KR20110022917A - Apparatus and method for user adapted activity recognition - Google Patents

Abstract

PURPOSE: An apparatus and a method for user adapted activity recognition are provided to detect the movement of a user regardless of the inclination change of a sensor by changing a reference inclination according to the movement pattern of the user. CONSTITUTION: A apparatus and a method for user adapted activity recognition comprises: a tilt measurement unit using a 3-axis acceleration sensor; a reference tilt calculation unit calculating the reference tilt of the sensor according to a type of wearing the sensor or user movement; and a tilt calculation unit calculates the difference between a calculated reference tilt and a current tilt to calculate the real tilt of the user.

Description

Apparatus and Method for User Adapted Activity Recognition

The present invention relates to a user behavior detecting apparatus and a method thereof, and more particularly to a user adaptive behavior detecting apparatus and method that can detect the user's behavior regardless of the change in the inclination of the wear sensor.

In recent years, a health care service for helping elderly people, pedestrians, or children, to live their daily lives with confidence in a rapidly aging society has emerged.

One of the health care services, the behavior monitoring service using an acceleration sensor, uses a three-axis acceleration sensor to classify a user's behavior.

In the user behavior monitoring method using the three-axis acceleration sensor, it is important to accurately detect the user's behavior or posture by wearing one or more acceleration sensors.

Most algorithms are based on tilt information of sensors worn on the user's body. However, if the reference axis of the sensor is not the wearer's forward direction during the user's action, the detected actions such as lying, standing, sitting, etc. cannot be trusted.

In fact, even if the user decides where and how to wear the sensor, the reference axis is different from person to person because the body shape and posture habits differ from person to person. The reference axis may also change during activities such as running, walking, sitting, or lying down. This may cause a detection error such as detecting a standing state as a sitting state or detecting a sitting state as a lying state.

FIG. 1 is a diagram illustrating a method of classifying behavior according to a tilt of a sensor, and FIG. 2 is a diagram illustrating a change in tilt according to a wearer's sitting habit and a tilt of a wearing sensor.

Referring to FIG. 1, the sensor wearer's behavior varies according to the inclination value of the z-axis direction vector. Referring to FIG. 2, a problem may occur because the wearer's sitting position is different for each wearer.

Referring to the upper part of FIG. 2, even if the wearer wears the sensor in the forward direction, the inclination may vary according to the wearer's sitting habit. Referring to the lower part of FIG. 2, even if the wearer's posture is correct, the inclination may be changed according to the wearer's sensor wearing angle. That is, even if there are two people with the same sitting posture, if the angles of the sensors worn by each person are different from the first and the third in the lower part of FIG. . Furthermore, more serious problems may arise when the sensor is worn in various locations, such as the waist, chest, or neck.

Therefore, in spite of the same sitting state, the difference in inclination can be increased when considering two factors such as the wearer's sitting habit and the change in the wearing angle of the sensor, so that the behavior is recognized only by the absolute value of the inclination using a fixed threshold. The method shown in 1 is problematic.

The technical problem to be achieved by the present invention is a user-adaptive behavior detection device that can detect the user's behavior regardless of the change in the inclination of the sensor by dynamically changing the reference tilt value of the user's wearing sensor according to the user's movement pattern and its To provide a way.

In order to solve the above technical problem, a user adaptive behavior detection apparatus according to an embodiment of the present invention,

A tilt measuring unit obtaining a current tilt by using the 3-axis acceleration data measured by the 3-axis acceleration sensor;

A reference slope calculator for calculating a reference slope of the sensor according to a change in the tilt of the sensor due to a sensor wearing method or a user movement; And

And a slope calculator which calculates a difference between the calculated reference slope and the current slope to obtain an actual slope of the user.

In this embodiment, the reference slope calculation unit calculates the reference slope of the sensor by analyzing the user's inclination pattern while the user repeats a low dependence on the inclination.

In this embodiment, the method may further include an action determination unit that determines an action of the user according to the actual slope of the user obtained by the slope calculator.

In this embodiment, a location information acquisition unit for obtaining location information of the user; And

The apparatus may further include a communication unit configured to notify the outside of the location information acquired by the location information acquisition unit and the user's behavior state determined by the action determination unit.

In this embodiment, the current slope may be obtained by weighted moving average (WMA) or low pass filtering the three-axis acceleration data.

In this embodiment, the calculated reference slope may be obtained by obtaining an average value of the tilt pattern of the user and performing weighted moving average (WMA) or low pass filtering on the 3-axis acceleration data based on the average value.

In this embodiment, the low dependence on tilt is walking or running.

A user adaptive behavior detection method according to an embodiment of the present invention,

Calculating a reference inclination of the sensor according to a sensor wearing method or a change in inclination of the sensor due to user movement;

Obtaining a current slope of the sensor using the 3-axis acceleration data measured by the 3-axis acceleration sensor;

Calculating an actual slope of a user by calculating a difference between the calculated reference slope and the current slope; And

Determining the user's behavior according to the actual inclination of the user.

In this embodiment, the reference slope may be calculated by analyzing the tilt pattern of the user while the user repeats a low dependence on the tilt.

In this embodiment, the step of obtaining the location information of the user; And

The method may further include notifying the acquired location information and the behavior state of the user determined in the behavior determination step of the user to the outside.

In this embodiment, the current slope of the user may be obtained by weighted moving average (WMA) or low pass filtering the 3-axis acceleration data.

Subsequently, the calculated reference slope may be obtained by obtaining an average value of the tilt pattern of the user and performing 3-way acceleration data based on the average value by weighted moving average (WMA) or low pass filtering.

In this embodiment, the low dependence on tilt is walking or running.

According to the user-adaptive behavior detection apparatus according to the present invention, regardless of the sensor wearing method or the change in the tilt of the sensor due to the user's movement, falling accidents that occur frequently in real life for the elderly or pedestrians with poor mobility You can detect behavior in real time.

And by notifying this to the rescue service system, there is an effect that can be actively prepared for inadvertent accidents caused by falling.

In addition, in order to detect falling accidents and the like that occur frequently to the elderly in everyday life, by using a miniaturized sensor module that is easy to wear and hassle-free inconvenience, There is an effect that allows the elderly to live their daily lives with no restrictions in place and time, and reassure their families.

In addition, the present invention is utilized in a nursing facility or silver town to enable a quick response in the event of accidents, to reduce the labor of caregivers, there is an effect that can reduce the social welfare costs due to aging.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is said to "include" a certain component, it means that it can further include other components, without excluding other components unless specifically stated otherwise. In addition, the terms “… unit”, “… unit”, “module”, etc. described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. have.

Hereinafter, a user adaptive behavior detecting apparatus for detecting a user's actual behavior by calculating a variable reference slope of the sensor will be described with reference to FIGS. 3 and 4.

3 is a block diagram illustrating an apparatus for detecting user adaptive behavior according to an embodiment of the present invention, and FIG. 4 is a diagram illustrating a method of obtaining an actual tilt of a user.

Referring to FIG. 3, the user adaptive behavior detecting apparatus 300 according to the present invention may include a tilt measuring unit 310, a reference slope calculating unit 320, a tilt calculating unit 330, and a behavior determining unit 340. Include. In addition, the location information acquisition unit 350 and the communication unit 360 may be included.

The inclination measuring unit 310 includes a three-axis acceleration sensor 315, which can obtain a tilt for three axes using the three-axis acceleration data measured by the three-axis acceleration sensor 315. Here, the slope includes the current slope of the sensor, and the three axes refer to the x-axis, y-axis, and z-axis.

The method of obtaining the inclination using the three-axis acceleration data will be described below.

The 3-axis acceleration sensor measures an acceleration signal that includes a motion acceleration component due to acceleration and deceleration and a gravity acceleration component due to an inclination during walking or action. In this case, the gravity acceleration component may be located in the low frequency component to extract the gravity acceleration value through low pass filtering or weighted moving average.

As a result, gradient changes in the x, y, and z axis directions, that is, the x, y, and z axis angles may be obtained through a low pass filtering value or a weighted moving average value of the 3-axis acceleration data.

Equation 1 below is a calculation of weighted moving average (WMA) for the x-axis using n acceleration values. w _n means the weight of each x _n value.

[Equation 1]

Therefore, the slope of the x-axis with a maximum value max (WMA _x) and the minimum value min _(x WMA) WMA of _x can be calculated as shown in Equation (2) below.

[Equation 2]

T _x = arccos ((max (WMA _x )-WMA _x ) / (max (WMA _x ) -min (WMA _x ))

The same applies to the y and z axes to obtain slope values for the x, y and z axes.

The reference slope calculation unit 320 calculates a reference slope of the sensor according to a sensor wearing method for a predetermined time or a change in the tilt of the wearing sensor due to a user's movement.

In order to generate an algorithm that is not affected by changes in the tilt of the sensor, it is necessary to dynamically understand the orientation of the user. For example, the sensor may be worn in an unexpected location, such as a pocket on the shirt chest or waist, so that the tilt of the sensor may vary while the user is moving.

In order to capture the direction of the dynamic posture, an example such as walking is described. The actions such as lying, standing, sitting, and falling are highly dependent on the tilt value, while walking is an action that does not greatly depend on the tilt angle or the posture direction while walking.

The slope calculation unit 320 analyzes the slope pattern during the corresponding time based on the current slope information transmitted from the slope measurement unit 310 when an activity such as walking with low dependency on the slope continues for a sufficient time. By averaging the values, an inclination value with respect to the reference axis of the sensor while walking can be obtained, and this can be calculated as the reference inclination of the sensor currently worn by the user.

The slope calculator 330 receives information on the current slope of the sensor from the slope measurer 310, receives information on the reference slope calculated from the reference slope calculator 320, calculates the difference, and then calculates the difference. Find the actual slope of.

Referring to FIG. 4, the entire box means a device, and the inclinations of the x, y, and z axes mean respective inclinations extracted from the three-axis acceleration sensor.

In general, Reference Postural Orientation (P _R ) refers to a vector value of a reference slope (an inclination when standing upright) of a currently worn user, and P _R shown in FIG. 4 is a reference gradient obtained during a walking (or running) operation. Means the vector value of. This reference slope is assumed to represent the mean value of the slope values obtained during a constant walking motion.

P _N (New Postural Orientation) means the vector value of the user's slope measured in real time.

P _D A (Difference between P and _R P _N) is based on the slope P _R vector and the value that indicates the difference between the current tilt vector P _N, it can be predicted the actual inclination of the user.

When the device to the wearer is repeated after some period of action, such as walk or run, three-axis by the mobile weighted tilt information an average can be calculated for the reference gradient vector P _R of the device, by finding the P _D to detect the inclination of the user Can be. This allows us to analyze various behaviors.

P _D can be obtained by Equation 3 below.

&Quot; (3) "

Here, P _R = (x _R , y _R , z _R ) and P _N = (x _N , y _N , z _N ).

The P _D thus obtained is converted into an angle using Equation 2, and as illustrated in FIG. 1, a user's behavior based on the angle may be detected.

The behavior determination unit 340 determines the user's behavior according to the actual slope of the user calculated by the slope calculation unit 330. In this case, if it is determined that the user does not depend heavily on the tilt of the user such as walking or running (these actions may be determined using acceleration information other than the tilt information obtained by the 3-axis acceleration sensor), the reference tilt Provided to the mountain government (320) is used to calculate the reference slope.

Therefore, according to the user-adaptive behavior detection apparatus according to the present invention, as time passes, the user's adaptive ability becomes stronger and more accurate behavior determination is possible.

Although the behavior determination unit 340 is shown as being included in the inside of the device in this embodiment, it exists outside the device to receive information about the actual tilt of the user from the device by wire or wirelessly, the behavior data corresponding to the slope Can also be imported to the device.

The location information acquisition unit 350 obtains current location information of the user from a GPS satellite. The location information acquisition unit 350 may obtain location information of the user by using a known location information acquisition technology instead of using GPS satellites.

The communicator 360 notifies the external information (for example, a rescue service system) of the location information acquired by the location information acquirer 350 and the information related to the behavior status of the user determined by the behavior determination unit 340.

The location information acquisition unit 350 and the communication unit 360 may additionally be included in the device, or may be located outside the device, and may be located by the user in a wired or wireless manner from the device. You can also enter). In this case, the notification information includes a user personal number.

Meanwhile, in an embodiment of the present invention, the case where the user adaptive behavior detecting apparatus 300 includes both walking and behavior pattern information and a collection and discrimination function is described as an example. It can be divided into modules and implemented separately.

That is, the sensor module includes a tilt measurement unit 310 including a three-axis acceleration sensor 315, a reference slope calculation unit 320, and a slope calculation unit 330 to obtain current slope information and calculated reference slope information. Collect periodically and calculate the actual tilt of the user.

In addition, the sensor module may include a communication unit 360 for transmitting the calculated data to the determination module, and may include a power switch for turning on / off power, an LED for informing an on state, and a state of charge of the battery. have.

Such a sensor module is preferably manufactured in a very small size so that a general user can easily wear it. A single axis module includes a 3-axis acceleration sensor for collecting walking and behavior patterns and a chip for wirelessly transmitting calculated data to a mobile terminal. It is desirable to integrate with.

On the other hand, the determination module includes a behavior determination unit 340 to determine the user's behavior from the data transmitted from the sensor module. In this case, the determination module may include a communication unit 360 for receiving data from the sensor module. This determination module may be applied to a portable terminal.

5 is a flowchart illustrating a user adaptive behavior detection method according to an embodiment of the present invention.

Since the specific embodiment of the user adaptive behavior detection method according to the embodiment of the present invention is as described above, the operation process will be briefly described herein.

The user may attach the plurality of sensors to various positions of the body at various angles.

Referring to FIG. 5, first, a reference tilt of a sensor is calculated according to a change in tilt of a sensor due to a sensor wearing method or a user's movement (S510).

Thereafter, the current slope of the user is obtained using the three-axis acceleration data measured by the three-axis acceleration sensor (S520).

Thereafter, the difference between the calculated reference slope and the current slope is calculated to obtain an actual slope of the user (S530).

Thereafter, the user's behavior is determined according to the obtained actual tilt of the user (S540).

In addition, the current location information of the user may be acquired (S550), and notification information including the current location information of the user and the behavior state of the user may be notified to an external (for example, a rescue service system) (S560). ).

6 is a table comparing the detection accuracy according to the position and behavior of the device before and after using the user adaptive behavior detection method according to an embodiment of the present invention.

Referring to FIG. 6, it can be seen that the behavior detection accuracy reaches 88.6% when the user adaptive behavior detection method according to the present invention is used. The accuracy indicated in FIG. 6 represents the number of actions correctly detected during the action cycle. Behavior classification is performed every second.

It can be seen that walking and running are very high in sensing accuracy, while sitting and standing are not, both before and after use of the user adaptive behavior detection method according to the present invention. However, as compared with before and after use of the user adaptive behavior detection method according to the present invention, the difference between the sitting and standing in the detection accuracy is very large (that is, the behavior detection accuracy after use compared to before use) Increases very significantly).

On the other hand, when the device is worn on the chest, the experimental results are worse than when the device is worn on the left waist or right waist. This is because the device vibrates more severely and frequently when the device is worn on the chest.

The embodiments of the present invention described above are not implemented only through the apparatus and the method, but may be implemented through a program for realizing a function corresponding to the configuration of the embodiment of the present invention or a recording medium on which the program is recorded. Implementation may be easily implemented by those skilled in the art from the description of the above-described embodiments.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concepts of the present invention defined in the following claims are also provided. It belongs to the scope of rights.

1 is a diagram illustrating a behavior classification method according to a tilt of a sensor.

2 is a view showing a change in inclination according to the wear habit of the wearer and the inclination of the wearing sensor.

3 is a block diagram illustrating an apparatus for detecting user adaptive behavior according to an embodiment of the present invention.

4 is a diagram illustrating a method of obtaining an actual tilt of a user.

5 is a flowchart illustrating a user adaptive behavior detection method according to an embodiment of the present invention.

6 is a table comparing the detection accuracy according to the position and behavior of the device before and after using the user adaptive behavior detection method according to an embodiment of the present invention.

Claims (13)

A tilt measuring unit obtaining a current tilt by using the 3-axis acceleration data measured by the 3-axis acceleration sensor; A reference slope calculator for calculating a reference slope of the sensor according to a sensor wearing method or a change in the tilt of the sensor due to user movement; And And a slope calculator configured to calculate a difference between the calculated reference slope and the current slope to obtain an actual slope of the user. The method of claim 1, And the reference slope calculator calculates the reference slope of the sensor by analyzing the user's inclination pattern while the user repeats a low dependency on the inclination. The method of claim 1, And a behavior determination unit for determining the behavior of the user according to the actual slope of the user obtained by the slope calculation unit. The method of claim 1, A location information acquisition unit for obtaining location information of the user; And And a communication unit for notifying outside of the location information obtained by the location information acquisition unit and the user's behavior state determined by the behavior determination unit. The method of claim 1, And the current slope is obtained by performing weighted moving average (WMA) or low pass filtering on the 3-axis acceleration data. The method of claim 2, The calculated reference slope obtains an average value of the user's slope pattern and obtains a 3-axis acceleration data based on the average value by weighted moving average (WMA) or low pass filtering. The method of claim 2, And wherein the low dependence on the tilt is walking or running. Calculating a reference inclination of the sensor according to a sensor wearing method or a change in inclination of the sensor due to user movement; Obtaining a current slope of the sensor using the 3-axis acceleration data measured by the 3-axis acceleration sensor; Calculating an actual slope of a user by calculating a difference between the calculated reference slope and the current slope; And And determining the user's behavior according to the actual inclination of the user. The method of claim 8, The reference slope is calculated by analyzing the user's tilt pattern while the user repeats a low dependence on the tilt. The method of claim 8, Obtaining location information of the user; And And notifying the acquired location information and the behavior state of the user determined in the behavior determination step of the user to the outside. The method of claim 8, And a user's current slope is obtained by performing weighted moving average (WMA) or low pass filtering on the 3-axis acceleration data. 10. The method of claim 9, The calculated reference slope calculates an average value of the slope pattern of the user and obtains the 3-axis acceleration data based on the average value by weighted moving average (WMA) or low pass filtering. 10. The method of claim 9, The act of low dependence on the slope is a user adaptive behavior detection method of walking or running.

Images