WO2023136433A1

WO2023136433A1 - Method for calculating exercise information related to wheelchair, and wearable device

Info

Publication number: WO2023136433A1
Application number: PCT/KR2022/016370
Authority: WO
Inventors: 김창훈; 배현진; 송진우; 최창림; 강산희; 권영서; 김용훈; 박소영; 박소진
Original assignee: 삼성전자 주식회사; 세종대학교산학협력단
Priority date: 2022-01-12
Filing date: 2022-10-25
Publication date: 2023-07-20

Abstract

Disclosed in various embodiments of the present invention are a method and a device, the device comprising: an inertial sensor; a display; a memory for storing a reference signal; and a processor operatively connected to at least one from among the inertial sensor, the display and the memory, wherein the processor is configured to: measure acceleration data by using the inertial sensor; store the measured acceleration data in an acceleration buffer of the memory; acquire a KLD output value by comparing, through a Kullback-Leivler divergence algorithm, a probability distribution of the measured acceleration data to a probability distribution of a reference signal stored in the memory; determine whether the acquired KLD output value is less than a KLD reference value; and recognize same to be a wheelchair pushing movement if the acquired KLD output value is less than the KLD reference value. Various embodiments are possible.

Description

Exercise information calculation method and wearable device related to wheelchair

Various embodiments of the present disclosure disclose a method for calculating exercise information related to a wheelchair and a wearable device.

With the development of digital technology, various types of electronic devices such as mobile communication terminals, personal digital assistants (PDAs), electronic notebooks, smart phones, tablet PCs (personal computers), or wearable devices are widely used. In order to support and increase functions of these electronic devices, hardware parts and/or software parts of electronic devices are continuously being improved.

For example, electronic devices such as wearable devices are capable of being in contact with (or worn on) a user's body, and are provided in various forms such as, for example, smart watches, smart glasses, and smart bands. The wearable device may provide various functions (eg, exercise information, health information) to the user by collecting and analyzing various information (eg, biometrics and activity) about the user.

Conventional wearable devices provide exercise information based on the number of steps taken. Accordingly, a special algorithm for calculating exercise information may be required to provide exercise information based on the number of steps when walking on crutches or wearing a prosthetic leg. Furthermore, since exercise information based on the number of steps cannot be provided when riding a wheelchair, a scheme for providing exercise information related to wheelchair use to a user in a wheelchair may be required.

In various embodiments, a probability distribution of acceleration data measured by an acceleration sensor included in a wearable device and a probability distribution of a reference signal pre-stored in the wearable device are compared through a Kullback-Leivler divergence (KLD) algorithm to obtain a KLD output value, , A method and apparatus for providing exercise information related to a wheelchair by recognizing the obtained KLD output value as a wheelchair pushing motion and counting the number of times of the recognized wheelchair pushing motion may be disclosed.

A wearable device according to various embodiments of the present disclosure includes an inertial sensor, a display, a memory for storing a reference signal, and a processor operatively connected to at least one of the inertial sensor, the display, or the memory, the processor Measures acceleration data using the inertial sensor, stores the measured acceleration data in an acceleration buffer of the memory, and uses a Kullback-Leivler divergence (KLD) algorithm to obtain a probability distribution of the measured acceleration data and the memory Obtain a KLD output value by comparing the probability distribution of the reference signal stored in , determine whether the obtained KLD output value is less than the KLD reference value, and set to recognize as a wheelchair pushing operation if the obtained KLD output value is less than the KLD reference value It can be.

A method of operating a wearable device according to various embodiments of the present disclosure includes measuring acceleration data using an inertial sensor included in the wearable device, and storing the measured acceleration data in an acceleration buffer of a memory included in the wearable device. An operation of storing, an operation of determining whether the acceleration data is stored as much as the size of the acceleration buffer, and when the acceleration data is stored as much as the size of the acceleration buffer, a probability distribution of the measured acceleration data and stored in the memory through the KLD algorithm Obtaining a KLD output value by comparing probability distributions of reference signals, determining whether the obtained KLD output value is less than the KLD reference value, and recognizing as a wheelchair pushing operation if the obtained KLD output value is less than the KLD reference value Actions may be included.

According to various embodiments, by comparing the probability distribution of acceleration data measured in a wearable device with a probability distribution of a reference signal using a KLD algorithm, recognizing a wheelchair pushing motion based on the KLD output value, and recognizing the recognized wheelchair pushing motion number of times can be counted.

According to various embodiments, the detection accuracy of the wheelchair pushing motion may be improved by determining the wheelchair pushing motion as a standard motion and determining all other motions as atypical motions.

According to various embodiments, detection accuracy of a wheelchair pushing motion may be improved by further using not only the KLD output value obtained through the KLD algorithm but also the KLD average value and acceleration variance value.

According to various embodiments, by counting the number of wheelchair pushing motions as well as counting the number of steps, motion information related to the wheelchair (eg, number of times of pushing the wheelchair, time of pushing the wheelchair) efficiently can provide

1 is a block diagram of a wearable device according to various embodiments.

2 is a front perspective view of a wearable device according to various embodiments.

3 is a diagram illustrating a rear perspective view of a wearable device according to various embodiments.

4 is a diagram illustrating acceleration data related to a wheelchair pushing motion according to various embodiments.

5 is a flowchart illustrating a method of operating a wearable device according to various embodiments.

6 is a diagram illustrating an example of storing acceleration data in an acceleration buffer according to time according to various embodiments.

7 is a flowchart illustrating a method of recognizing a wheelchair pushing motion of a wearable device according to various embodiments.

8A to 8C are diagrams illustrating an example of utilizing a buffer included in a wearable device according to various embodiments.

9 is a diagram illustrating an example of classifying motion based on acceleration data in a wearable device according to various embodiments.

10 is a flowchart illustrating a method of counting a wheelchair pushing motion of a wearable device according to various embodiments.

11 is a diagram illustrating an example of setting a window for acceleration data in a wearable device according to various embodiments.

12 is a diagram illustrating an example of selecting an adaptive threshold in a wearable device according to various embodiments.

13 is a diagram illustrating an example of preventing a count error of a wheelchair pushing motion in a wearable device according to various embodiments.

14 is a diagram illustrating a difference in acceleration data according to movement in a wearable device according to various embodiments.

15 is a diagram illustrating a user interface related to pushing a wheelchair in a wearable device according to various embodiments.

Various embodiments of this document and terms used therein are not intended to limit the technical features described in this document to specific embodiments, but should be understood to include various modifications, equivalents, or substitutes of the embodiments. In connection with the description of the drawings, like reference numbers may be used for like or related elements. The singular form of a noun corresponding to an item may include one item or a plurality of items, unless the relevant context clearly dictates otherwise.

In this document, "A or B", "at least one of A and B", "at least one of A or B", "A, B or C", "at least one of A, B and C", and "A Each of the phrases such as "at least one of , B, or C" may include any one of the items listed together in that phrase, or all possible combinations thereof. Terms such as "first", "second", or "first" or "secondary" may simply be used to distinguish that component from other corresponding components, and may refer to that component in other respects (eg, importance or order) is not limited. A (eg, first) component is said to be "coupled" or "connected" to another (eg, second) component, with or without the terms "functionally" or "communicatively." When mentioned, it means that the certain component may be connected to the other component directly (eg by wire), wirelessly, or through a third component.

The term "module" used in various embodiments of this document may include a unit implemented in hardware, software, or firmware, and is interchangeable with terms such as, for example, logic, logical blocks, parts, or circuits. can be used as A module may be an integrally constructed component or a minimal unit of components or a portion thereof that performs one or more functions. For example, according to one embodiment, the module may be implemented in the form of an application-specific integrated circuit (ASIC).

1 is a block diagram of a wearable device according to various embodiments.

Referring to FIG. 1 , the wearable device 101 includes a memory 110, a processor 120, an input module 130, a display module 140, an inertial sensor 150, a communication module 160, and a microphone 170. , a speaker 175, a charging module 180, a battery 185, and/or an interface 190. The wearable device 101 may communicate with at least one of an external electronic device or a server through a network (eg, a short-distance wireless communication network or a long-distance wireless communication network). In the wearable device 101, at least one of the components described above may be omitted or one or more other components (eg, a camera module) may be added. The input module 130 and the display module 140 may be integrated into one module (eg, a touch screen).

The memory 110 may store various data used by at least one component of the wearable device 101 (eg, the processor 120 or the inertial sensor 150). The data may include, for example, input data or output data for software (eg, a program) and commands related thereto. The memory 110 may include volatile memory or non-volatile memory. The program may be stored as software in the memory 110 and may include, for example, an operating system, middleware, or applications.

The processor 120 may execute software to control at least one other component (eg, hardware or software component) of the wearable device 101 connected to the processor 120, and perform various data processing or calculations. can According to one embodiment, as at least part of data processing or operation, the processor 120 stores instructions or data received from other components (eg, the inertial sensor 150 or the communication module 160) in volatile memory and , it can process commands or data stored in volatile memory, and store the resulting data in non-volatile memory. According to one embodiment, the processor 120 may include a main processor (eg, a central processing unit or an application processor) or a secondary processor (eg, a graphics processing unit, a neural processing unit (NPU)) that may operate independently or together therewith. , image signal processor, sensor hub processor, or communication processor).

The processor 120 measures acceleration data (or an acceleration value or an acceleration sensed value) using the inertial sensor 150, stores the measured acceleration data in an acceleration buffer (or first acceleration buffer), and stores the measured acceleration data in the acceleration buffer. The probability distribution of the stored acceleration data and the probability distribution of the reference signal are compared, the comparison result is obtained as a KLD output value, and the wheelchair pushing motion can be recognized based on the obtained KLD output value. The reference signal is previously stored in the memory 110, and the processor 120 may compare a probability distribution of the measured acceleration data with a probability distribution of the reference signal using a KLD algorithm. The processor 120 determines whether acceleration data equal to the size of the acceleration buffer is stored, measures acceleration data until acceleration data equal to the size of the acceleration buffer is stored, and stores the measured acceleration data in the acceleration buffer. can be repeated. The processor 120 may recognize the wheelchair pushing motion when the KLD output value is less than the KLD reference value, and recognize it as not the wheelchair pushing motion when the KLD output value is greater than or equal to the KLD reference value.

According to various embodiments, when acceleration data is stored as much as the size of the acceleration buffer, the processor 120 may calculate KLD and store the KLD output value in the KLD buffer. The processor 120 may repeatedly perform operations of calculating KLD based on the acceleration data stored in the acceleration buffer and storing the KLD output value in the KLD buffer until KLD output values equal to the size of the KLD buffer are stored. When KLD output values equal to the size of the KLD buffer are stored, the processor 120 may calculate a KLD average value based on the KLD output values stored in the KLD buffer. When acceleration data corresponding to the size of the acceleration buffer is stored, the processor 120 may identify the last acceleration data stored in the acceleration buffer and store the identified acceleration data in a second acceleration buffer. The processor 120 repeatedly performs an operation of identifying the last acceleration data (eg, the latest acceleration data) stored in the acceleration buffer and storing it in the second acceleration buffer until acceleration data equal to the size of the second acceleration buffer is stored. can do. When acceleration data corresponding to the size of the second acceleration buffer is stored, the processor 120 may calculate an acceleration variance value based on the acceleration data stored in the second acceleration buffer.

According to various embodiments, the processor 120 performs a wheelchair pushing operation when the KLD output value is less than the KLD reference value, the calculated KLD average value is less than the KLD average threshold value, and the calculated acceleration variance value exceeds the variance threshold value. can be recognized as The processor 120 performs the wheelchair pushing operation when at least one of the KLD output value is greater than the KLD reference value, the calculated KLD average value is greater than or equal to the KLD average threshold value, and the calculated acceleration variance value is less than or equal to the variance threshold value. It can be judged that this is not the case.

According to various embodiments, the processor 120 may set a window to a set number of acceleration data from among the acceleration data stored in the second acceleration buffer when recognizing the wheelchair pushing motion. The processor 120 may detect an acceleration peak within a window, and return (or identify) acceleration data detected as a peak when an acceleration peak within the window is detected. When an acceleration peak within the window is not detected, the processor 120 may update the window by adding a window index. The processor 120 may detect a minimum value of acceleration data stored in the second acceleration buffer, and select (or determine) an adaptive threshold based on the minimum value of the acceleration data.

According to various embodiments, the processor 120 determines whether the returned acceleration data is less than the adaptive threshold, and if the returned acceleration data is less than the adaptive threshold, the two windows detected as acceleration peaks It may be determined whether a difference between indices exceeds an index threshold. The processor 120 may count the number of pushing motions of the wheelchair when a difference between two window indices detected as an acceleration peak exceeds an index threshold. The processor 120 may not count the wheelchair pushing motion when the returned acceleration data is equal to or greater than the adaptive threshold or when a difference between two window indices detected as an acceleration peak is equal to or less than the index threshold.

For example, when the wearable device 101 includes a main processor and an auxiliary processor, the auxiliary processor may use less power than the main processor or may be set to be specialized for a designated function. A secondary processor may be implemented separately from, or as part of, the main processor. A secondary processor may, for example, perform a wearable device on behalf of the main processor while the main processor is in an inactive (eg sleep) state, or together with the main processor while the main processor is in an active (eg application execution) state. It is possible to control at least some of functions or states related to at least one of the components (eg, the display module 140, the inertial sensor 150, or the communication module 160). According to one embodiment, an auxiliary processor (eg, an image signal processor or a communication processor) may be implemented as part of other functionally related components (eg, the communication module 160).

According to an embodiment, the auxiliary processor (eg, a neural network processing device) may include a hardware structure specialized for processing an artificial intelligence model. AI models can be created through machine learning. Such learning may be performed, for example, in the wearable device 101 itself where the artificial intelligence model is performed, or may be performed through a separate server. The learning algorithm may include, for example, supervised learning, unsupervised learning, semi-supervised learning or reinforcement learning, but in the above example Not limited. The artificial intelligence model may include a plurality of artificial neural network layers. Artificial neural networks include deep neural networks (DNNs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), restricted boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recurrent deep neural networks (BRDNNs), It may be one of deep Q-networks or a combination of two or more of the foregoing, but is not limited to the foregoing examples. The artificial intelligence model may include, in addition or alternatively, software structures in addition to hardware structures.

The input module 130 may receive a command or data to be used in a component (eg, the processor 120) of the wearable device 101 from the outside of the wearable device 101 (eg, a user). For example, the input module 130 may include a microphone, mouse, keyboard, key (eg, button), or digital pen (eg, stylus pen).

The display module 140 may visually provide information to the outside of the wearable device 101 (eg, a user). For example, the display module 140 may include a display, a hologram device, or a projector and a control circuit for controlling the device. According to one embodiment, the display module 140 may include a touch sensor set to detect a touch or a pressure sensor set to measure the intensity of force generated by the touch.

The inertial sensor 150 is a sensor that detects force, and may also be referred to as a 6-axis sensor by combining acceleration (3 axes) and Gyro (3 axes). Alternatively, the inertial sensor 150 may be referred to as a game rotation vector sensor by integrating an acceleration sensor and a gyro sensor. The accelerometer measures dynamic forces such as acceleration, vibration, and shock of an object. It may be a sensor that The acceleration sensor can detect the motion state of an object and can be used for various purposes. The wearable device 101 may recognize a wheelchair pushing motion based on acceleration data measured by the inertial sensor 150 . A gyroscope may be a sensor that measures the angular velocity of an object. Unlike an acceleration sensor that measures the acceleration of an object, the gyro sensor may measure the angular velocity of the object. The angular velocity may mean a rotational speed (or angle) per time. The gyro sensor may also be referred to as a gyroscope.

The communication module 160 may support establishment of a direct (eg, wired) communication channel or wireless communication channel between the wearable device 101 and an external electronic device or server, and communication through the established communication channel. The communication module 160 may include one or more communication processors that operate independently of the processor 120 (eg, an application processor) and support direct (eg, wired) communication or wireless communication. According to one embodiment, the communication module 160 is a wireless communication module (eg, a cellular communication module, a short-distance wireless communication module, or a global positioning system (GPS) communication module) or a wired communication module (eg, a local area network (LAN)). communication module or power line communication module).

Among these communication modules, the corresponding communication module is a short-range communication network such as Bluetooth, wireless fidelity (WiFi) direct, or infrared data association (IrDA), or a legacy cellular network, a 5G network, a next-generation communication network, the Internet, or a computer network (eg, LAN). or WAN) may communicate with an external electronic device through a long-distance communication network. These various types of communication modules may be integrated as one component (eg, a single chip) or implemented as a plurality of separate components (eg, multiple chips). The wireless communication module may identify or authenticate the wearable device 101 within a communication network such as a short-distance communication network or a long-distance communication network using subscriber information (eg, International Mobile Subscriber Identity (IMSI)) stored in the subscriber identification module. .

The microphone 170 may convert sound into an electrical signal or vice versa. According to one embodiment, the microphone 170 may acquire sound (or audio) and convert it into an electrical signal.

The speaker 175 may output an audio (or sound) signal to the outside of the wearable device 101 . The speaker 175 may include a receiver. The speaker 175 may be used for general purposes such as multimedia playback or recording playback. A receiver may be used to receive an incoming call. According to one embodiment, the receiver may be implemented separately from the speaker 175 or as part of it.

The charging module 180 may manage power supplied to the wearable device 101 . The charging module 180 may charge the battery 185 with power received through the interface 190 . The charging module 180 may be implemented as at least part of a power management integrated circuit (PMIC).

The battery 185 may supply power to at least one component of the wearable device 101 . According to one embodiment, the battery 185 may include, for example, a non-rechargeable primary battery, a rechargeable secondary battery, or a fuel cell.

The interface 190 may support one or more specified protocols that may be used to directly or wirelessly connect the wearable device 101 to an external electronic device (or charger). According to one embodiment, the interface 190 may include a high definition multimedia interface (HDMI), a universal serial bus (USB) interface, an SD card interface, or an audio interface. The interface 190 may include a connector through which the wearable device 101 is physically connected to an external electronic device.

According to an embodiment, commands or data may be transmitted or received between the wearable device 101 and an external electronic device through a server. The external electronic device may be the same as or different from the wearable device 101 . According to an embodiment, all or part of operations executed in the wearable device 101 may be executed in an external electronic device. For example, when the wearable device 101 needs to perform a certain function or service automatically or in response to a request from a user or other device, the wearable device 101 instead of executing the function or service by itself Alternatively or additionally, one or more external electronic devices may be requested to perform the function or at least part of the service. One or more external electronic devices receiving the request may execute at least a part of the requested function or service or an additional function or service related to the request, and deliver a result of the execution to the wearable device 101 . The wearable device 101 may process the result as it is or additionally and provide it as at least part of a response to the request.

To this end, for example, cloud computing, distributed computing, mobile edge computing (MEC), or client-server computing technology may be used. The wearable device 101 may provide an ultra-low latency service using, for example, distributed computing or mobile edge computing. In another embodiment, the external electronic device may include an internet of things (IoT) device. The server may be an intelligent server using machine learning and/or neural networks. The wearable device 101 can be applied to intelligent services (eg, smart home, smart city, smart car, or health care) based on 5G communication technology and IoT-related technology.

Various embodiments of this document may be implemented as software including one or more instructions stored in a storage medium readable by a machine (eg, the wearable device 101). For example, the processor 120 of the wearable device 101 may call at least one command among one or more commands stored from a storage medium and execute it. This enables the device to be operated to perform at least one function according to the at least one command invoked. The one or more instructions may include code generated by a compiler or code executable by an interpreter. The device-readable storage medium may be provided in the form of a non-transitory storage medium. Here, 'non-temporary' only means that the storage medium is a tangible device and does not contain a signal (e.g. electromagnetic wave), and this term refers to the case where data is stored semi-permanently in the storage medium. It does not discriminate when it is temporarily stored.

2 is a front perspective view of a wearable device according to various embodiments, and FIG. 3 is a rear perspective view of a wearable device according to various embodiments.

Referring to FIGS. 2 and 3 , a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments has a first surface (or front surface) 210A and a second surface (or rear surface) 210B. , And a housing 210 including a side surface 210C surrounding a space between the first surface 210A and the second surface 210B, and connected to at least a portion of the housing 210 and the wearable device 200 It may include

fastening members

250 and 260 capable of being detached from the user's body (eg, wrist, ankle, etc.). In another embodiment (not shown), the housing may refer to a structure forming some of the first face 210A, the second face 210B, and the side face 210C of FIG. 2 .

According to one embodiment, the first surface 210A may be formed by a front plate 201 (eg, a glass plate or a polymer plate including various coating layers) that is substantially transparent at least in part. The second face 210B may be formed by the substantially opaque back plate 207 . Back plate 207 may be formed, for example, of coated or tinted glass, ceramic, polymer, metal (eg, aluminum, stainless steel (STS), or magnesium), or a combination of at least two of the foregoing. can The side surface 210C is coupled to the front plate 201 and the back plate 207 and may be formed by a side bezel structure (or "side member") 206 including metal and/or polymer. In some embodiments, the back plate 207 and the side bezel structure 206 may be integrally formed and may include the same material (eg, a metal material such as aluminum). The binding

members

250 and 260 may be formed of various materials and shapes. Integral and plurality of unit links may be formed to flow with each other by woven material, leather, rubber, urethane, metal, ceramic, or a combination of at least two of the above materials.

According to various embodiments, the wearable device 200 includes a display 220 (eg, the display module 140 of FIG. 1 ),

audio modules

205 and 208 , and a sensor module 211 (eg, the inertial module of FIG. 1 ). sensor 150),

key input devices

202, 203, 204 (eg, the input module 130 of FIG. 1) and a connector hole 209. In some embodiments, the wearable device 200 may omit at least one of the components (eg, the

key input devices

202, 203, and 204, the connector hole 209, or the sensor module 211) or have other components. Additional elements may be included.

The display 220 may be exposed through a substantial portion of the front plate 201 , for example. The shape of the display 220 may be a shape corresponding to the shape of the front plate 201, and may have various shapes such as a circle, an ellipse, or a polygon. The display 220 may be coupled to or disposed adjacent to a touch sensing circuit, a pressure sensor capable of measuring the intensity (pressure) of a touch, and/or a fingerprint sensor.

The

audio modules

205 and 208 may include a microphone hole 205 and a speaker hole 208 . A microphone for acquiring external sound may be disposed inside the microphone hole 205, and in some embodiments, a plurality of microphones may be disposed to detect the direction of sound. The speaker hole 208 can be used as an external speaker and a receiver for a call.

The sensor module 211 may generate an electrical signal or data value corresponding to an internal operating state of the wearable device 200 or an external environmental state. The sensor module 211 may include, for example, a biometric sensor module 211 (eg, an HRM sensor) disposed on the second surface 210B of the housing 210 . The wearable device 200 includes a sensor module not shown, for example, a gesture sensor, a gyro sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a grip sensor, a color sensor, an IR (infrared) sensor, a biometric sensor, a temperature sensor, At least one of a humidity sensor and an illuminance sensor may be further included.

The

key input devices

202, 203, and 204 include a wheel key 202 disposed on a first surface 210A of the housing 210 and rotatable in at least one direction, and/or a side surface 210C of the housing 210. ) may include side

key buttons

202 and 203 disposed on. The wheel key may have a shape corresponding to the shape of the front plate 202 . In other embodiments, the wearable device 200 may not include some or all of the

key input devices

202, 203, and 204, and the

key input devices

202, 203, and 204 that are not included may be displayed on the display 220. It may be implemented in other forms such as soft keys on the screen. The connector hole 209 may accommodate a connector (eg, a USB connector) for transmitting and receiving power and/or data to and from an external electronic device and a connector for transmitting and receiving an audio signal to and from an external electronic device. Other connector holes (not shown) may be included. The wearable device 200 may further include, for example, a connector cover (not shown) that covers at least a portion of the connector hole 209 and blocks external foreign substances from entering the connector hole.

The binding

members

250 and 260 may be detachably attached to at least a partial region of the housing 210 using the locking

members

251 and 261 . The

fastening members

250 and 260 may include one or more of a fixing member 252 , a fixing member fastening hole 253 , a band guide member 254 , and a band fixing ring 255 .

The fixing member 252 may be configured to fix the housing 210 and the

fastening members

250 and 260 to a part of the user's body (eg, wrist, ankle, etc.). The fixing member fastening hole 253 corresponds to the fixing member 252 to fix the housing 210 and the

fastening members

250 and 260 to a part of the user's body. The band guide member 254 is configured to limit the movement range of the fixing member 252 when the fixing member 252 is fastened to the fixing member fastening hole 253, so that the

fastening members

250 and 260 are attached to a part of the user's body. It can be tightly bonded. The band fixing ring 255 may limit the movement range of the

fastening members

250 and 260 in a state in which the fixing member 252 and the fixing member fastening hole 253 are fastened.

Referring to FIG. 4 , wheelchair pushing motions can be largely classified into four types. For example, the wheelchair pushing motion classification 410 shows a hand movement according to a wheelchair pushing motion, and may show a situation in which the wheelchair moves in a left direction. The wheelchair pushing action may include ① grabbing the handle of the wheelchair (holding), ② pushing the wheelchair wheel (pushing), ③ releasing the hand from the handle (releasing), and ④ recovering the hand (recovery). . Although it is classified into four operations for description of the invention, it may be classified into more or less than four operations depending on the implementation. Below, acceleration characteristics that can be detected when a user rides a wheelchair while wearing a wearable device (eg, the wearable device 101 of FIG. 1 ) will be described.

① The operation of grabbing the handle is an operation of grabbing the wheel to rotate the wheelchair, and as the user's hand touches the wheel and stops, the wearable device 101 receives acceleration, and when the operation of retrieving the hand is followed, the direction of movement of the hand is change, and large accelerations can occur. The grabbing motion may generate acceleration data in the form of a peak.

② The action of pushing the wheel is an action of directly rotating the wheel of the wheelchair. Although the speed may continuously increase, the acceleration may not occur significantly, so a peak may not occur. In the case of motion No. 2, the length of the motion of pushing the wheel may change according to the pushing speed of the wheelchair.

③ The motion of releasing the hand is a motion in which the user's hand changes from the forward direction of the wheelchair to the opposite direction. As the second motion is performed, the increased speed temporarily stops and large acceleration may occur. Accordingly, the largest peak among accelerations performing the entire pushing motion may occur.

④ The hand-retrieving action is an action in which the hand is retrieved in the opposite direction of the wheelchair forward after performing the action when the user releases his hand, and since it occurs without a change in direction, a large peak action may not occur.

A graph 450 shows acceleration data generated when a wheelchair pushing motion is performed. Looking at the graph 450, it can be seen that the acceleration characteristics appear differently according to ①②③④ corresponding to the wheelchair pushing motion classification 410, and the acceleration characteristics repeatedly occur. In the

graph

450, ③ the action of releasing the hand is an action that generates the largest peak in acceleration data (451, 455), and the action of holding the knob is an action that generates the smallest peak in acceleration data (453, 457) may be marked with

According to various embodiments, acceleration characteristics according to the wheelchair pushing motion classification 410 may appear more distinctly in an axis of the acceleration sensor that coincides with the direction of the user's fingertip. An axis of the acceleration sensor that coincides with the direction of the user's fingertip may vary depending on a position where the acceleration sensor (eg, the inertial sensor 150 of FIG. 1 ) is disposed on the wearable device 101 . In the present invention, the axis of the acceleration sensor corresponding to the direction of the user's fingertip is set as the 'x-axis', and the pushing motion of the wheelchair can be recognized based on the acceleration data of the x-axis. However, the axis of the acceleration sensor used for pushing the wheelchair may be the y-axis or the z-axis according to the position where the acceleration sensor is disposed in the wearable device 101 .

According to various embodiments, a user may wear the wearable device 101 on his or her left hand or right hand. As the user wears the wearable device 101 on the left hand or the right hand, the x-axis of the acceleration sensor may be divided into a + direction or a - direction. When the acceleration data measured on the x-axis of the accelerometer has a -value, the wheelchair pushing motion can be recognized using sign inversion of the x-axis.

The wearable device 101 recognizes the wheelchair pushing motion by using the acceleration data measured by the acceleration sensor according to the wheelchair pushing motion as an input of the KLD (Kullback-Leivler divergence) algorithm, and recognizes the wheelchair pushing motion, and then the measured acceleration data It is possible to count the number of wheelchair pushing motions based on .

A wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments of the present invention includes an inertial sensor (eg, the inertial sensor 150 of FIG. 1 ), a display (eg, the display module 140 of FIG. 1 ) )), a memory for storing a reference signal (eg, memory 110 of FIG. 1 ), and a processor operatively connected with at least one of the inertial sensor, the display, or the memory (eg, processor 120 of FIG. 1 ). )), wherein the processor measures acceleration data using the inertial sensor, stores the measured acceleration data in an acceleration buffer of the memory, and uses a Kullback-Leivler divergence (KLD) algorithm to measure the measured acceleration data. A KLD output value is obtained by comparing the probability distribution of the acceleration data with the probability distribution of the reference signal stored in the memory, determining whether the obtained KLD output value is less than the KLD reference value, and the obtained KLD output value is less than the KLD reference value. , can be set to be recognized as a wheelchair pushing motion.

The processor determines the size of the acceleration buffer as much as the size of the reference signal, determines whether the acceleration data is stored as much as the size of the acceleration buffer, and when the acceleration data is stored as much as the size of the acceleration buffer, the measured It may be set to compare the probability distribution of the acceleration data with the probability distribution of the reference signal stored in the memory.

The processor may be configured to store acceleration data corresponding to an axis set according to a position of the inertial sensor disposed in the wearable device in the acceleration buffer.

The processor calculates a KLD average value based on the obtained KLD output value, calculates an acceleration variance value based on the measured acceleration data, and calculates an acceleration variance value based on at least one of the KLD output value, the KLD average value, or the acceleration variance value. It can be set to recognize the pushing motion of the wheelchair.

The processor stores the obtained KLD output value in the KLD buffer, determines whether the KLD output value is stored as much as the size of the KLD buffer, and when the KLD output value is stored as much as the size of the KLD buffer, the KLD output value stored in the KLD buffer It can be set to calculate an average value.

The acceleration buffer is referred to as a first acceleration buffer, and the processor stores the acceleration data stored in the first acceleration buffer in a second acceleration buffer when the size of the first acceleration buffer is stored, and It may be set to determine whether acceleration data is stored as much as the size and, when acceleration data is stored as much as the size of the second acceleration buffer, to calculate an acceleration variance value of the acceleration data stored in the second acceleration buffer.

The processor may be configured to store the last acceleration data stored in the first acceleration buffer in the second acceleration buffer when the size of the first acceleration buffer is stored.

The processor may be set to recognize the wheelchair pushing motion when at least one of the KLD output value, the KLD average value, and the acceleration variance value corresponds to a set condition.

The set condition may be set to include at least one of the following: the KLD output value is less than the KLD reference value, the KLD average value is less than the KLD average threshold value, and the acceleration variance value exceeds the variance threshold value.

The acceleration buffer is referred to as a first acceleration buffer, and the processor stores the acceleration data stored in the first acceleration buffer in a second acceleration buffer when it is recognized as the wheelchair pushing motion, and the acceleration stored in the second acceleration buffer A window is set for a set number of acceleration data among the data, an acceleration peak within the window is detected, it is determined whether the acceleration data detected as an acceleration peak within the window is less than an adaptive threshold, and according to the determination result Based on this, it may be set to count the number of times of the wheelchair pushing motion.

The processor may be set to update the window by identifying acceleration data detected as the acceleration peak when an acceleration peak within the window is detected, and adding a window index when an acceleration peak within the window is not detected. .

The processor detects a minimum value of acceleration data stored in the second acceleration buffer, determines an adaptive threshold based on the minimum value of the acceleration data, and determines whether acceleration data detected as an acceleration peak within the window is less than the adaptive threshold. It can be set to determine whether or not.

The processor may be configured to change the adaptive threshold whenever new acceleration data is stored in the second acceleration buffer or the window is updated.

The processor, when the acceleration data detected as the acceleration peak within the window is less than the adaptive threshold, determines whether the difference between the two window indices in which the acceleration peak is detected exceeds the index threshold, and the two window indices in which the acceleration peak is detected When the difference between the values exceeds the index threshold, the number of times of the wheelchair pushing motion may be set to be counted.

5 is a flowchart 500 illustrating a method of operating a wearable device according to various embodiments.

Referring to FIG. 5 , in operation 501, a processor (eg, processor 120 of FIG. 1 ) of a wearable device (eg, wearable device 101 of FIG. 1 ) according to various embodiments generates an acceleration sensor (eg, FIG. Acceleration data (or acceleration value, acceleration sensed value) may be measured using the inertial sensor 150 of 1). Acceleration characteristics according to the pushing motion of the wheelchair may appear more distinctly in the axis of the inertial sensor 150 that coincides with the direction of the user's fingertip. The axis of the inertial sensor 150 coinciding with the direction of the user's fingertip may vary depending on the position where the inertial sensor 150 is disposed on the wearable device 101 . The processor 120 may identify (or extract) acceleration data of the 'x-axis' that matches the direction of the user's fingertip from among the acceleration data measured by the inertial sensor 150 . The acceleration data measured by the inertial sensor 150 includes acceleration data of the x-axis, y-axis, and z-axis, and the processor 120 does not use the acceleration data of the y-axis and z-axis to recognize the wheelchair pushing motion, and Only axis acceleration data can be used. Depending on the position where the inertial sensor 150 is disposed on the wearable device 101, the axis of the acceleration sensor used for pushing the wheelchair may be the y-axis or the z-axis. Hereinafter, it is described using the x-axis, but the present invention is not limited by the description.

In operation 503, the processor 120 may store the measured acceleration data in an acceleration buffer (or a first acceleration buffer). The first acceleration buffer may store acceleration data measured by the inertial sensor 150 in real time. The processor 120 may process (eg, filter) measured acceleration data (eg, raw data) and store the measured acceleration data (eg, raw data) in the first acceleration buffer. When the size of the first acceleration buffer is 120, the processor 120 may store 120 pieces of acceleration data in the first acceleration buffer in the measured order. The processor 120 may repeatedly perform

operations

501 and 503 until acceleration data corresponding to the size of the first acceleration buffer is stored. The processor 120 may perform operation 505 when all acceleration data is stored as much as the size of the first acceleration buffer. Although operation 501 and operation 503 are separately described in the drawing, the processor 120 performs operation 501 and operation 503 as one operation, and stores the measured acceleration data in the first acceleration buffer whenever acceleration data is measured. can be saved

In operation 505, the processor 120 may compare a probability distribution of the measured acceleration data with a probability distribution of the reference signal. The processor 120 may use a probability distribution comparison method, for example, a KLD algorithm, to determine whether the measured acceleration data corresponds to a wheelchair pushing motion. Using the fact that the same signal characteristics occur when performing a wheelchair pushing motion, acceleration data generated when a wheelchair pushing motion (eg ①②③④) is performed once can be defined as a reference signal. . A probability distribution of the defined reference signal may be calculated using a kernel function, and the calculated reference signal may be previously stored in a memory of the wearable device 101 (eg, the memory 110 of FIG. 1 ). The reference signal may be a reference value for determining whether the measured acceleration data corresponds to a wheelchair pushing motion.

According to various embodiments, in order to use the probability distribution comparison method, the size of the first acceleration buffer may be set as much as the size at which the reference signal is stored. For example, when 120 reference signals are stored, the size of the first acceleration buffer is set to store 120 acceleration data, and when 150 reference signals are stored, the size of the first acceleration buffer is also set. It can be set so that 150 pieces of acceleration data can be stored.

In operation 507, the processor 120 may obtain the comparison result as a KLD output value. For example, the processor 120 calculates a probability distribution of 120 acceleration data stored in the first acceleration buffer (eg, first to 120th measured acceleration data) and 120 reference signals by using the KLD algorithm. One KLD output value can be obtained by comparing the probability distributions. KL-divergence can be considered as an information entropy difference between the probability distribution of the measured acceleration data and the probability distribution of the reference signal. If the calculated KL-divergence (eg, KLD output value) has a large value, it means that the information entropy difference is large, which may mean that the similarity between the probability distribution of the measured acceleration data and the probability distribution of the reference signal is small. Conversely, when KL-divergence has a small value, it means that the information entropy difference is small, which means that the probability distribution of the measured acceleration data and the probability distribution of the reference signal have similar probability distributions. Since the KLD algorithm corresponds to a known technology, a detailed description thereof may be omitted.

In operation 509, the processor 120 may recognize a wheelchair pushing motion based on the obtained KLD output value. If the obtained KLD output value is less than the KLD reference value, the processor 120 may determine that the measured acceleration data and the reference signal have a high similarity and recognize the wheelchair pushing motion. If the obtained KLD output value is greater than or equal to the KLD reference value, the processor 120 may determine that the measured acceleration data and the reference signal have a small similarity and may not recognize the wheelchair pushing motion. The KLD reference value may mean a threshold for determining the similarity of two probability distributions as KLD. The KLD reference value may be previously stored in the memory 110. The present invention is for recognizing a wheelchair pushing motion, and if it is not a wheelchair pushing motion (eg, walking, running, etc.), it can be determined as an atypical motion. If it is determined in operation 509 that the wheelchair pushing motion is not, the processor 120 may perform an algorithm for obtaining exercise information (eg, number of steps, walking time, running time, etc.) according to the atypical motion. Since operations such as calculating the number of steps, calculating the walking time, or calculating the running time correspond to the prior art, a detailed description thereof may be omitted.

When it is recognized as a wheelchair pushing motion in operation 509, the processor 120 repeatedly performs operations 501 to 509 to obtain a probability distribution of newly measured acceleration data (eg, second to 121st measured acceleration data). It is possible to determine the wheelchair pushing motion by comparing the probability distribution of the reference signal. Also, if the processor 120 recognizes the wheelchair pushing motion, the processor 120 may perform the operation of FIG. 10 to count the number of times of the wheelchair pushing motion.

Referring to FIG. 6 , a processor (eg, processor 120 of FIG. 1 ) of a wearable device (eg, wearable device 101 of FIG. 1 ) according to various embodiments includes an acceleration sensor (eg, inertial sensor of FIG. 1 ). (150) may be used to measure acceleration data (or acceleration value or acceleration sensing value). Referring to first reference numeral 610, the processor 120 stores acceleration data measured in real time according to a time sequence (eg, time index 601) in an acceleration buffer (or first acceleration buffer) 603. can For example, in the time index 601, the first acceleration data measured at the first time (1) is stored in the acceleration buffer 603, and the second acceleration data measured at the second time (2) is stored in the acceleration buffer 601. can The processor 120 continues to measure and store acceleration data in the acceleration buffer 603 until the acceleration data is stored as much as the size of the acceleration buffer 603 (eg, 120), and stores the acceleration data as much as the size of the acceleration buffer 603. When is stored, a first KLD output value may be obtained by comparing a probability distribution of acceleration data (eg, first to 120th acceleration data) stored in the acceleration buffer with a probability distribution of the reference signal.

Referring to the second reference numeral 630, the 121st acceleration data is stored in the acceleration buffer 603 according to the 121st new time index 601, and the processor 120 stores the acceleration data stored in the acceleration buffer (eg, the second A second KLD output value may be obtained by comparing the probability distribution of the acceleration data to the 121st acceleration data) with the probability distribution of the reference signal. Then, when the time index 601 is the 122nd, the 122nd acceleration data is stored in the acceleration buffer 603, and the processor 120 stores the acceleration data stored in the acceleration buffer (eg, the third acceleration data to the 122nd acceleration data). A third KLD output value can be obtained by comparing the probability distribution with the probability distribution of the reference signal.

The processor 120 stores new acceleration data in the acceleration buffer 601 according to the sampling time of the inertial sensor 150, and the acceleration buffer 601 can be continuously updated. For example, if acceleration data is sampled at 100 Hz, each time acceleration data is measured, the acceleration buffer 601 may store the latest acceleration data, remove the oldest acceleration data, and continue updating.

7 is a flowchart 700 illustrating a method of recognizing a wheelchair pushing motion of a wearable device according to various embodiments.

Referring to FIG. 7 , in operation 701, a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments generates a first acceleration buffer (eg, the wearable device 101 of FIG. 1 ). : It may be determined whether acceleration data is stored as much as the size of the acceleration buffer 603 in FIG. 6 . In order to use the probability distribution comparison method using the KLD algorithm, acceleration data measured in real time as much as the magnitude of the reference signal may be required. The processor 120 may determine the size of the first acceleration buffer by the size of the reference signal. For example, if the size of the reference signal is 100, the size of the first acceleration buffer may also be set to 100, and if the size of the reference signal is 120, the size of the first acceleration buffer may also be set to 120. When acceleration data is not stored as much as the size of the first acceleration buffer, a probability distribution comparison with a reference signal may not be possible. The processor 120 may perform operation 703 when acceleration data equal to the size of the first acceleration buffer is stored, and return to operation 501 of FIG. 5 when acceleration data is not stored equal to the size of the first acceleration buffer. .

When returning to operation 501, the processor 120 may repeatedly perform an operation of measuring acceleration data and storing the measured acceleration data in the first acceleration buffer. The processor 120 may repeatedly perform an operation of measuring acceleration data and storing the measured acceleration data in the first acceleration buffer until acceleration data equal to the size of the first acceleration buffer is stored.

When acceleration data equal to the size of the first acceleration buffer is stored, in operation 703, the processor 120 may calculate KLD and identify the acceleration data. The processor 120 uses the KLD algorithm to determine the probability distribution of the acceleration data (eg, from the first acceleration data to the 120th acceleration data) stored in the first acceleration buffer and the reference signal stored in the memory (eg, the memory 110 of FIG. 1). KLD can be calculated by comparing the probability distributions of (120 reference signals). The calculated comparison value may be a KLD output value. Also, the processor 120 may identify acceleration data last stored in the first acceleration buffer. Hereinafter, operations 705 to 709 may be performed simultaneously with operations 706 to 710 .

According to various embodiments, the processor 120 may recognize the wheelchair pushing motion using the KLD output value, but may further use the KLD average value or acceleration variance value to increase the accuracy of the wheelchair pushing motion. To this end, the processor 120 may further perform the operation of FIG. 7 . First, the operation of calculating the KLD average value will be described. The present invention is not limited by the description.

At operation 705, the processor 120 may store the KLD output value in a KLD buffer. The KLD output value may be calculated and stored in the KLD buffer whenever the first acceleration buffer is updated. The size of the KLD buffer may be the same as or different from the size of the first acceleration buffer. For example, the size of the KLD buffer may be larger than the size of the first acceleration buffer. When the size of the first acceleration buffer is 100, the size of the KLD buffer may be 120. Alternatively, when the size of the first acceleration buffer is 120, the size of the KLD buffer may be 150.

In operation 707, the processor 120 may determine whether as many KLD output values as the size of the KLD buffer are stored. When the size of the first acceleration buffer is 120 and the 120th acceleration data is stored in the first acceleration buffer, the processor 120 may perform operation 703 to obtain a first KLD output value. The processor 120 may store the obtained first KLD output value in the KLD buffer, and determine whether as many KLD output values are stored as the size of the KLD buffer.

The processor 120 may perform operation 709 when the KLD output value is stored as much as the size of the KLD buffer, and perform operation 721 when the KLD output value is not stored as much as the size of the KLD buffer.

When the KLD output value is not stored as much as the size of the KLD buffer, in operation 721, the processor 120 measures new acceleration data (eg, 121st acceleration data) and stores the measured acceleration data in the first acceleration buffer. can If operation 721 is performed, the processor 120 returns to operation 703 and compares the probability distribution of the acceleration data (eg, the second acceleration data to the 121st acceleration data) stored in the first acceleration buffer with the probability distribution of the reference signal. A second KLD output value can be obtained. After performing step 703, the processor 120 may perform step 705 to store the obtained second KLD output value in the KLD buffer, and perform step 707 to determine whether KLD output values corresponding to the size of the KLD buffer are stored. . The processor 120 may repeatedly perform

operations

703, 705, 707, and 721 until KLD output values corresponding to the size of the KLD buffer are stored.

When KLD output values equal to the size of the KLD buffer are stored, in operation 709, the processor 120 may calculate a KLD average value based on the KLD output values. When the size of the KLD buffer is 150, the processor 120 may calculate an average value of 150 KLD output values. The processor 120 may perform operation 711 after calculating the KLD average value.

An operation of calculating an acceleration variance value will be described.

In operation 706, the processor 120 may store the identified acceleration data in a second acceleration buffer. Similar to the first acceleration buffer, acceleration data may be stored in the second acceleration buffer. The difference is that after storing as much as the size of the first acceleration buffer, the last acceleration data stored in the first acceleration buffer may be stored in the second acceleration buffer. When acceleration data is stored as much as the size of the first acceleration buffer (eg, 120), the 120th acceleration data may be the last acceleration data stored in the first acceleration buffer. The processor 120 may identify the last acceleration data stored in the first acceleration buffer in operation 703 and store the identified acceleration data in the second acceleration buffer.

In operation 708, the processor 120 may determine whether acceleration data is stored as much as the size of the second acceleration buffer. The size of the second acceleration buffer may be the same as or different from the size of the first acceleration buffer or the size of the KLD buffer. For example, the size of the second acceleration buffer may be greater than the size of the first acceleration buffer and the same as the size of the KLD buffer. When the size of the first acceleration buffer is 100, the size of the second acceleration buffer may be 120. Alternatively, when the size of the first acceleration buffer is 120, the size of the second acceleration buffer may be 150.

The processor 120 may perform operation 710 when acceleration data equal to the size of the second acceleration buffer is stored, and perform operation 722 when acceleration data is not stored equal to the size of the second acceleration buffer. there is.

When acceleration data is not stored as much as the size of the second acceleration buffer, in operation 722, the processor 120 measures new acceleration data (eg, 121st acceleration data) and converts the measured acceleration data to the first acceleration data. can be stored in a buffer. Operation 722 may be the same as or similar to operation 721 . After performing operation 722, the processor 120 may return to operation 703 to identify the last acceleration data (eg, 121st acceleration data) stored in the first acceleration buffer. After performing step 703, the processor 120 performs operation 706 to store the identified acceleration data in the second acceleration buffer, and performs operation 708 to determine whether acceleration data corresponding to the size of the second acceleration buffer is stored. can judge The processor 120 may repeatedly perform

operations

703, 706, 708, and 722 until acceleration data corresponding to the size of the second acceleration buffer is stored.

When acceleration data corresponding to the size of the second acceleration buffer is stored, in operation 710, the processor 120 may calculate an acceleration variance value based on the acceleration data. When the size of the second acceleration buffer is 150, the processor 120 may calculate variance values of 150 acceleration data. The processor 120 may perform operation 711 after calculating the acceleration dispersion value.

In operation 711, the processor 120 may determine whether or not a set condition is met. The set condition may include at least one of a KLD output value less than a KLD reference value, a KLD average value less than a KLD average threshold value, and an acceleration variance value exceeding a variance threshold value. When the KLD output value obtained in operation 507 of FIG. 5 is less than the KLD reference value, the processor 120 may determine that the set condition is satisfied. Alternatively, the processor 120 determines the set conditions when the KLD output value is less than the KLD reference value, the KLD average value calculated in operation 709 is less than the KLD average threshold value, and the acceleration variance value calculated in operation 710 exceeds the variance threshold value. can be judged to be relevant.

The processor 120 may perform operation 713 when the set conditions are met, and perform operation 715 when the set conditions are not met.

In operation 713, if the set conditions are met, the processor 120 may recognize (or determine) the wheelchair pushing motion. When the processor 120 determines that the pushing motion is the wheelchair, the processor 120 may repeatedly perform operations 701 to 711 to determine whether the pushing motion of the wheelchair is continuously performed. Also, if the processor 120 recognizes the wheelchair pushing motion, the processor 120 may perform the operation of FIG. 10 to count the number of times of the wheelchair pushing motion.

If the set conditions are not met, in operation 715, the processor 120 may determine that the pushing motion is not the wheelchair. The processor 120 determines whether the KLD output value is greater than or equal to the KLD reference value, the KLD average value calculated in operation 709 is greater than or equal to the KLD average threshold value, or the acceleration variance value calculated in operation 710 is less than or equal to the variance threshold. It may be determined that the set conditions are not met. When the KLD output value, the KLD average value, or the acceleration variance value does not correspond to any of the set conditions, the processor 120 may determine that the wheelchair pushing motion is not. The processor 120 may determine that the motion other than the wheelchair pushing motion is an atypical motion such as walking or running. The processor 120 may perform an algorithm for acquiring exercise information (eg, walking time, number of steps, running time, etc.) according to the atypical motion. Since operations such as calculating the number of steps, calculating the walking time, or calculating the running time correspond to the prior art, a detailed description thereof may be omitted.

Referring to FIG. 8A , a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments includes an acceleration sensor (eg, the inertial sensor of FIG. 1 ). (150) may be used to measure acceleration data (or acceleration value or acceleration sensing value). The processor 120 may store acceleration data measured in real time according to a time sequence (eg, the time index 601) in a first acceleration buffer (eg, the acceleration buffer 603 of FIG. 6). The size of the first acceleration buffer 603 may be set equal to the size of the reference signal stored in the memory of the wearable device 101 (eg, the memory 110 of FIG. 1 ). Hereinafter, the size of the first acceleration buffer 603 is described as '120', but the present invention is not limited by the description. For example, in the time index 601, the first acceleration data measured at the first time (1) is stored in the first acceleration buffer 603, and the second acceleration data measured at the second time (2) is stored in the first acceleration buffer ( 601) can be stored. The processor 120 may continuously measure and store acceleration data in the first acceleration buffer 603 until the acceleration data is stored as much as the size of the first acceleration buffer 603 (eg, 120).

When acceleration data corresponding to the size of the first acceleration buffer 603 is stored, the processor 120 converts the probability distribution of the acceleration data stored in the first acceleration buffer 603 (eg, the first acceleration data to the 120th acceleration data) into a reference signal. The first KLD output value 811 can be obtained by comparing with the probability distribution of . The processor 120 may determine whether it is a wheelchair pushing motion based on the first KLD output value 811 . Alternatively, the processor 120 may further use the KLD average value or acceleration variance value in order to more accurately determine whether the motion is a wheelchair pushing motion. To this end, when the processor 120 obtains the first KLD output value 811, the first KLD output value 811 is stored in the KLD buffer 801, and the last acceleration data stored in the first acceleration buffer 603 (eg, 120 th acceleration data) may be stored in the second acceleration buffer 803 . The size of the KLD buffer 801 may be the same as or different from the size of the first acceleration buffer 603 or the second acceleration buffer 803 . For example, the size of the KLD buffer 801 or the size of the second acceleration buffer 803 is 150, which is the same and may be different from that of the first acceleration buffer 603. The present invention is not limited by the description.

Thereafter, the processor 120 stores the newly measured acceleration data (eg, 121st acceleration data) in the first acceleration buffer 603, and stores the acceleration data (eg, second acceleration data) stored in the first acceleration buffer 603. The second KLD output value may be obtained by comparing the probability distribution of the 121st acceleration data) with the probability distribution of the reference signal. When the processor 120 obtains the second KLD output value, the second KLD output value is stored in the KLD buffer 801, and the last acceleration data (eg, the 121st acceleration data) stored in the first acceleration buffer 603 is stored in the second acceleration buffer. (803).

Referring to FIG. 8B , the processor 120 stores the 269th acceleration data measured at the time when the time index 601 is the 269th in the first acceleration buffer 603, and stores the acceleration data stored in the first acceleration buffer 603. The 150th KLD output value 817 may be obtained by comparing the probability distribution of the data (eg, the 150th acceleration data to the 269th acceleration data) with the probability distribution of the reference signal. When the processor 120 obtains the 150th KLD output value 817, it stores the 150th KLD output value 817 in the KLD buffer 801, and stores the last acceleration data stored in the first acceleration buffer 603 (eg, the 269th KLD output value 817). acceleration data) may be stored in the second acceleration buffer 803 . When the KLD output value is stored as much as the size of the KLD buffer 801 (eg, 150), the processor 120 may calculate the first KLD average value 813 based on the KLD output value stored in the KLD buffer 801. In addition, when acceleration data (eg, 120th to 269th acceleration data) is stored as much as the size (eg, 150) of the second acceleration buffer 803, the processor 120 stores the data stored in the second acceleration buffer 803. A first acceleration variance value 815 may be calculated based on the acceleration data.

The processor 120 may perform motion classification 830 based on at least one of the 150th KLD output value 817 , the first KLD average value 813 , and the first acceleration variance value 815 . The motion classification 830 may determine (or classify) a typical motion or an irregular motion. The standard motion may include a wheelchair pushing motion, and the atypical motion may include all motions (eg, walking, number of steps) except for the wheelchair pushing motion. For example, if the 150th KLD output value 817 corresponds to less than the KLD reference value, the processor 120 may determine the wheelchair pushing motion in the motion classification 830. When the 150th KLD output value 817 is equal to or greater than the KLD reference value, the processor 120 may determine that the wheelchair pushing motion is not the motion classification 830 .

Alternatively, the processor 120 determines that the 150th KLD output value 817 is less than the KLD reference value, the first KLD average value 813 is less than the KLD average threshold value, and the first acceleration variance value 815 exceeds the variance threshold value, In the motion classification 830, it may be determined as a wheelchair pushing motion. The processor 120 determines at least one of the fact that the 150th KLD output value 817 is greater than or equal to the KLD reference value, the first KLD average value 813 is greater than or equal to the KLD average threshold value, or the first acceleration variance value 815 is less than or equal to the variance threshold value. If applicable, it may be determined that the motion classification 830 is not a wheelchair pushing motion.

Referring to FIG. 8C , the processor 120 stores the 270th acceleration data measured at the time when the time index 601 is the 270th in the first acceleration buffer 603, and stores the acceleration data stored in the first acceleration buffer 603. The 151st KLD output value 831 may be obtained by comparing the probability distribution of the data (eg, the 151st acceleration data to the 270th acceleration data) with the probability distribution of the reference signal. When the processor 120 obtains the 151st KLD output value 831, it stores the 151st KLD output value 831 in the KLD buffer 801, and stores the last acceleration data stored in the first acceleration buffer 603 (eg, the 270th KLD output value 831). acceleration data) may be stored in the second acceleration buffer 803 . When the KLD output value is stored as much as the size of the KLD buffer 801 (eg, 150), the processor 120 may calculate the second KLD average value 833 based on the KLD output value stored in the KLD buffer 801. In addition, when acceleration data is stored as much as the size of the second acceleration buffer 803 (eg, 150), the processor 120 stores the acceleration data stored in the second acceleration buffer 803 (eg, the 121st acceleration data to the 270th acceleration data). Data), the second acceleration variance value 835 may be calculated.

The processor 120 may perform motion classification 850 based on at least one of the 151st KLD output value 831 , the second KLD average value 833 , and the second acceleration variance value 835 . For example, if the 151st KLD output value 831 corresponds to less than the KLD reference value, the processor 120 may determine the wheelchair pushing motion in the motion classification 850. When the 151st KLD output value 831 is equal to or greater than the KLD reference value, the processor 120 may determine that the wheelchair pushing motion is not the motion classification 850 .

Alternatively, the processor 120 determines that the 151st KLD output value 831 is less than the KLD reference value, the second KLD average value 833 is less than the KLD average threshold value, and the second acceleration variance value 835 exceeds the variance threshold value, In the motion classification 850, it may be determined as a wheelchair pushing motion. The processor 120 determines at least one of the fact that the 151st KLD output value 831 is greater than or equal to the KLD reference value, the second KLD average value 833 is greater than or equal to the KLD average threshold value, or the second acceleration variance value 835 is less than or equal to the variance threshold value. If applicable, it may be determined that the motion classification 850 is not a wheelchair pushing motion.

Referring to FIG. 9 , since the probability distribution of acceleration data measured during a wheelchair pushing motion is not general and has a nonlinear distribution, it may not be easy to select a probability density function model. According to various embodiments, a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) estimates a probability density function using acceleration data. method can be used. In the present invention, among non-parametric density estimation methods, a kernel function-based density estimation method using a Gaussian function as a kernel function may be used. The first graph 910 may represent acceleration data 911 appearing during a wheelchair pushing motion and acceleration data 913 appearing when not a wheelchair pushing motion (eg, atypical motion).

The second graph 930 may indicate a kernel function generation result when a wheelchair pushing motion (eg, a orthopedic motion) is performed. The first signal 933 represents a kernel function estimation result of the reference signal, and the second signal 931 may mean acceleration data actually measured during a wheelchair pushing motion. The reference signal may refer to acceleration data generated when a wheelchair pushing motion (eg, ①②③④ in FIG. 4) is performed once. Comparing the first signal 933 and the second signal 931, it can be seen that they have similar characteristics.

The third graph 950 may indicate a kernel function generation result when the wheelchair pushing motion is not (eg, atypical motion). A third signal 953 indicates a result of estimating the kernel function of the reference signal, and a fourth signal 951 may mean acceleration data actually measured when the wheelchair is not pushing. Comparing the third signal 953 and the fourth signal 951, it can be seen that they do not have similar characteristics.

Therefore, in the case of a wheelchair pushing motion, by comparing the probability distribution of a predefined reference signal with the probability distribution of actually measured acceleration data using the KLD algorithm, it is possible to efficiently recognize the wheelchair pushing motion according to the similarity of the two probability distributions. there is.

10 is a flowchart 1000 illustrating a method of counting a wheelchair pushing motion of a wearable device according to various embodiments. 10 may be performed after being recognized as a wheelchair pushing motion by FIG. 5 or 7 .

Referring to FIG. 10 , in operation 1001, a processor (eg, processor 120 of FIG. 1 ) of a wearable device (eg, wearable device 101 of FIG. 1 ) according to various embodiments sets a second acceleration buffer (eg, processor 120 of FIG. 1 ). : New acceleration data may be stored in the second acceleration buffer 803 of FIGS. 8A to 8C. To recognize the wheelchair pushing motion, the processor 120 may determine whether the KLD output value is less than the KLD reference value, the KLD average value is less than the KLD average threshold value, and the acceleration variance value exceeds the variance threshold value. In order to calculate the acceleration dispersion value, acceleration data corresponding to the size of the second acceleration buffer 803 may be stored. Then, when a new acceleration is measured by an acceleration sensor (eg, the inertial sensor 150 of FIG. 1 ), the processor 120 may store new acceleration data in the second acceleration buffer 803 . For example, when the size of the second acceleration buffer 803 is 150, the 270th acceleration data may be stored as new acceleration data in the second acceleration buffer 803.

In operation 1003, the processor 120 may set a window for the set number of acceleration data. For example, the processor 120 may set one window to five acceleration data and assign (or allocate) a window index. Hereinafter, it is described that windows are set for 5 acceleration data, but windows may be set for more than 5 acceleration data (eg, 7, 9, 11). Examples are only for helping understanding of the invention, and the present invention is not limited by the examples. The second acceleration buffer 803 may store 121st to 270th acceleration data. The processor 120 may set the first window index to the 121st to 125th acceleration data. When updating the window, the processor 120 may set the second window index to the 122nd to 126th acceleration data. The processor 120 may sequentially set windows for acceleration data stored in the second acceleration buffer 803 in a window sliding manner.

At operation 1005, the processor 120 may detect an acceleration peak within the window. The processor 120 defines the five acceleration data included in the window as d(1), d(2), d(3), d(4), and d(5), respectively, through the following determination formula. Acceleration peaks can be determined.

Condition = {d(1) > d(3) ∧ d(2) > d(3) ∧ d(4) > d(3) ∧ d(5) > d(3)}

When the Condition value is 1, the processor 120 may determine that an acceleration peak within the window is detected, and when the Condition value is 0, it may be determined that an acceleration peak within the window is not detected. The processor 120 may detect an acceleration peak based on third acceleration data (eg, d(3)), which is an intermediate value within the window. The above determination method is a method of detecting the minimum value of acceleration data as a peak, and this may be because the axis of the inertial sensor 150 used when determining a wheelchair pushing motion (e.g., ①②③④ in FIG. 4) is set to the x-axis. When the x-axis of the inertial sensor 150 performs an operation (e.g., ③) to return to the wheelchair after the wheelchair pushing operation, the direction of the user's hand changes to the opposite direction of the pushing, and a large acceleration is generated to more easily detect the peak. this could be possible

According to various embodiments, the axis of the inertial sensor 150 coincident with the direction of the user's fingertip may vary depending on the location where the inertial sensor 150 is disposed on the wearable device 101 . In the present invention, the axis of the acceleration sensor corresponding to the direction of the user's fingertip is set as the 'x-axis', and the pushing motion of the wheelchair can be recognized based on the acceleration data of the x-axis. However, the axis of the acceleration sensor used for pushing the wheelchair may be the y-axis or the z-axis according to the position where the acceleration sensor is disposed in the wearable device 101 . Also, the processor 120 may detect an acceleration peak using a maximum value of acceleration data.

In operation 1005, when an acceleration peak within the window is detected (Condition value is 1), the processor 120 may determine that an acceleration peak is detected and perform operation 1007. In addition, when no acceleration peak is detected within the window (Condition value is 0), the processor 120 may determine that no acceleration peak is detected and perform operation 1006 .

If it is determined that the acceleration peak has not been detected, in operation 1006, the processor 120 may add a window index. If the window index is 1 (eg, the first window index) when operation 1003 is first performed, the window index may become 2 (eg, the second window index) when operation 1006 is performed. Processor 120 may perform operation 1006 and return to operation 1003 . When returning to operation 1003, the processor 120 may update the window and set a window corresponding to the added window index to the set number of acceleration data. For example, the processor 120 may set the second window index to the 122nd to 126th acceleration data.

According to various embodiments, if the window index is 2 (eg, the second window index) when operation 1003 is performed a second time, the window index may become 3 (eg, the third window index) when operation 1006 is performed. . Processor 120 may perform operation 1006 and return to operation 1003 . Returning to operation 1003, the processor 120 may set a window for the set number of acceleration data. For example, the processor 120 may set the third window index to the 123rd to 127th acceleration data.

When it is determined that an acceleration peak has been detected, in operation 1007, the processor 120 may return (or identify) acceleration data detected as a peak. For example, the processor 120 may return the acceleration data having the minimum value among the 121st to 125th acceleration data.

In operation 1009 , the processor 120 may detect a minimum value of acceleration data stored in the second acceleration buffer 803 . The detection of the minimum value of the acceleration data may be because acceleration data having the minimum value is detected as an acceleration peak when the acceleration peak is detected in operation 1005 . When the acceleration data having the maximum value is detected as the acceleration peak upon detection of the acceleration peak, the processor 120 may detect the maximum value of the acceleration data stored in the second acceleration buffer 803 . Hereinafter, in order to help understanding of the present invention, it is described that the minimum value of acceleration data is detected, but the present invention is not limited by the description.

While performing a wheelchair pushing motion (eg, ①②③④ in FIG. 4 ), a number of acceleration peaks may be detected due to the user's hand motion and the noise of the inertial sensor 150 . Among the wheelchair pushing motions, the situation that receives the greatest acceleration may be the part where the direction of the hands changes for the motion to return to the wheelchair pushing motion (e.g., ③) after the hand pushes off the handle of the wheelchair. In order to consider only the largest peak, only peaks lower than a threshold may be considered. At this time, fixing the threshold to a constant may cause a peak detection error (or error) because the magnitude of the acceleration data varies according to the strength of the wheelchair pushing motion. The processor 120 may use an adaptive threshold to reduce a peak detection error occurring when the user performs a wheelchair pushing motion.

In operation 1011, the processor 120 may select (or determine) an adaptive threshold based on the minimum value of the acceleration data. The processor 120 may determine a set reference value (eg, acceleration data corresponding to 10% of the minimum value) of the minimum value of acceleration data stored in the second acceleration buffer 803 as an adaptive threshold. When the acceleration peak is detected, the processor 120 may set an adaptive threshold based on the minimum value of the acceleration data because it detects the acceleration data having the minimum value as the acceleration peak. When the processor 120 detects the acceleration data having the maximum value as the acceleration peak upon detection of the acceleration peak, the set reference value of the maximum value of the acceleration data stored in the second acceleration buffer 803 (e.g., corresponds to 90% of the maximum value). acceleration data) may be determined as an adaptive threshold.

Whenever new acceleration data is measured, the new acceleration data is stored in the second acceleration buffer 803, so that the adaptive threshold can be continuously changed. When the acceleration data stored in the second acceleration buffer 803 is changed, the adaptive threshold may be dynamically changed because the adaptive threshold is determined based on the acceleration data stored in the second acceleration buffer 803. Alternatively, the processor 120 may determine the adaptive threshold whenever a new window is updated. When an acceleration peak is detected in a new window, operations 1007 to 1011 are performed, so that the processor 120 can determine the adaptive threshold every time a new window is updated.

For example, a first adaptive threshold determined when

operations

1009 and 1011 are first performed may be the same as or different from a second adaptive threshold determined when

operations

1009 and 1011 are performed a second time.

In operation 1013, the processor 120 may determine whether the returned acceleration data is less than the adaptive threshold. Since the adaptive threshold is determined based on the “minimum value” of the acceleration data in

operations

1009 and 1011, the processor 120 may determine whether the returned acceleration data is less than the adaptive threshold. If the adaptive threshold is determined based on the “maximum value” of the acceleration data, the processor 120 may determine whether the returned acceleration data exceeds the adaptive threshold. The returned acceleration data may be obtained in operation 1007 and the adaptive threshold may be obtained in operation 1011 . The processor 120 may perform operation 1015 when the returned acceleration data is less than the adaptive threshold, and may perform operation 1019 when the returned acceleration data is greater than or equal to the adaptive threshold.

When the returned acceleration data is less than the adaptive threshold, in operation 1015, the processor 120 may determine whether a difference (or interval) between window indices exceeds an index threshold. The window index may be a window index given in operation 1003. It may take a certain amount of time to perform the wheelchair pushing motion (eg, ①②③④ in FIG. 4 ) once. Even if the wheelchair pushing motion is performed quickly or slowly, it may take a certain amount of time to perform one wheelchair pushing motion. The index threshold may be used to reduce count errors when counting wheelchair pushing motions.

If the interval (or difference) between the two window indices detected as an acceleration peak is too small, it may be difficult to assume that one wheelchair pushing motion has been performed. The processor 120 may count the wheelchair pushing motion only when the interval between the window index determined as the previous wheelchair pushing motion and the window index determined as the subsequent wheelchair pushing motion exceeds the index threshold. The index threshold may be previously stored in the memory of the wearable device 101 (eg, the memory 110 of FIG. 1 ). The index threshold may be set such that an interval between two window indices in which an acceleration peak is detected is determined as one wheelchair pushing motion.

For example, if the window index where the first acceleration peak is detected (eg, the previous window index) is too close to the window index where the second acceleration peak is detected (eg, the current window index), the processor 120 performs one wheelchair pushing motion. may not be recognized as a detected acceleration peak. When the difference (or interval) between the window index in which the first acceleration peak is detected and the window index in which the second acceleration peak is detected exceeds the index threshold, the processor 120 converts the wheel chair pushing motion to the detected acceleration peak. Recognizable.

The processor 120 performs operation 1017 when the difference between the two window indices detected as the acceleration peak exceeds the index threshold, and when the difference between the two window indices detected as the acceleration peak is less than or equal to the index threshold, the processor 120 performs operation 1017. 1019 can be done.

When the difference between the two window indices detected as the acceleration peak exceeds the index threshold, in operation 1017, the processor 120 may count the wheelchair pushing motion. The processor 120 may count the wheelchair pushing motion once every time operations 1001 to 1017 are performed. The processor 120 may store the counted number in the memory 110 and display the counted number on a display (eg, the display module 140 of FIG. 1 ). The processor 120 may calculate the wheelchair pushing time based on the number of times or time in which the wheelchair pushing motion is counted. The processor 120 may provide the number of times of pushing the wheelchair or the time of pushing the wheelchair as activity information (or exercise information). The processor 120 may analyze and provide a movement pattern of the user based on the wheelchair pushing motion. According to various embodiments, the processor 120 may provide a wheelchair mode, such as a walking mode or a running mode, as an exercise mode. When the wearable device 101 is set to the wheelchair mode, the processor 120 performs the operation of FIG. 5 or 7 to recognize the wheelchair pushing motion, and when recognized as a wheelchair pushing motion, the processor 120 performs the motion of FIG. 10 to push the wheelchair The number of operations can be counted.

According to various embodiments, the processor 120 may calculate calorie consumption based on the number of wheelchair pushing motions and provide the calculated calorie consumption to the user. For example, the processor 120 may calculate calories consumed during the day by adding a basal metabolic rate or additional calories each time the number of wheelchair pushing motions is counted. The processor 120 may calculate the calorie consumption based on the speed of the wheelchair pushing motion. The processor 120 calculates additional calories consumed when the speed of the wheelchair pushing motion is fast (eg, exceeds a speed threshold), and calculates additional calories consumed when the speed of the wheelchair pushing motion is slow (eg, below the speed threshold). may not be calculated. The processor 120 may add the additional calories burned to the calories burned. Alternatively, the processor 120 may recognize an active wheelchair pushing motion when the speed of the wheelchair pushing motion is fast, and recognize a normal wheelchair pushing motion when the speed of the wheelchair pushing motion is slow.

According to various embodiments, the processor 120 may output a guide to exercise when the wheelchair pushing motion is not recognized for a set period of time (eg, 3 days, 7 days, etc.). The inertial sensor 150 can measure acceleration data 24 hours a day with low power. The processor 120 may determine whether the wheelchair pushing motion is recognized at set intervals (eg, 1 day, 3 days) based on the measured acceleration data. The guide may include at least one of text, image, video, or voice. The guide may display a user interface on the display module 140 or output a voice through a speaker (eg, the speaker 175 of FIG. 1 ). The processor 120 may provide a user interface for setting a target value for the number of pushing the wheelchair. The user may set the number of times of pushing the wheelchair (eg, 100 times or 200 times) through the user interface similarly to the target number of steps.

The processor 120 may count the wheelchair pushing motion once in operation 1017 and then return to operation 1001 . The processor 120 may count the wheelchair pushing motion by repeatedly performing operations 1001 to 1017 .

If the returned acceleration data is greater than or equal to the adaptive threshold, or if the difference between two window indices detected as acceleration peaks is less than or equal to the index threshold, the processor 120 may not count the wheelchair pushing motion in operation 1019. The processor 120 may return to operation 1001 without counting the wheelchair pushing motion. The processor 120 may perform the operation of FIG. 5 or 7 while performing FIG. 10 to recognize the wheelchair pushing motion. If the operation of FIG. 5 or 7 is not recognized as a wheelchair pushing motion, the processor 120 may not perform the operation of FIG. 10 . The processor 120 may perform the operation of FIG. 10 when the operation of FIG. 5 or 7 is recognized as a wheelchair pushing motion. According to various embodiments, when the operation 1019 is performed a set number of times (eg, 10 or 20 times), the processor 120 may stop the operation of FIG. 10 and perform the operation of FIG. 5 or 7. can

Referring to FIG. 11 , a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments generates a second acceleration buffer (eg, FIGS. 8A to 100 ). A window may be set for acceleration data stored in the second acceleration buffer 803 of FIG. 8C. The window setting 1110 may be set in a window sliding manner. The size of the second acceleration buffer 803 is 150, and 150 pieces of acceleration data can be stored in the second acceleration buffer 803. Referring to the window setting 1110, the processor 120 sets the first to fifth acceleration data stored in the second acceleration buffer 803 as the first window index 1111, and the second acceleration data to the fifth acceleration data. Up to 6 acceleration data is set as the 2nd window index 1112, from the 3rd acceleration data to the 7th acceleration data is set as the 3rd window index 1113, from the 4th acceleration data to the 8th acceleration data is set as the 4th window index (1114) can be set.

The processor 120 may set the first window index 1111 and perform operation 1005 of FIG. 10 . When an acceleration peak is not detected within the first window index 1111 by performing operation 1005, in operation 1006, the processor 120 may set the second window index 1112 in operation 1003 by adding one window index. there is. In addition, when the acceleration peak is not detected within the second window index 1112 by performing operation 1005, in operation 1006, the processor 120 adds one window index to obtain the third window index 1113 in operation 1003. can be set

Referring to the graph 1130, the first window index 1111 includes first acceleration data 1131, second acceleration data 1133, third acceleration data 1135, fourth acceleration data 1137, and fifth acceleration data 1137. Acceleration data 1139 may be included. The processor 120 determines whether the third acceleration data 1135 is less than the first acceleration data 1131 or the third acceleration data 1135 is less than the second acceleration data 1133 according to the determination method described with reference to FIG. 10 . , It may be determined whether the third acceleration data 1135 is less than the fourth acceleration data 1137 or whether the third acceleration data 1135 is less than the fifth acceleration data 1139. When the third acceleration data 1135 is less than the first acceleration data 1131, the second acceleration data 1133, the fourth acceleration data 1137, and the fifth acceleration data 1139 (eg, Condition If the value is 1), it can be determined that an acceleration peak within the first window index 1111 is detected.

When the third acceleration data 1135 is greater than or equal to the first acceleration data 1131, the second acceleration data 1133, the fourth acceleration data 1137, or the fifth acceleration data 1139 (eg, Condition If the value is 0), it may be determined that the acceleration peak within the first window index 1111 is not detected. In the graph 1130, an example in which an acceleration peak in the first window index 1111 is not detected may be shown.

The processor 120 may set a window for the second window index 1112, the third window index 1113, or the fourth window index 1114 in the same way, and detect an acceleration peak within the window.

Referring to FIG. 12 , a first graph 1210 and a second graph 1230 show a second acceleration buffer (eg, the second acceleration buffer of FIGS. 8A to 8C ) of a wearable device (eg, the wearable device 101 of FIG. 1 ). An example in which an adaptive threshold is determined based on a minimum value of acceleration data stored in the acceleration buffer 803 may be shown. During the wheelchair pushing motion (e.g., ①②③④ in FIG. 4), a number of acceleration peaks may be detected due to the user's hand movement type and the noise of the acceleration sensor (e.g., the inertial sensor 150 in FIG. 1). Among the wheelchair pushing motions, the situation that receives the greatest acceleration may be the part where the direction of the hands changes for the motion to return to the wheelchair pushing motion (e.g., ③) after the hand pushes off the handle of the wheelchair. In order to consider only the largest peak, only peaks smaller than the threshold can be considered. If the threshold is fixed at one value, an acceleration peak detection error may occur because the magnitude of acceleration data varies depending on the intensity of the wheelchair pushing motion. A processor of the wearable device 101 (eg, the processor 120 of FIG. 1 ) may use an adaptive threshold to reduce a peak detection error generated when a user performs a wheelchair pushing motion.

In the first graph 1210, the first signal 1211 means acceleration data stored in the second acceleration buffer 803, and the second signal 1213 is based on the acceleration data stored in the second acceleration buffer 803. It shows the adaptive threshold determined by The second graph 1230 may be an enlarged view of a certain section of the first graph 1210 . In the second graph 1230, the third signal 1231 means acceleration data stored in the second acceleration buffer 803, and the fourth signal 1233 is based on the acceleration data stored in the second acceleration buffer 803. It shows the adaptive threshold determined by

Looking at the second signal 1231 and the fourth signal 1233 of the first graph 1210 and the second graph 1230, it can be seen that the peak (eg, upper end) of the adaptive threshold is continuously changed. Whenever new acceleration data is measured, the new acceleration data is stored in the second acceleration buffer 803, so that the adaptive threshold can be continuously changed. When the acceleration data stored in the second acceleration buffer 803 is changed, the adaptive threshold may be dynamically changed because the adaptive threshold is determined based on the acceleration data stored in the second acceleration buffer 803.

Referring to FIG. 13, a first graph 1310 and a second graph 1330 are provided to prevent a count error of a wheelchair pushing motion based on whether a difference between two window indices detected as an acceleration peak exceeds an index threshold. An example may be shown. Referring to the first graph 1310, a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments generates a first acceleration peak detected. After 1 window index 1311, when the second window index 1313 where the acceleration peak is detected is detected, the difference (or interval) between the first window index 1311 and the second window index 1313 is the index threshold. It can be judged whether it exceeds . If the interval between the two window indices detected as the acceleration peak is too small, it may be difficult for the processor 120 to see that a single wheelchair pushing motion has been performed.

The index threshold may be previously stored in the memory of the wearable device 101 (eg, the memory 110 of FIG. 1 ). The index threshold may be set such that an interval between two window indices in which an acceleration peak is detected is determined as one wheelchair pushing motion. For example, if the difference between the first window index 1311 and the second window index 1313 is less than or equal to an index threshold, the processor 120 determines that the wheelchair pushing motion is detected as an error, and does not count it as the number of times of the wheelchair pushing motion. may not be The processor 120 determines the difference between the first window index 1311 and the third window index 1315 when the third window index 1315 in which the acceleration peak is detected is detected after the second window index 1313 is detected. It may be determined whether or not exceeds the index threshold. When the difference between the first window index 1311 and the third window index 1315 exceeds an index threshold, the processor 120 may count the number of wheelchair pushing motions (eg, increase the wheelchair pushing motion by one time). .

Referring to the second graph 1330, a third graph 1331 before applying the index threshold and a fourth graph 1335 after applying the index threshold may be included. Looking at the third graph 1331, a plurality of window indices (eg, the first window index 1353, the second window index 1353, the third window index 1355, the fourth window index 1356, etc.) can be detected. If it is determined whether the difference between two window indices consecutive in time exceeds the index threshold, it may be determined that the second window index 1353 and the fourth window index 1356 exceed the index threshold. As shown in the third graph 1331, the processor 120 may count the wheelchair pushing motion 4 or more times before applying the index threshold, but may count the wheelchair pushing motion 2 times when the index threshold is applied.

Referring to FIG. 14 , a processor (eg, processor 120 of FIG. 1 ) of a wearable device (eg, wearable device 101 of FIG. 1 ) according to various embodiments recognizes a wheelchair pushing motion, and The operation of counting may be repeatedly performed. At a first reference numeral 1400, a first graph 1401 represents signals according to acceleration data and motion classification when a wheelchair pushing motion is performed. The first signal 1410 may represent acceleration data measured by an acceleration sensor (eg, the inertial sensor 150 of FIG. 1 ) of the wearable device. The second signal 1413 indicates a signal according to motion classification, and may be a non-zero constant (eg, 30 in FIG. 14 ) when recognized as a wheelchair pushing motion, or 0 when not recognized as a wheelchair pushing motion. In the first graph 1401, a circle 1411 may mean that the wheelchair pushing motion is counted. In a first reference numeral 1400, a second graph 1403 represents acceleration data and signals according to motion classification when the wheelchair is not a pushing motion. The third signal 1431 is acceleration data measured by the inertial sensor 150, and the fourth signal 1433 represents a signal according to motion classification, and may be 0 when not recognized as a wheelchair pushing motion.

Referring to a second reference numeral 1450, a third graph 1405 represents signals according to acceleration data and motion classification when a wheelchair pushing motion is performed. The fifth signal 1460 is acceleration data measured by the inertial sensor 150, and the sixth signal 1463 represents a signal according to motion classification, which is a non-zero constant when recognized as a wheelchair pushing motion, and a wheelchair pushing motion. It can be 0 when it is not recognized as . In the third graph 1405, a circle 1461 may mean that the wheelchair pushing motion is counted. At second reference numeral 1450, a fourth graph 1407 shows signals according to acceleration data and motion classification when the wheelchair is not a pushing motion. The seventh signal 1471 is acceleration data measured by the inertial sensor 150, and the eighth signal 1473 indicates a signal according to motion classification, and may be 0 when not recognized as a wheelchair pushing motion.

Referring to FIG. 15 , a processor (eg, the processor 120 of FIG. 1 ) of a wearable device (eg, the wearable device 101 of FIG. 1 ) according to various embodiments counts a wheelchair pushing motion, and calculates the counted number of times. may be displayed on a display (eg, the display module 140 of FIG. 1). The first user interface 1510 may represent an example of providing the number of times of pushing a wheelchair or the time of pushing a wheelchair as activity information (or exercise information). The processor 120 may analyze and provide a movement pattern of the user based on the wheelchair pushing motion. According to various embodiments, the processor 120 may provide a wheelchair mode, such as a walking mode or a running mode, as an exercise mode. When the wearable device 101 is set to the wheelchair mode, the processor 120 performs the operation of FIG. 5 or 7 to recognize the wheelchair pushing motion, and when recognized as a wheelchair pushing motion, the processor 120 performs the motion of FIG. 10 to push the wheelchair The number of operations can be counted.

According to various embodiments, the processor 120 may calculate calorie consumption based on the number of wheelchair pushing motions and provide the calculated calorie consumption to the user. The second user interface 1530 may include at least one of the number of times of pushing the wheelchair, the time of pushing the wheelchair, and calories burned. For example, the processor 120 may calculate calories consumed during the day by adding a basal metabolic rate or additional calories each time the number of wheelchair pushing motions is counted. The processor 120 may calculate the calorie consumption based on the speed of the wheelchair pushing motion.

According to various embodiments, the processor 120 may output a guide to exercise when the wheelchair pushing motion is not recognized for a set period of time (eg, 3 days, 7 days, etc.). The guide may include at least one of text, image, video, or voice. The guide may display a user interface on the display module 140 or output a voice through a speaker (eg, the speaker 175 of FIG. 1 ). The processor 120 may provide a user interface for setting a target value for the number of pushing the wheelchair. The user may set the number of times of pushing the wheelchair (eg, 100 times or 200 times) through the user interface similarly to the target number of steps.

An operating method of a wearable device (eg, the wearable device 101 of FIG. 1) according to various embodiments of the present invention uses an inertial sensor (eg, the inertial sensor 150 of FIG. 1) included in the wearable device. An operation of measuring acceleration data, an operation of storing the measured acceleration data in an acceleration buffer of a memory (eg, memory 110 of FIG. 1) of the wearable device, and determining whether the acceleration data is stored as much as the size of the acceleration buffer operation, when the acceleration data is stored as much as the size of the acceleration buffer, obtaining a KLD output value by comparing the probability distribution of the measured acceleration data with the probability distribution of the reference signal stored in the memory through the KLD algorithm; An operation of determining whether one KLD output value is less than the KLD reference value, and an operation of recognizing as a wheelchair pushing motion when the obtained KLD output value is less than the KLD reference value may be included.

The recognizing operation includes an operation of calculating a KLD average value based on the obtained KLD output value, an operation of calculating an acceleration variance value based on the measured acceleration data, the KLD output value is less than the KLD reference value, and the KLD average value is If the acceleration variance value is less than the KLD average threshold and corresponds to at least one of those exceeding the variance threshold, an operation of recognizing the wheelchair pushing operation may be included.

The operation of calculating the KLD average value includes the operation of storing the obtained KLD output value in the KLD buffer, the operation of determining whether the KLD output value is stored as much as the size of the KLD buffer, and the KLD output value as much as the size of the KLD buffer When stored, An operation of calculating an average value of KLD output values stored in the KLD buffer may be included.

In the case where the acceleration buffer is referred to as a first acceleration buffer and the acceleration variance value is calculated, the acceleration data stored in the first acceleration buffer is stored in a second acceleration buffer when the acceleration data is stored as much as the size of the first acceleration buffer. , an operation of determining whether acceleration data equal to the size of the second acceleration buffer is stored, and an operation of calculating an acceleration variance value of the acceleration data stored in the second acceleration buffer when acceleration data equal to the size of the second acceleration buffer is stored. can include

When the motion is recognized as pushing the wheelchair, the method stores acceleration data stored in the first acceleration buffer in a second acceleration buffer, and sets a window to a set number of acceleration data among the acceleration data stored in the second acceleration buffer. operation, detecting an acceleration peak within the window, determining whether acceleration data detected as an acceleration peak within the window is less than an adaptive threshold, and counting the number of times of the wheelchair pushing operation based on the determination result. It may further include an operation to do.

Determining whether the threshold is less than the adaptive threshold may include detecting a minimum value of acceleration data stored in the second acceleration buffer, determining the adaptive threshold based on the minimum value of the acceleration data, and determining an acceleration peak within the window. It may include an operation of determining whether the detected acceleration data is less than the adaptive threshold.

The counting operation may include, when the acceleration data detected as the acceleration peak within the window is less than the adaptive threshold, an operation of determining whether a difference between two window indices in which an acceleration peak is detected exceeds an index threshold, and an acceleration peak is detected and counting the number of times of the wheelchair pushing operation when the difference between the two window indices exceeds an index threshold.

According to one embodiment, the method according to various embodiments disclosed in this document may be included and provided in a computer program product. Computer program products may be traded between sellers and buyers as commodities. A computer program product is distributed in the form of a device-readable storage medium (eg compact disc read only memory (CD-ROM)), or through an application store (eg Play Store ^TM ) or on two user devices ( It can be distributed (eg downloaded or uploaded) online, directly between smart phones. In the case of online distribution, at least part of the computer program product may be temporarily stored or temporarily created in a device-readable storage medium such as a manufacturer's server, an application store server, or a relay server's memory.

According to various embodiments, each component (eg, module or program) of the above-described components may include a single object or a plurality of entities, and some of the plurality of entities may be separately disposed in other components. there is. According to various embodiments, one or more components or operations among the aforementioned corresponding components may be omitted, or one or more other components or operations may be added. Alternatively or additionally, a plurality of components (eg modules or programs) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the plurality of components identically or similarly to those performed by a corresponding component of the plurality of components prior to the integration. . According to various embodiments, the actions performed by a module, program, or other component are executed sequentially, in parallel, iteratively, or heuristically, or one or more of the actions are executed in a different order, or omitted. or one or more other actions may be added.

Various embodiments of the present invention disclosed in the present specification and drawings are only presented as specific examples to easily explain the technical content of the present invention and help understanding of the present invention, and are not intended to limit the scope of the present invention. Therefore, the scope of the present invention should be construed as including all changes or modifications derived based on the technical idea of the present invention in addition to the embodiments disclosed herein.

Claims

In the wearable device,

inertial sensor;

display;

a memory for storing a reference signal; and

a processor operatively coupled with at least one of the inertial sensor, the display, or the memory, the processor comprising:

Measuring acceleration data using the inertial sensor;

Store the measured acceleration data in an acceleration buffer of the memory;

Obtaining a KLD output value by comparing a probability distribution of the measured acceleration data with a probability distribution of a reference signal stored in the memory through a Kullback-Leivler divergence (KLD) algorithm;

Determine whether the obtained KLD output value is less than the KLD reference value,

When the obtained KLD output value is less than the KLD reference value, a wearable device configured to recognize a wheelchair pushing motion.
The method of claim 1, wherein the processor,

determining the size of the acceleration buffer by the size of the reference signal;

Determine whether the acceleration data is stored as much as the size of the acceleration buffer;

The wearable device configured to compare a probability distribution of the measured acceleration data with a probability distribution of a reference signal stored in the memory when the acceleration data is stored as much as the size of the acceleration buffer.
The method of claim 1, wherein the processor,

The wearable device configured to store acceleration data corresponding to an axis set according to a position of the inertial sensor disposed in the wearable device in the acceleration buffer.
The method of claim 1, wherein the processor,

Calculate a KLD average value based on the obtained KLD output value,

Calculate an acceleration dispersion value based on the measured acceleration data,

A wearable device configured to recognize the wheelchair pushing motion based on at least one of the KLD output value, the KLD average value, or the acceleration variance value.
The method of claim 4, wherein the processor,

Store the obtained KLD output value in a KLD buffer,

Determine whether the KLD output value is stored as much as the size of the KLD buffer,

A wearable device configured to calculate an average value of KLD output values stored in the KLD buffer when KLD output values are stored as much as the size of the KLD buffer.
According to claim 4,

The acceleration buffer is referred to as a first acceleration buffer,

the processor,

When the size of the first acceleration buffer is stored, the acceleration data stored in the first acceleration buffer is stored in a second acceleration buffer,

Determine whether acceleration data is stored as much as the size of the second acceleration buffer;

The wearable device configured to calculate an acceleration variance value of the acceleration data stored in the second acceleration buffer when acceleration data is stored as much as the size of the second acceleration buffer.
The method of claim 6, wherein the processor,

The wearable device configured to store the last acceleration data stored in the first acceleration buffer in the second acceleration buffer when the size of the first acceleration buffer is stored.
The method of claim 4, wherein the processor,

A wearable device set to recognize the wheelchair pushing motion when at least one of the KLD output value, the KLD average value, or the acceleration variance value corresponds to a set condition.
According to claim 8,

The set condition is set to include at least one of the KLD output value being less than the KLD reference value, the KLD average value being less than the KLD average threshold value, and the acceleration variance value exceeding the variance threshold value.
According to claim 1,

The acceleration buffer is referred to as a first acceleration buffer,

the processor,

When the motion is recognized as pushing the wheelchair, the acceleration data stored in the first acceleration buffer is stored in a second acceleration buffer;

Setting a window to the set number of acceleration data among the acceleration data stored in the second acceleration buffer;

detecting an acceleration peak within the window;

Determining whether acceleration data detected as an acceleration peak within the window is less than an adaptive threshold;

A wearable device configured to count the number of times of the wheelchair pushing motion based on the determination result.
The method of claim 10, wherein the processor,

When an acceleration peak within the window is detected, identifying acceleration data detected as the acceleration peak;

A wearable device configured to update a window by adding a window index when an acceleration peak within the window is not detected.
The method of claim 11, wherein the processor,

detecting a minimum value of acceleration data stored in the second acceleration buffer;

determine an adaptive threshold based on the minimum value of the acceleration data;

The wearable device configured to determine whether acceleration data detected as an acceleration peak within the window is less than an adaptive threshold.
The method of claim 12, wherein the processor,

The wearable device configured to change the adaptive threshold whenever new acceleration data is stored in the second acceleration buffer or the window is updated.
The method of claim 10, wherein the processor,

When acceleration data detected as an acceleration peak within the window is less than an adaptive threshold, determining whether a difference between two window indices in which an acceleration peak is detected exceeds an index threshold;

The wearable device configured to count the number of times of the pushing motion of the wheelchair when a difference between two window indices in which an acceleration peak is detected exceeds an index threshold.
In the operating method of the wearable device,

measuring acceleration data using an inertial sensor included in the wearable device;

storing the measured acceleration data in an acceleration buffer of a memory included in the wearable device;

determining whether the acceleration data is stored as much as the size of the acceleration buffer;

obtaining a KLD output value by comparing a probability distribution of the measured acceleration data with a probability distribution of a reference signal stored in the memory through a KLD algorithm when the acceleration data is stored as much as the size of the acceleration buffer;

determining whether the obtained KLD output value is less than a KLD reference value; and

And if the obtained KLD output value is less than the KLD reference value, recognizing it as a wheelchair pushing motion.