 and the coaching guide generator 122 may be physically dispersed. For example, the metric calculator 121 may be formed on a single substrate in the form of an integrated circuit with the sensor signal collecting unit 110 to perform various calculations, or may be formed in a form such as a separate computer. . In this case, the coaching guide generating unit 122 may be located in the coaching guide outputting unit 130 in the form of a software module. Alternatively, the sensor signal collecting unit 110 , the metric calculating unit, and the coaching guide generating unit 122 may be physically located as one hardware and software module. A person skilled in the art may appropriately distribute and locate the metric calculating unit 121 and the coaching guide generating unit 122 in consideration of hardware limitations such as battery and processor performance.

The coaching guide output unit 130 recognizes the guide information including the warning signal and/or the stride correction amount generated by the coaching guide generating unit 122 in this way, a user including at least one of a sound, an illustration, or an image. It can be converted and output as possible information.

For example, if the coaching guide output unit 130 is an earphone and the stride correction amount is calculated and it is necessary to reduce the stride length, voice coaching such as “reduce the stride length” or a warning sound is made to indicate that the user is not the optimal stride length. It can be recognized and induce a change in walking posture. Alternatively, the coaching guide output unit 130 may output coaching guide information as an illustration or an image in the form of a smartphone, computer, or dedicated display, wearable display (eg, Google Glass), or the like.

In addition, the coaching guide generating unit 122 or the coaching guide outputting unit 130 may be configured to transmit and accumulate the walking posture derived by the metric calculating unit 121 to the external database 140 . Users who require such walking or driving motion analysis may be ordinary people who walk or jog daily for health promotion, or experts who train to improve physical ability, and such motion analysis data is accumulated and time It is of course desirable to be configured so that changes can be seen. In addition, when exercise analysis data is accumulated and stored in a large amount in this way, such data can be utilized as big data to be used for various statistics or analysis, and various uses are possible.

Meanwhile, the coaching guide output unit 130 may provide coaching to the user by generating and playing a guide message with respect to the coaching target metric having the highest priority. The coaching guide output unit 130 may output the guide message to the user using an auditory method or a visual method. For example, the coaching guide output unit 130 may generate a coaching target metric with the highest priority in the case of voice coaching as a natural language-based message and notify the user. This may be applied to various sound reproduction devices such as earphones, headphones, and speakers, and may provide values and evaluations for corresponding coaching metrics, and may provide correction methods.

In addition, in the case of coaching using a visual display, it can be applied to devices using various visual displays such as mobile phone screens, smart watches, monitor screens, AR displays, LCD displays, etc. It can be provided as text or video.

As a specific application example, the coaching guide output unit 130 performs coaching such as "Be careful of M1, M2, and M3" in the case of voice coaching, and "The impact force is too large at 150N/s. Run more lightly." can be provided to users. Alternatively, the coaching guide output unit 130 may display coaching such as "The up-and-down movement is too large. Increase the reward, and take a small leap when jumping", "Your stride is too large. Please raise your reward" in the case of visual display coaching, using text or emoticons Alternatively, a short video may be outputted through a visual display device and provided to the user.

The automatic coaching method according to the present invention described above may be set so that the entire cycle is periodically repeated every n seconds of the operation period. This automatic coaching method can be applied to the automatic coaching system described above.

Configure the metric calculator

The metric calculation unit 121 of the control unit 120 receives a signal from the sensor signal collection unit 110, and uses the 3-axis acceleration and/or position signal to calculate walking or driving motion information, posture information, and risk of injury. Derive various related metrics. The generated metrics are, for example, as shown in Table 1 below.

In this case, the various metrics are basically generated by using the sensing values such as acceleration in each direction collected by the sensor signal collecting unit 110 or other metrics, and for convenience, in the present specification, the values calculated using the other metrics are used as sub-metrics. Metrics using basic sensing values are indicated as upper metrics. Some of the metrics may be derived using, for example, a GPS location sensor of the coaching guide output, or the like.

Metric values provided to users Korean English exercise information hour Time distance Distance [km] speed pace [m/s] virtual speed virtual pace [m/s] calorie consumption Calories [kcal] Altitude Altitude [m] number of steps Step number [steps] detailed information pay per minute Cadence [steps / min] stride Stride length [m] >Advanced distance traveled by sticking to the ground >Forward distance traveled during stance phase [m] >The distance traveled in the air >Forward distance traveled during flight phase [m] interpolation Step width [cm] complement Step angle [deg] head angle Head angle [deg] vertical travel distance Vertical oscillation [m] >Vertical movement distance when supporting the left foot >Vertical oscillation L [m] >Vertical movement distance when supporting the right foot >Vertical oscillation R [m] >Vertical distance moved by attaching to the ground >Vertical oscillation during stance phase [m] >Vertical distance traveled in the air >Vertical oscillation during flight phase [m] support time in contact with the ground Ground contact time [s] >Left foot support time >Ground contact time L [s] >Right foot support time >Ground contact time R [s] airtime flight time [s] >Left foot >Flight time L [s] >right foot >Flight time R [s] Support time-float time ratio Contact-Flight Ratio
Tf/Tc = [%] ground reaction force ground reaction force; GRF center of pressure Center of pressure; COP
injury
Risks
information Maximum vertical load Max Load on Legs [N] >Left foot full load >Max Load on Legs L [N] >Right foot full load >Max Load on Legs R [N] shock load factor Impact Loading Rate [N/s] left-right balance Symmetry [%] uniformity Stability [%] >Left Stability - L [%] >Oh Stability - R [%] leg stiffness Leg stiffness [BW/m] BW = Body weight, m*g >Leg stiffness >Leg stiffness [N/m] > Instantaneous vertical load factor estimate Estimates for IVLR; (Instantaneous Vertical Loading Rate) [BW/S, N/S) BW = Body weight, m*g

> is the child metric derived from the parent metric

Implementation example of metric calculation related to exercise posture

The metric calculator 121 of the present invention detects the user's motion and calculates various metrics related to the distinction between walking/running or posture determination. Hereinafter, a metric calculation example of the present invention will be described in detail with reference to FIGS. 3 to 8 .

For example, stride length, interpolation, complement angle, and left and right asymmetry, which are metrics for deriving an exercise posture, may include a step of estimating a center of pressure path, a step of determining an exercise type, and a step of calculating an exercise posture-related metric as shown in FIG. 3 .

For example, in the step of estimating the center of pressure path, the acceleration vector at the position of the center of mass using the motion state values of the user's center of mass calculated using the three-axis accelerations (ax, ay, az) sensed by the sensor unit 111 are used. Estimate the center of pressure path by projecting it onto the ground in a direction.

The metric calculation unit 121 integrates the 3-axis acceleration collected by the 3-axis acceleration sensor 111 of the sensor unit 111 or uses the position information of the position measurement sensor of the sensor unit 111 to determine the speed, position save In this case, it is preferable to convert and analyze the data value at the user's center of mass based on the basic data (key information) previously input from the user and the sensor data collected by the sensor unit 111 . In this case, the exercise data value at the user's center of mass may be calculated by appropriately multiplying a gain value obtained in advance by using body information such as user height information.

Meanwhile, the center of pressure path may be estimated from the value of the motion state at the user's center of mass (acceleration/velocity/position per time in each direction, frequency analysis, etc.).

The human body moves by using the reaction pressure applied to the supporting foot when walking or running. The sum of these reaction pressures is called the ground reaction force (GRF), and the center of this pressure is called the center of pressure (COP). It has been found that the ground reaction force generated at this time has a characteristic toward the center of mass (COM) from the center of pressure. 4 schematically illustrates a relationship diagram between the center of mass and the center of pressure. In this embodiment, the center of pressure is inferred by projecting to the ground in the vector direction of the force measured from the center of mass.

5 is a view for explaining the determination of the direction of the pressure center and the analogy of the position. The center of pressure direction refers to a direction from the center of mass to the center of pressure. In the step of estimating the center of pressure path, the direction of the center of pressure is first determined, and the position of the center of pressure is inferred by projecting in this direction. In more detail, in the first step of determining the direction of the pressure center, as shown in FIG. 5 , the ratio of the left-right acceleration (ax) to the sum of the vertical acceleration (az) and the gravitational acceleration (g) and the vertical acceleration (az) ) and the ratio of the forward-backward acceleration (ay) to the sum of the gravitational acceleration (g) determines the direction of the pressure center. When the direction of the pressure center is determined in this way, in the next step of inferring the position of the pressure center, it is assumed that the center of mass is located at a height determined by multiplying the pre-measured user key information by a predetermined constant for inference, and the pressure center The position of the center of pressure is inferred by projecting on the ground in the direction determined in the direction determination step. Here, the inference constant refers to the height of the center of mass according to the height of the user. In general, it is well known that the center of mass of children is proportionally higher than that of adults, and that the center of mass of men is proportionally higher than that of women, and of course the ratio is also known. As a specific example, it is known that the center of mass of an adult male is located at 55.27% of the height on average, and in this case, the inference constant is 0.5527. Therefore, for example, when inputting user key information, by having children/adults and male/female division information be input together, an appropriate analog constant can be selected and used for calculation.

In order to further increase the accuracy of the pressure center position obtained in this way, a pressure center position correction step in which the pressure center position inferred in the pressure center position inference step is multiplied by a predetermined constant for front and rear and left and right directions may be further performed. . Here, the constant for front-rear and left-right direction correction is a constant that can statistically match the position of the pressure center obtained by the projection method as described above with the actual center of pressure in the front-rear and left-right directions.

In the exercise type determination step, it is determined whether walking or running from the pattern of the vertical acceleration (az) graph. 6 is an example of an origin pattern obtained as an estimated center of pressure path. As shown, it can be seen that the left and right feet alternately support the ground while progressing.

On the other hand, what distinguishes walking from running is that, in the case of walking, one or both feet are always in contact with the ground, whereas in driving, one or both feet are always floating from the ground. 7 illustrates an example of a vertical acceleration graph with respect to time during walking and driving. In the case of the walking graph shown in Fig. 7(a), it can be seen that the peak occurs the moment both feet touch the ground, and in the case of the driving graph shown in Fig. 7(B), both feet are floating from the ground. It can be confirmed that there is a constant value section in which the instantaneous vertical acceleration az becomes the minimum value. As such, it is possible to determine whether the user's current exercise is walking or running by using the different patterns of the vertical acceleration (az) graph in each case during walking and driving.

In the exercise posture deriving step, posture information including stride length, interpolation, complement angle, and left and right asymmetry is derived based on the pressure center path estimate and the three-axis acceleration (ax, ay, az). It will be described in more detail with reference to the example of the center of pressure path of FIG. 6 and the example of vertical acceleration during walking or driving of FIG. 7 .

First, it has been described above that the user's movement is slightly different from walking and running, and of course, there are some things that appear in common. As described above, in the case of walking, one or both feet are always in contact with the ground, and in the case of running, one or both feet are always floating from the ground. That is, there is a section in which only one foot is supported in common during walking and driving. In consideration of these points, in the exercise posture derivation step, first, the intermediate support point determination step of determining the intermediate support point, and the section classification determination step of determining the two-foot support section, the one-foot support section, and the floating section are made, It forms basic information for deriving posture while distinguishing walking and driving.

First, let's describe the gait motion by unraveling it as follows. First, at the moment when the heel of one foot hits the ground, the toe of the other foot does not even come off the ground, that is, it starts with both feet supported. In this state, while supporting the ground with only one foot, the other foot falls from the ground, and as the other foot moves forward while stirring the air, the body of the person also moves forward. And at the moment when the heel of the other foot hits the ground, the state where the toe of one foot has not yet come off the ground, that is, the state in which both feet are supported, is reestablished, and one-step gait is made. In this process, while the person's body is moving forward while supported by only one foot, the person's head does not shake greatly in the up-down direction (a local minimum is formed from the up-down acceleration (az)), while the moment of stepping It shakes the most in the up-down direction (a peak value is formed in the up-down direction acceleration (az)).

That is, the gait motion can be divided into a section in which both feet are on the ground and a section in which only one foot is on the ground, and the vertical shaking is the least among the states in which only one foot is on the ground. The aspect of this movement is well shown in Fig. 7(A) and as shown in this example, in the intermediate support point determination step, when the user's movement is walking, the local minimum in the vertical acceleration (az) measured in the time domain is defined as an intermediate support point. In addition, in the section classification determining step, when the user's movement is walking, a section in which a peak value is formed in the vertical acceleration (az) measured in the time domain is determined as a two-foot support section, and the remaining section is determined as a single-foot support section.

Next, let's describe the driving motion by unraveling it as follows. First, it starts with the moment when one foot protruding forward kicks off the ground (at this moment the other foot is floating in the air). In this state, one foot spurs up from the ground and the body moves forward with both feet floating in the air. When the other foot that comes out in front touches the ground, the moment when it spurs the ground again occurs, and the driving takes place one step at a time. In this process, the human head shakes the most in the vertical direction (local maximum is formed from the vertical acceleration (az)) at the moment when one foot is kicked off the ground, whereas while floating in the air, it hardly shakes in the vertical direction. (a constant value is formed in the vertical acceleration (az)).

That is, the driving motion can be divided into a section in which both feet are floating in the air and a section in which only one foot is on the ground, and the vertical shaking is the least among both feet floating in the air. The aspect of this movement is well shown in Fig. 7(B) and as shown in this example, in the intermediate support point determination step, when the user's movement is driving, the local maximum in the vertical acceleration (az) measured in the time domain is defined as an intermediate support point. In addition, in the section classification determining step, when the user's motion is driving, a section represented by a constant value in the vertical acceleration (az) measured in the time domain is determined as the floating section, and the remaining section is determined as a one-foot support section. Here, the constant value shown in the floating section is a preset value of the signal level level when no external force other than gravity acts on the accelerometer, and may be appropriately determined as a value close to approximately zero. That is, the constant value is a reference value that allows the current stance to be determined. In this sense, it can be called a stance phase constant. If the section is large, it is determined as a one-foot support section. As such, when the basic high-order metrics for posture derivation are derived, it becomes possible to derive walking or driving postures such as stride length, interpolation, supplemental angle, and left and right asymmetry.

Stride length: First, the average speed is calculated by measuring user location information at predetermined time intervals. Next, a walking frequency is calculated by measuring the number of intermediate support points during the time interval. Finally, by dividing the average speed by the walking frequency, the user's stride length can be accurately calculated.

Interpolation: Interpolation in the left and right directions can be calculated using the pressure center position value corresponding to the intermediate support point. That is, if the time value corresponding to the intermediate support point shown in FIG. 7(A) or (B) is applied to the pressure center position value shown in FIG. 6 and the pressure center position corresponding to this time value is found, the left foot stepped on the ground The position and the position where the right foot stepped on the ground are displayed, and the user's interpolation can be accurately calculated by measuring the distance between them.

Complementary angle: Complementary angle may be calculated using the pressure center position value corresponding to the start time of the one foot support period and the pressure center position value corresponding to the end time of the one foot support period. To explain it, the heel steps on the ground at the start of the one-foot support section, and the toe steps on the ground at the end of the one-foot support section. That is, as described above, to obtain the angle between the pressure center positions means to obtain the angle formed between the heel position and the toe position at the moment of step on the ground, that is, the complement angle, that is, in this way, the user's complement angle can be accurately calculated. .

Left-right asymmetry: First, the supporting foot is identified based on the sign of the left-right acceleration (ax) measured in the time domain. Next, a peak value, a valley value, and a difference value of the vertical acceleration (az) measured in the time domain are compared. That is, by comparing the peak value and the trough value when the left foot is supported and when the right foot is supported, it is possible to accurately calculate the left and right asymmetry of the user. In addition, the repeatability of walking or running can be calculated in the same way. FIG. 8 is an example of an acceleration signal measurement result, and it can be seen that the left-right asymmetry of the vertical acceleration az is strongly shown in the bottom graph of FIG. 8 .

Metrics such as stride length, interpolation, complement angle, left-right asymmetry, etc. may be calculated by the method as described above.

An example of calculating an injury risk-related metric

The detailed configuration of the metric calculation unit 121 to which the technical idea of the present invention can be applied is briefly described with reference to FIG. 1 in relation to the calculation of the metric related to the risk of injury, and then the instantaneous vertical load factor (IVLR) which is a metric related to the risk of injury ; instantaneous vertical loading rate) metric calculation is described in detail.

As shown in FIG. 1, the metric calculator 121 of the present invention basically includes a high-pass filter (HPF, 124) for extracting a high-frequency signal above a preset frequency from the collected vertical acceleration signal, and a high-pass filter ( 124), the peak value between the first time point and the second time point is detected from the filtered high frequency signal, the average value of the detected peak values is calculated, and the average value is multiplied by the estimation coefficient k1 to obtain the instantaneous vertical load factor ( IVLR) may include a calculation unit 125 for calculating the estimated value.

At this time, the estimated coefficient value (k1) is the average value of the calculated peak values, the measured value (or the value calculated through the measured value) collected from the force plate and the infrared motion capture system, and regression analysis It is determined by the regression coefficient obtained through

In general, the instantaneous vertical load factor (IVLR) is very difficult to obtain from data collected from a sensor worn on the user's body, such as the present system. Because, in terms of power, most wearable devices use a battery as a power source, so the available power is quite limited, whereas the instantaneous vertical load factor (IVLR) shows a slope value, so a high sampling rate (typically 1000Hz or more) ) can be precisely measured.

Sensors having a high sampling rate are very difficult to be worn on a wearable device due to a power problem. In general, sensors that can be used in a mobile device or a wearable device have a sampling rate of less than 200 Hz. With such a low sampling rate, the accuracy of IVLR measurement is very poor.

However, the applicants of the present invention are the average value of the peak value of the high frequency component of the vertical acceleration signal among metrics that can be collected by using a sensor in a mobile device or a wearable device, or the sum of the high frequency power signal of the Fourier transform of the vertical acceleration signal It was found through an experiment that it had a high correlation with the IVLR value. Therefore, as shown in FIGS. 19(a) and 19(b), regression analysis of the sum of IVLR and the average value of the peak values of the high-frequency components of the vertical acceleration signal, and the sum of the high-frequency power signal of IVLR and the Fourier transform of the vertical acceleration signal It was applied to the present invention that the IVLR value can be estimated quite precisely using the estimation coefficients k1 and k2 obtained through .

Here, the unit of IVLR may be N/s, which is an absolute unit indicating force per unit time, or may be expressed as g/s. Alternatively, since the sensor applied to the present invention basically collects acceleration data rather than force, data is collected in BW/s, which is a relative unit (here, BW is body weight). If this is multiplied by the user's mass m, it can also be expressed as an absolute unit, N/s((N = m*g)/s). In this case, the user's mass m may be included in the estimation coefficient (k1 or k2), or may be calculated as the estimation coefficient (k1 or k2) × the user's mass (m) as a separate coefficient.

On the other hand, a high-frequency signal may mean, for example, a vertical acceleration signal of 10 Hz or more, which is difficult to generate an acceleration vertical signal of 5 Hz or more by the user's voluntary movement, and the vertical acceleration signal generated in a situation such as an impact is generally high. Since it exists in the band, the present invention intends to use it to estimate the instantaneous vertical load factor (IVLR) using a high-frequency signal of 10 Hz or higher.

The calculator 125 detects a peak value between a first time point and a second time point from the filtered high frequency signal, and the first time point may be, for example, a user's ground landing time, and the second time point is, for example, This may be the midpoint of the ground reaction force peak. This time point will be described later with reference to FIG. 9 .

Additionally, the metric calculator 120 includes a frequency domain converter 127 that converts the vertical acceleration signal collected by the sensor 111 from the time domain to the frequency domain, and a power from the vertical acceleration signal in the frequency domain. It may further include a power signal detection unit 126 for detecting a signal. In this case, the high-pass filter 124 extracts a high-frequency power signal having a frequency higher than or equal to a preset frequency from the detected power signal, and the calculating unit 125 calculates the sum of the high-frequency power signals, and an estimation coefficient ( k2) to calculate an estimate of the instantaneous vertical load factor (IVLR).

The frequency domain transform unit 124 may operate to transform the vertical acceleration signal from the time domain to the frequency domain using Fast Fourier Transform (FFT).

On the other hand, in this case, the high-frequency power signal may mean, for example, force signals of 10 Hz or higher, which is difficult to generate a force signal of 5 Hz or higher by the user's voluntary movement, and the force signal generated in situations such as an impact is generally Since it exists in a high band, in the present invention, the instantaneous vertical load factor (IVLR) can be estimated and calculated as a metric using a high frequency power signal of 5 Hz or more, preferably 10 Hz or more.

Hereinafter, a method of calculating the instantaneous vertical load factor will be described in detail. Prior to this, first, a vertical acceleration (a _z ) graph will be described. 9( a ) is a diagram illustrating a vertical acceleration graph during driving. As shown, the vertical acceleration (a _z ) appears in a periodic form with respect to time (it is natural because walking or running itself is a periodic motion). Let's unpack and describe the driving motion as follows.

First, it starts with the moment when one foot protruding forward kicks off the ground (at this moment the other foot is floating in the air). In this state, one foot spurs up from the ground and the body moves forward with both feet floating in the air. When the other foot that came out in front landed on the ground, the moment it kicked off the ground again occurred, and a single step of driving was made. In this process, at the moment of landing on one foot, the person's head shakes the most in the vertical direction (a local maximum is formed from the vertical acceleration (a _z )), while in the state of moving forward while floating in the air, it hardly shakes in the vertical direction (a constant value is formed in the vertical acceleration (a _z )).

The moment the foot lands, the most impact is applied to the joint, and this impact appears in the form of a first peak in the vertical acceleration graph as shown in FIG. 9(a). The risk of injury varies depending on the degree of impact at this time, and it can be calculated as a metric and used as a basis for quantified judgment of the risk of injury. As such a determination metric, the average slope of the vertical acceleration (a _z ), the maximum slope of the vertical acceleration (a _z ), the maximum impact force, the amount of impact, and the like may be used.

9( b ) is a view showing a slope in a vertical acceleration graph during driving. Through this, the average slope and the maximum slope of the vertical acceleration a _z can be derived. In the vertical acceleration graph, the slope corresponds to the instantaneous vertical load factor (IVLR), and the instantaneous vertical load factor (IVLR) can be estimated using the algorithm in the present invention.

10 (a) to 10 (c) are graphs for explaining the calculation of the estimation value of the instantaneous vertical load factor of the present invention.

First, the vertical acceleration signal collected by the sensor unit 111 appears in the form of a period as shown in FIG. 10( a ).

Using the high-pass filter 124 from the collected vertical acceleration signal, a high-frequency signal of a preset frequency or higher is extracted. For example, as described above, a high-frequency signal of 5 Hz or more or preferably 10 Hz or more may be extracted. FIG. 10(b) shows that a high-frequency signal of 10 Hz or higher is extracted from the vertical acceleration signal.

The peak value (*) between the user's ground landing time (▷) and the midpoint (◁) of the ground reaction force peak is detected from the extracted high-frequency signal. 10( c ), the user's ground landing time (▷), the midpoint of the ground reaction force peak (◁), and the peak value (*) therebetween are shown.

After calculating the average value of the detected peak values, the estimated value of the instantaneous vertical load factor (IVLR) can be calculated as a metric by multiplying this average value by the estimated coefficient k1. This can be used as a criterion for judging the risk of injury. For example, when the estimated value of the instantaneous vertical load factor IVLR is greater than the preset value n, it may be determined that there is a risk of injury.

In addition, this can be used as a criterion for evaluating the correct posture (energy efficiency). For example, when the amount of impact is larger than the preset value (m), it is judged to be inefficient because the loss of mechanical energy is large because the amount of impact is large. can do. The impact amount will be described later in more detail.

When the metric calculator 121 further includes the frequency domain converter 127 and the power signal detector 126 , the estimated value of the instantaneous vertical load factor IVLR may be calculated as a metric using the power signal.

For example, the acceleration signal in the vertical direction appears in the form of a period as shown in FIG. 10( a ). At this time, the metric calculator 121 converts the collected vertical acceleration signal from the time domain to the frequency domain, and detects power (Power, the square of the FFT size), which is a force signal, from the vertical acceleration signal in the frequency domain. . 11( a ) shows an example in which a vertical acceleration signal is transformed into a frequency domain using a fast Fourier transform (FFT), and power is detected therefrom.

High-frequency power greater than or equal to a preset frequency is extracted from the detected power, and the sum of the extracted high-frequency powers is calculated. For example, the high-frequency power is 5 Hz or more, preferably 10 Hz or more, and FIG. 11( b ) is illustrated for the high-frequency power of 10 Hz or more.

The instantaneous vertical load factor IVLR may be estimated by multiplying the sum of the high frequency powers by the estimation coefficient k2. As described above, it can be used as a criterion for determining the risk of injury using this. For example, when the instantaneous vertical load factor IVLR is greater than a preset value n, it may be determined that there is a risk of injury.

In addition, this can be used as a criterion for evaluating the correct posture (energy efficiency). For example, when the amount of impact is larger than the preset value (m), it is judged to be inefficient because the loss of mechanical energy is large because the amount of impact is large. can do.

12 is a view showing the amount of impact on a vertical acceleration graph during driving. The maximum impact force value can be calculated using the following equation.

Maximum impact force = m × a _z (t _m )

(where, a _z : acceleration in the vertical direction, m: user mass, tm: end time of impact)

As described above, the impact end time is the time at which the first peak value appears, so naturally, the time at which the maximum impact force appears is the impact end time. In FIG. 16 , the first peak (1st peak) of the up-down acceleration (a _z ) is displayed, and the value obtained by multiplying this by the user mass (m) becomes the maximum impact force value.

On the other hand, the value of the impulse can be calculated using the following equation.

(where, a _z : vertical acceleration, m: user mass, tc: shock start time, tm: shock end time)

In FIG. 12 , the vertical acceleration (a _z ) graph area between the impact start time and the impact end time is displayed, and the value obtained by multiplying this area by the user's mass (m) becomes the impulse value.

Automatic coaching method using calculated metrics

The coaching guide generating unit 122 uses the metric value derived by the metric calculating unit 121 to obtain coaching guide information including information on posture correction when the current walking or exercise posture needs to be corrected. plays a role in creating Here, the generating of the coaching guide information may include: determining a coaching target metric indicating a high risk of injury due to an inappropriate walking/running posture or lowering of exercise efficiency; selection of an alarm phrase according to the coaching target metric or the level of badness of the coaching target metric; Or it is a concept including at least one of generation of an alarm phrase including a correction amount of the coaching target metric.

The automatic coaching operation of the coaching guide generating unit 122 is as shown in FIG. 13 .

First, the metric calculation unit 121 of the control unit 120 calculates metrics related to the exercise posture and the risk of injury based on the sensor data collected by the sensor unit 111 . (S1310)

After calculating the metric, it is checked whether there is a coaching guide message currently being reproduced in the coaching guide output unit 130 or whether a coaching guide message is being output within a predetermined time. (S1320) At this time, if there is a coaching guide message being reproduced or a coaching guide message is output within a predetermined time, the metric calculation step (S1310) is repeated again.

Next, the coaching guide generating unit 122 checks whether the coaching guide operation is within a preset state guidance period. (S1330) In this case, the coaching guide generating unit 122 may set a state guidance period to guide the state at every predetermined distance or for each predetermined time based on the distance or time. For example, the coaching guide generating unit 122 may brief summary information of states such as an exercise distance, a coaching target reminder, and an exercise time every 500 or 1000 m during exercise through the coaching guide output unit 130 .

Next, the coaching guide generating unit 122 selects a new coaching target metric. (S1340) A selection algorithm for a new coaching target metric will be described later.

Next, the coaching guide generating unit 122 selects a new metric improvement. (S1350) That is, it is checked whether there is a part in which the exercise posture of the user is improved. A specific method of selecting a new metric improvement will be described later in detail in the description with respect to FIG. 15 .

Finally, the coaching guide generation unit 122 checks whether a new coaching target metric has occurred and a new coaching target new release metric is checked, and when there is no message alarm for a reference time or longer, reports on the current state. For example, the report on the current status may be a reminder for the coaching target list because there is no coaching target metric, so exercising with the correct posture or because there is no improvement in the coaching target list.

New coaching target metrics selection and alarm

Hereinafter, selection of a new coaching target metric will be described in detail with reference to FIGS. 14 and 17 . 14 is a flowchart illustrating a step of selecting a new coaching target metric, and FIG. 17 is a flowchart illustrating a step of checking whether a new coaching candidate metric is generated.

First, the coaching guide generating unit 122 checks whether a new coaching candidate metric is generated in order to select a new coaching target metric after the status guide cycle coaching guide check operation ( S1330 ). (S1410)

More specifically, referring to FIG. 17 , the coaching guide generating unit 122 first checks whether a metric for a review target is in the coaching target list. (S1411) Here, the coaching target list is, for example, a list in which a metric for which a coaching guide message has been output for a certain period is stored, and is a reference list when judging improvement or deterioration of metrics.

If it is not on the coaching target list, the difference between the corresponding metric and the metric reference value is calculated, and the error rate is calculated. (S1412) The error rate P is obtained as shown in Equation 1 below when the metric is X _i and the metric reference value is M _ref .

(Equation 1)

P = (M _ref - X _i )/M _ref

In this case, the metric reference value is stored in advance in the coaching guide generating unit 122 . For example, the coaching guide generating unit 122 may store the optimal key-step length relation data for each walking and running speed as a reference value of the stride length metric. Alternatively, a reference value of an instantaneous vertical load factor metric according to height and/or weight may be stored.

On the other hand, the error rate P is compared with the reference error rate P _ref , and it is checked whether the corresponding metric is worse than the reference error rate. (S1413)

In this case, whether the corresponding metric has deteriorated may be determined by a different criterion for each metric. For example, when the impulse amount metric or the instantaneous vertical load factor (IVLR) metric is increased, it puts a strain on the user's body, so an increase in the corresponding metric value means deterioration. As another example, the stride length or the interpolation metric has a reference range having upper and lower limits, and may be judged to be deteriorated when it deviates from the reference range. Finally, in the case of a metric with the same left-right balance, it can be determined that the metric has deteriorated when the corresponding metric value is lower than the metric reference value.

When the corresponding metric deteriorates by more than the reference error rate P _ref , it is determined that a new coaching candidate metric has occurred (S1414), and in the opposite case, it is determined that the new coaching candidate metric has not occurred.

Referring again to FIG. 14 , when a new coaching candidate metric is generated, the priority of the new coaching candidate metric is determined. (S1420) However, when a new coaching candidate metric does not occur, the process proceeds to a new metric improvement selection step (S1350).

Meanwhile, the priority is determined according to the size and frequency of the error rate. For example, when a new coaching candidate metric is generated, it is first determined whether the corresponding metric has been coached for a reference number of times (n) or less during a reference time. That is, when the frequency of coaching guides of the new coaching candidate metric is less than or equal to the reference number, the priority Ki is determined by multiplying the error rate p of the corresponding metric by the weight value Wi for each metric.

(Equation 2)

Ki=p*Wi

In this case, if the number of coaching guide frequencies of the new coaching candidate metric is equal to or greater than the reference number (n times), the corresponding metric is excluded from the new coaching candidate metric. This is because, when the alarm for the corresponding metric is excessively generated, the user feels uncomfortable, and the awareness of the corresponding metric is reduced, which interferes with efficient exercise. Accordingly, the user experience discomfort is reduced by appropriately adjusting the reference number of times.

The weight for each metric is a set factor because the importance of each metric is different. For example, the weight of the metric representing the risk of injury may be set high, and the weight of the metric representing the exercise posture may be set relatively low compared to the weight of the metric related to the risk of injury.

Therefore, according to the present invention, since an alarm can be notified only when an important exercise metric deteriorates while reducing the user's discomfort due to frequent alarms, the efficiency of exercise is remarkably increased.

After the priority Ki is determined, a metric having the highest priority among new coaching candidate metrics is selected as a new coaching target metric, and a coaching message is outputted through the coaching guide output unit 130 . (S1430) A user interface such as a display or voice alarm that can be utilized during exercise, such as a mobile phone or a wearable device, may present only fairly limited information to the user. Therefore, the coaching alarm is issued only for the metric having the highest priority.

Meanwhile, after the coaching alarm, the coaching candidate metric having the highest priority is registered in the coaching target list. (S1440) The coaching target list is used as basic data to store the alarm history and reduce user fatigue due to repeated alarms. Then, it returns to the metric calculation step (S1310) and repeats the overall operation.

See if metrics improve

Hereinafter, a method for determining whether a new metric improvement occurs will be described in detail with reference to FIGS. 15 and 18 . 15 is a flowchart illustrating a step of confirming whether or not a new metric improvement has occurred.

First, the coaching guide generating unit 122 checks whether a new coaching target metric is generated (S1340) and then whether a new metric improvement occurs. (S1510)

More specifically, referring to FIG. 18 , the coaching guide generator 122 first calculates an error rate of metrics in the coaching target list. (S1511) Since the calculation of the error rate is the same as the calculation of the error rate at the time of determining the new coaching target metric, a description thereof will be omitted. The error rate P of the metrics in the calculated coaching target list is compared with the improvement reference error rate P _refi . Preferably, the improvement reference error rate (P _refi ) is set different (eg, smaller) than the reference error rate (P _ref ) when selecting a new coaching target metric. For example, the improvement reference error rate (P _refi ) is set at a level of 85 to 95% of the reference error rate (P _ref ) when selecting a new coaching target metric.

The reason for setting the improvement reference error rate and the deterioration reference error rate differently is as follows. When a metric is collected at a value close to the reference error rate, it may be repeated that the corresponding metric value exceeds or falls below the reference error rate. In this case, the metric may be repeatedly classified as improvement and deterioration unnecessarily, and there is a possibility that the metric may be classified as improved even though the improvement has not been made clearly. Therefore, by setting the standard deterioration criterion error rate to be 5 to 15% higher than the improvement criterion error rate, the status of the metric is set to improved after the metric has been improved.

Meanwhile, the coaching guide generating unit 122 checks whether the corresponding metric is improved compared to the improvement reference error rate as a result of the comparison. (S1513)

In this case, whether the corresponding metric has been improved may be determined based on different criteria for each metric. For example, an impulse metric or an instantaneous vertical load factor (IVLR) metric would imply improvement when reduced. As another example, the stride length or the interpolation metric may be determined to be improved when the deviation from the reference range is reduced because there is a reference range having an upper limit and a lower limit. Finally, in the case of a metric with the same left-right balance, it can be determined that the metric value is improved when the metric value is higher than the metric reference value.

If the corresponding metric is improved by more than the improvement reference error rate P _refi , it is determined that the new metric improvement has occurred (S1514), and in the opposite case, it is determined that the new metric improvement has not occurred. (S1515)

Referring again to FIG. 15 , when a new metric improvement occurs, the improved metric is removed from the coaching target list. (S1520) Then, it is checked whether the coaching target list is all removed. (S1530)

At this time, when all of the coaching target list is removed, an alarm message indicating that all metrics are good is output through the coaching guide output unit 130 . When the coaching target list is not all removed, an alarm message indicating that the corresponding metric is improved only for the improved metric is outputted through the coaching guide output unit 130, and the process returns to the metric calculation step (S1310) to repeat the overall operation. .

See if metrics improve

Finally, with reference to FIG. 16 , the current exercise state notification step will be described in detail. 16 is a flowchart of a current exercise state notification step.

If the new coaching target metric and metric improvement, which are the previous steps, are not newly selected, the last step, the current exercise state notification step ( S1360 ) is entered.

First, the coaching guide generating unit 122 checks whether there is no message notification for a reference time (eg, 5 minutes) when a new coaching target metric and metric improvement do not occur. (S1610) In this case, if there is no notification for the reference time, it returns to the metric calculation step (S1310) and repeats the overall operation.

However, if there is a notification during the reference time, it is checked whether there is no coaching target metric in the coaching target list. (S1610)

At this time, when there is no coaching target list, an alarm message indicating that all metrics are good is output through the coaching guide output unit 130 . (S1630) When the coaching target list still remains, an alarm message to remind at least some of the metrics in the coaching target list is output through the coaching guide output unit 130 . (S1640) After all processes are completed, the flow returns to the metric calculation step (S1310) and the entire operation is repeated.

Therefore, according to the present invention, by using equipment that is portable and can be easily worn on the body such as the head and waist, the general public can receive coaching on their own exercise posture, or the risk of injury during driving can be very easily It has a great effect that it can be measured.

In addition, in terms of device configuration, according to the present invention, there is a great advantage that only a sensor that measures a user's dynamic physical quantity, such as an acceleration sensor, can be used. Accordingly, effects such as improved user convenience and improved economic feasibility for each user or producer can also be obtained.

In addition, from a user experience point of view, if an alarm is given too frequently for coaching, the user may feel uncomfortable and may greatly interfere with efficient exercise, so the automatic coaching system of the present invention should be improved most importantly and urgently You can first correct the exercise posture to be performed.

It is to be understood that the above-described embodiments are illustrative in all respects and not restrictive, and the scope of the present invention will be indicated by the following claims rather than the foregoing detailed description. And it should be construed that all changes and modifications derived from the meaning and scope of the claims as well as equivalent concepts are included in the scope of the present invention.

100: automatic coaching system
110: sensor signal collection unit
120: control unit
121: metric calculator
122: coaching guide generation unit
130: coaching guide output unit

Claims (9)

As an automatic coaching method for a user's exercise state applied in an automatic coaching system,
a metric calculation step of calculating at least one metric by using data collected from an acceleration sensor or a location sensor worn by the user;
a new coaching target metric selection step of calculating an error rate between the calculated metric and a previously stored metric reference value, selecting a new coaching target metric based on the error rate, and outputting a coaching message related to the new coaching target metric;
a new metric improvement checking step of confirming the occurrence of new metric improvement when the new coaching target metric is not selected and outputting an alarm when the new metric improvement occurs; and
An exercise state output step of outputting an alarm for the exercise state when there is no metric improvement and the new coaching target metric,
determining the priority of the new coaching target metric by multiplying the error rate by a weight corresponding to the new coaching candidate metric,
Auto Coaching Method. According to claim 1,
The metric calculation step includes an instantaneous vertical load factor (IVLR) estimate metric based on data that has been subjected to a high-pass filter having a cutoff frequency region of 5 Hz or more after fast FFT-transformation (FFT) of the data collected from the acceleration sensor. calculating a risk-related metric;
Auto Coaching Method. 3. The method of claim 2,
In the step of calculating the injury risk-related metrics including the instantaneous vertical load factor (IVLR) estimate metric, the vertical acceleration data collected by the acceleration sensor for a certain period is converted into high-speed pre-converted data, and then the high-pass filter is applied to the data. Comprising the step of calculating an estimate based on the calculated power value,
Auto Coaching Method. The method of claim 1,
The calculating of the metric includes calculating an estimate of the center of pressure path based on the location of the user's center of mass, and calculating an exercise posture related metric including the calculated stride length metric and the interpolation metric based on the estimate of the center of pressure path. doing,
Auto Coaching Method. According to claim 1,
In the new coaching target metric selection step, after selecting a new coaching candidate metric based on the error rate, at least one metric with the highest priority among the selected new coaching candidate metrics is selected as a new coaching target metric,
Auto Coaching Method. The method of claim 1,
When selecting the new coaching target metric, the number of alarm frequencies of the new coaching candidate metric is considered,
Auto Coaching Method. According to claim 1,
The step of selecting a new coaching target metric includes storing the selected metric as the new coaching target metric in a coaching target list
Auto Coaching Method. 8. The method of claim 7,
The new metric improvement checking step includes checking whether at least one of the metrics stored in the coaching target list is improved by more than an improvement reference error rate,
Auto Coaching Method. a sensor signal collecting unit including an acceleration sensor or a position sensor worn by the user;
a metric calculator configured to calculate at least one metric by using the data collected by the sensor signal collecting unit;
Calculate the error rate between the calculated metric and the previously stored metric reference value, select a new coaching target metric based on the error rate, generate a coaching message related to the new coaching target metric selected in the coaching guide, and the new coaching target metric When this is not selected, a new metric improvement is confirmed, an alarm message is generated for the new metric improvement, and an alarm message is generated for the exercise state when the metric improvement and the new coaching target metric do not exist. wealth; and
and a coaching guide output unit for outputting a coaching message related to the new coaching target metric, an alarm message for improving the new metric, and an alarm message for the exercise state,
determining the priority of the new coaching target metric by multiplying the error rate by a weight corresponding to the new coaching candidate metric,
automatic coaching system.

Images