TWM524527U - A human active state detecting device - Google Patents

Abstract

A human active state detecting device comprises a motion identify device and a receiving device, wherein the motion identify device comprises a necklace and a detecting device fixing on the surface of the necklace and a connecting part. A receiving device receives human motion signal outing from the motion identify device, wherein the motion indentify device figures out the human motion signal and the receiving device recognizes the motion of user through the human motion signal. This utility model has the following features: 1. The motion identify device recognizes the motion of user precisely. 2. Recording every motion of user to figure out the ratio of each kind of human motion in a total detecting time.

Description

Human activity state recognition device

A human activity state recognition device, in particular, an instant human activity state recognition device that can be worn on a human body.

With the advancement of technology, notebook computers, tablet computers, and smart phones have changed our lives, but the ensuing civilized diseases have affected daily life. Modern people work hard, do not hesitate to fight for long-term work, maintain computers in the same sitting position for several hours; after busy, they will find that the muscles of the neck, waist and other parts of the body are maintained in the same position due to long periods of inactivity. pain. These pains may be caused by prolonged use of the same muscle group (for example, a long sitting posture that causes the waist muscles to tighten, or a long time to use the mouse to cause shoulder and neck muscle pain, etc.). The record of human posture can be used as an important reference data for judging human health and living habits. It can be used as a means of judging the daily habits of users by recording various postures in the daily life of the user, and using relevant information as an adjustment or improvement of bad posture. in accordance with. However, in the prior art, measuring the movement of the human body requires complicated instruments, so that the measurement process affects the daily life of the user, or the measured data cannot truly reveal the human body motion, and it is difficult to achieve simple and efficient operation. Detection.

In order to solve the problem that the human body motion cannot be easily and effectively monitored in the prior art, the present invention provides a motion recognition device, which includes a motion recognition device and a receiving device. The motion recognition device includes a collar and is respectively fixed on the collar surface. The monitoring body and a binding component, the receiving device receives a human body motion parameter output by the motion recognition device, wherein the monitoring body comprises a microprocessor and a sensing module, and the sensing module senses a human body After the operation, a sensing signal is outputted to the microprocessor, and the microprocessor calculates a human body motion parameter by the sensing signal; the receiving device performs a human body motion calculation according to the human motion parameter calculation, and the human motion calculation determines the use. And the action of the action is displayed on the receiving device; and the connector comprises two corresponding connectors, the contact faces of the two connectors are respectively provided with a positive terminal, a positive socket and each contact a pair of negative contact points on the surface, the positive terminal is correspondingly inserted into the positive socket, and the two positions correspond to the position of the negative contact Generating two electrical connection of the contacts of the connector when the face should be combined.

Wherein, the surface of the contact surface is magnetic, so that the connector is quickly combined to form an electrical connection.

The sensing module includes a gyroscope and a three-axis acceleration sensing module, and the gyroscope measures a geographic or azimuth angle and outputs the sensing result to the microprocessor, the three-axis acceleration sensing module. The group senses the moving speed and acceleration information and outputs to the microprocessor.

The monitoring body includes a wireless transmission module, and the wireless transmission module receives or transmits a wireless signal to the receiving device.

The receiving device is coupled to the motion recognition device, and after the receiving device is activated, a call signal is sent to search for the motion recognition device.

The receiving device verifies an identity signal of the motion recognition device, determines that the identity signal is correct, and completes the link between the receiving device and the motion recognition device.

The receiving device sets an action identification sampling frequency, and the action identification sampling frequency controls a time interval in which the receiving device captures the sensing signal.

The motion recognition device indicates the measured sensing signal by a right angle coordinate system, and the user's action decomposes a plurality of human body motion parameters according to the sensing signal.

The receiving device performs the human body motion calculation, and uses the human body sensing parameter as a data basis of the human body motion calculation.

The receiving device displays the action determination result.

The human motion calculation includes a first algorithm, and the human motion is determined by the first algorithm.

The receiving device successively performs a plurality of personal body motion determinations, and each of the human body motion determination outputs a corresponding motion recognition result.

Wherein, the human motion judgment is performed by measuring a plurality of the human motion parameters as a criterion for determining the human motion, and each of the human motions corresponds to a human motion parameter combination, wherein the human motion parameter combination includes each of the human body A plurality of the human motion parameters corresponding to the action.

Wherein, the human motion calculation includes a second algorithm, and the second algorithm calculates a eigenvalue of each of the human motion parameters and performs a neural network operation. This type of neural network operation outputs the motion recognition result.

The feature value is used to calculate the value of the human motion parameter in a time interval by a feature value algorithm.

The microprocessor, the sensing module, the wireless transmission module and the battery are all covered by an outer casing to prevent liquid from affecting its function.

Wherein, the monitoring body comprises a vibration motor module and a reservoir, the vibration motor module is controlled by the microprocessor to generate vibration, and the storage device stores an action maintaining threshold, according to the value of the percentage of the human body action, After the value of the percentage of the human body movement is greater than the threshold of the motion maintaining, a warning signal is output to the vibration motor, and the vibration motor module is controlled to start vibrating.

Among them, the microprocessor performs such neural network operations.

The receiving device sets the number of neurons in the neural network, the input layer hidden layer or the output layer.

The receiving device sets the weighting value and the offset weight in a random random number.

Wherein, the receiving device reads the neural network training sample.

The receiving device performs a correction calculation of the weighting value and the biasing value.

The receiving device updates the weighting value and the biasing value.

The receiving device determines whether the network converges.

The receiving device stores the weighting value and the biasing value.

The receiving device receives a global satellite signal, and the global satellite signal calculates a displacement distance of the user, wherein the global satellite signal includes information such as the longitude, latitude, altitude, and time of the user.

10‧‧‧Action identification device

11‧‧‧ collar

13‧‧‧Monitor ontologies

131‧‧‧Microprocessor

133‧‧‧Storage

134‧‧‧Vibration motor module

135‧‧‧Sensor module

1351‧‧‧Gyro

1352‧‧‧Three-axis acceleration sensing module

137‧‧‧Wireless transmission module

138‧‧‧Battery

151‧‧‧Connector

1511‧‧‧Contact surface

1512‧‧‧ positive terminal

1513‧‧‧ positive socket

1514‧‧‧negative contact

20‧‧‧Cloud database

30a, 30b‧‧‧ receiving devices

60‧‧‧Users

Figure 1 is a perspective view of a preferred embodiment of the present invention.

2 is a perspective view of a preferred embodiment of the present invention.

Figure 3 is a partial side elevational view of the preferred embodiment of the present invention.

Figure 4 is a partial side elevational view of the preferred embodiment of the present invention.

Figure 5 is a block diagram of a module of the preferred embodiment of the present invention.

Figure 6 is a three-axis vector diagram of a preferred embodiment of the present invention.

Figure 7 is a flow chart of the identification of the preferred embodiment of the present invention.

FIG. 8 is a flow chart of identification of the preferred embodiment of the present invention.

Figure 9 is a schematic view of the interface of the preferred embodiment of the present invention.

Figure 10 is a flow chart of the identification of the preferred embodiment of the present invention.

Figure 11 is a schematic view of a wave pattern of a preferred embodiment of the present invention.

Figure 12 is a perspective view of the human body action angle of the preferred embodiment of the present invention.

Figure 13 is a flow chart of the identification of the preferred embodiment of the present invention.

Figure 14 is a diagram showing the steps of a neural network calculation of a preferred embodiment of the present invention.

Figure 15 is a flow chart of the identification of the preferred embodiment of the present invention.

Figure 16 is a schematic view of a wave pattern of a preferred embodiment of the present invention.

Please refer to FIG. 1 to FIG. 6 , which is a preferred embodiment of the human body activity state recognition device, which includes a motion recognition device 10 and a cloud data base 20 and a receiving device that can be connected to the motion recognition device 10 . 30a, 30b, wherein the receiving device 30a, 30b can be a mobile communication device, a computer, or a network interface. The action The sensing device 10 senses a human body motion and outputs a sensing signal to the receiving device 30a, 30b, and performs a human body motion calculation by the receiving device 30a, 30b, or is stored in the cloud database 20 as the human body motion calculation. Based on the data, the result of the human motion calculation determines the current action of the user 60 and displays it on the receiving devices 30a, 30b.

The motion recognition device 10 includes a collar 11 and a monitoring body 13 and a coupling member respectively fixed to the surface of the collar 11. The collar 11 is annular and can be sleeved at a neck position of the human body. The monitoring body 13 includes a housing having waterproof performance, a microprocessor 131, and a storage unit 133 and a vibration motor respectively. The module 134, a sensing module 135, a wireless transmission module 137, and a battery 138.

The microprocessor 131, the sensing module 135, the wireless transmission module 137, and the battery 138 are all covered by the outer casing to prevent the device from being affected by liquid perspiration caused by human body perspiration or other factors. The memory 133 is a memory device for storing data for the microprocessor 131 to operate or access data; the vibration motor module 134 is controlled by the microprocessor 131 to generate vibration. The sensing module 135 senses the motion of the user 60 and outputs the sensing signal. The sensing signal is transmitted to the microprocessor 131. The sensing module 135 includes a gyroscope 1351 and a three-axis acceleration. The sensing module 1352 is controlled by the microprocessor 131 to measure a geographic or azimuth angle and output the sensing result to the microprocessor 131. The three-axis acceleration sensing module 1352 senses the movement. Speed and acceleration information are output to the microprocessor 131. The wireless transmission module 137 receives or transmits wireless signals to the receiving devices 30a, 30bb under the control of the microprocessor 131. The battery 138 provides the components of the embodiment required Electricity.

The coupling member is disposed on the collar 11 and includes two corresponding connectors 151. The contact faces 1511 of the two connectors 151 are respectively provided with a positive terminal 1512, a positive socket 1513 and each contact surface 1511. A pair of negative contacts 1514, wherein the positive terminal 1512 can be correspondingly inserted into the positive socket 1513, and the two negative contacts 1514 are correspondingly positioned, so that the two contact faces 1511 of the connector 151 are combined to generate electricity. The sexual connection enables the cervical vertebra monitoring device of the embodiment to produce a conduction effect, and can start normal operation. Further, the surface of the contact surface 1511 can be magnetically placed, so that the connector 151 can be quickly joined to form an electrical connection, reducing the time required for alignment or connection. In use, the cervical vertebra monitoring device is sleeved on the neck of the user 60 to sense the movement of the neck of the user 60.

Referring to FIGS. 6-9, the receiving device 30a, 30b performs the human motion calculation based on the sensing signal as a computing data basis, and the human motion calculation determines the motion of the user 60, and displays an action judgment result on the receiving. The devices 30a, 30b are shown, for example, in FIG.

This creation has the following steps:

Use step one, device link

The receiving devices 30a, 30b are linked to the motion recognition device 10. After the receiving devices 30a, 30b are activated, a call signal is sent to search for the motion recognition device 10, such as a wireless signal or a Bluetooth signal. After the receiving device 30a, 30b searches for the motion recognition device 10, the receiving device 30a, 30b verifies a body signal of the motion recognition device 10, determines whether the identity signal is correct, and after confirming that the identity signal is correct. The link between the receiving devices 30a, 30b and the motion recognition device 10 is completed.

Use step 2 to set the action to identify the sampling frequency.

A motion recognition sampling frequency is set in the receiving device 30a, 30b, and the motion recognition sampling frequency controls the receiving device 30a, 30b to capture a time interval Δt of the sensing signal.

Use step three to calculate human body motion parameters

As shown in FIG. 6, the motion recognition device 10 that is worn on the neck of the user 60 indicates the sensed signal measured by a right angle coordinate system. According to the sensing signal, the motion of the user 60 can be decomposed into a plurality of human body motion parameters including: an acceleration generated for the x-axis (t), acceleration Ÿ(t) generated in the y-axis, displacement Z(t) generated in the z-axis, and acceleration (t), roll angle α(t) and angular acceleration generated on the x-axis (t), forward or backward pitch angle γ(t) and angular acceleration generated on the z-axis (t) and yaw angle y (t) and angular acceleration for the y-axis (t).

Further, the human body sensing parameter includes a global satellite signal received by the receiving device 30a, 30ba, b to calculate a displacement distance (d) of the user 60, wherein the global satellite signal includes the longitude of the user 60. Information such as latitude, altitude and time. Alternatively, the displacement distance (d) may be calculated by an external device and transmitted to the receiving device 30a, 30b, for example, the external device may be a server host that receives the global satellite signal.

Use step four to perform the human motion calculation

The receiving device 30a, 30b performs the human body motion calculation, and the human body in step three The sensing parameters serve as the data base for the human motion calculation. The action judgment result of the calculation distinguishes a plurality of the human body actions, for example, the human body motion, sitting, standing, walking, running, jumping, or the like while lying down.

Use step five to display the result

As shown in FIG. 9, the receiving device 30a, 30b displays the action determination result to the user 60, and the display manner may be the maintenance time, the action frequency or the percentage of each of the human body actions, for example, the action judgment result is matched with a total sense. After the time is measured, the proportion of each of the human body actions in the total sensing time can be calculated.

In the implementation of the present invention, the calculation method of the maintenance time of each of the human body actions as a percentage of the total sensing time is to separately calculate the maintenance time of each of the human body actions, for example: TS ₁ = maintenance time of the inverted posture Σ (S ₁ × △ t); TS ₂ = maintenance time of the lying posture Σ (S ₂ × Δt); TS ₃ = maintenance time of the sitting posture estimation Σ (S ₃ × Δt); TS ₄ = maintenance time of the station attitude estimation Σ (S ₄ × Δt); TS ₅ = maintenance time of the attitude estimation Σ (S ₅ × Δt); TS ₆ = maintenance time of the running attitude estimation S (S ₆ × Δt); TS ₇ = maintenance of the hop attitude estimation Time Σ (S ₇ × Δt); or TS ₈ = maintenance time Σ (~ × Δt) of other posture estimation; wherein, S ₁ ~ S ₈ respectively represent an action recognition result of the human motion calculation to determine the human motion ; Δt is a time interval for the action recognition sampling frequency to control the receiving device 30a, 30b to capture the sensing signal; the total sensing time is the sum of TS ₁ ~ TS ₈ .

After calculating the maintenance time of each of the human body actions, the maintenance time of the single body motion is divided by the total sensing time, and the maintenance time of each of the human body actions is calculated as a percentage of the total sensing time.

In the new embodiment, the human motion calculation includes a first algorithm and a second algorithm, and the human motion is determined by the first algorithm and the second algorithm. Referring to FIG. 10 to FIG. 12, in the first algorithm, the receiving device 30a, 30b successively performs a plurality of personal motion determinations according to the human body sensing parameter in the time interval Δt, and each of the human body motion determination outputs Corresponding to the motion recognition result (S ₁ to S ₈ ), for example, if the value of the motion recognition result of the human body operating in the "reverse posture" is "1", it indicates that the human body motion is determined as "reverse posture". If the value of the motion recognition result is “0”, it indicates that the human body motion is determined not to be “inverted posture”, and the first algorithm continues to perform other human body motion determination. In the present embodiment, the human motion determination includes "reverse posture", "lying posture", "sitting posture", "station posture", "walking posture", "running posture", "jumping posture", and "other postures". "." For example, if the value of the action recognition result of the plurality of human body motion determinations is “0”, the human body motion is determined as “other posture”.

As shown in FIG. 11 and FIG. 12, in the first algorithm, the human motion judgment is performed by measuring a plurality of the human motion parameters as a reference for determining the human motion, for example, sampling with a certain human motion parameter. The highest value (peak _(max) ) or the lowest value (peak _(min) ) that is expressed in this interval. In the new embodiment, the first algorithm defines a combination of a human motion parameter corresponding to each human motion, and the human motion parameter combination includes the human motion parameter corresponding to each human motion, as shown in the following table. :

Wherein, the human body action parameter is:

(2) ab=a+b; abc=a+b+c

In the new embodiment, "a" is a "station pose" defined in the x-z plane. The range of the moving angle of the human body; "b" is the range of the moving angle of the human body defining the "sitting posture" in the xz plane; "c" is the range of the moving angle of the human body defining the "lying posture" in the xz plane; "d" The range of moving angles of the human body that defines the "inverted posture" in the xz plane.

(3) t _i (i=0,1,2,...,n) time series of the time interval

(4) Gravity acceleration

₍₅₎ peak _(max) Maximum Wave peak, peak _(min) minimum Wave value

Please refer to FIGS. 13 to 16, in which the second algorithm, a neural network calculation performed through the calculation of a feature value of each parameter of the human action, the action recognition result output s ₁ after the operation of such neural network ~s ₁ .

As shown in FIG. 16, in the new embodiment, the feature value is used to calculate the value of the human motion parameter in a time interval by using a feature value algorithm, and the feature value algorithm is:

As shown in FIGS. 14 and 15, the neural algorithm has the following steps: a neural-like algorithm step 1. The receiving device 30a, 30b sets the number of neurons in the neural network, the input layer hidden layer or the output layer; The second step of the calculation is to set the weighted value and the partial weight value by the uniform random random number; the third-like numerometric calculation step, the reading-like neural network training sample; the neural-like calculus step four, the correction calculation of the weighting value and the partial weight value; Step 5 of the neuro-calculus calculation, update the weighted value and the partial weight value; Step 6 of the neuro-calculus calculation, determine whether the network converges, and if the judgment result is "Yes" Then enter the nerve-like calculation step VII; otherwise, return to the neuro-calculus step 3; the neuro-calculus step VII, store the weighted value and the partial weight; the neuro-calculus step VIII, determine whether to re-network training, if the judgment result is "Yes Then complete the neurological algorithm; instead, return to the second neurological calculation step 2.

Further, the memory 133 stores an action maintaining threshold, and according to the value of the percentage of the human body action, after a certain value of the human body motion percentage is greater than the action maintaining threshold, an alarm signal is output to the vibration motor to control the vibration. The motor module 134 begins to vibrate to alert the user 60 that the action should be changed.

As can be seen from the above, the present invention has the following advantages:

1. The action recognition device can accurately determine the user's motion, and recognize the human body motion according to the human motion calculation.

2. Record each of the human body actions in detail, and record the proportion of the human body action in the total sensing time according to the action time.

3. The motion recognition device is connected to the receiver device, so that the user can conveniently read the result of the human body motion calculation.

10‧‧‧Action identification device

11‧‧‧ collar

13‧‧‧Monitor ontologies

151‧‧‧Connector

Claims (10)

A human activity state recognizing device includes a motion recognizing device and a receiving device, the motion recognizing device comprising a collar and a monitoring body and a joint member respectively fixed on the surface of the collar, the receiving device receiving the motion recognizing device The output of the human body motion parameter, wherein the monitoring body comprises a microprocessor and a sensing module, the sensing module senses a human body motion and outputs a sensing signal to the microprocessor, the microprocessor The human body motion parameter is calculated by the sensing signal; the receiving device performs a human body motion calculation according to the human motion parameter calculation, and the human motion calculation determines a user's motion, and displays an action judgment result on the receiving device; The bonding component comprises two corresponding connectors. The contact faces of the two connectors are respectively provided with a positive terminal, a positive socket and a pair of negative contacts on each contact surface, and the positive terminal is correspondingly inserted into the positive electrode. In the socket, the two positions correspond to the position of the negative electrode, so that the contact faces of the two connectors are electrically connected when correspondingly combined. According to the human activity state recognition device of claim 1, the surface of the contact surface is magnetic, so that the connector is quickly combined to form an electrical connection. For example, in the human activity state recognition device of claim 2, the sensing module includes a gyroscope and a three-axis acceleration sensing module, and the gyroscope measures a geographic or azimuth angle and outputs the sensing result to The microprocessor, the three-axis acceleration sensing module senses moving speed and acceleration information, and outputs the same to the microprocessor. Such as the human activity state recognition device of claim 3, the supervision The measurement body includes a wireless transmission module, and the wireless transmission module receives or transmits a wireless signal to the receiving device, wherein: the receiving device is linked to the motion recognition device, and the receiving device starts to send a call signal to search for the motion recognition. The receiving device verifies a body signal of the motion recognition device, determines that the identity signal is correct, and completes the link between the receiving device and the motion recognition device; and sets a motion recognition sampling frequency at the receiving device, the motion recognition sampling The frequency control unit captures a time interval of the sensing signal; the motion recognition device indicates the measured sensing signal by a right angle coordinate system, and the user's motion decomposes the plurality of human body actions according to the sensing signal The receiving device performs the human body motion calculation, and uses the human body sensing parameter as a data basis of the human body motion calculation; and the receiving device displays the motion determination result. For example, in the human activity state recognition device of claim 4, the human motion calculation includes a first algorithm, and the human motion is determined by the first algorithm, wherein: the receiving device successively performs a plurality of personal motion judgments, each The human motion judgment outputs the corresponding motion recognition result; and the human motion judgment is performed by measuring the plurality of the human motion parameters as the judgment criterion of the human motion, and each of the human motion corresponds to a human motion parameter In combination, the human motion parameter combination includes a plurality of the human motion parameters corresponding to each of the human body actions. For example, in the human activity state recognition device of claim 4, the human motion calculation includes a second algorithm, and the second algorithm calculates a characteristic value of each of the human motion parameters and performs a neural network operation. The neural network operation outputs the motion recognition result, wherein: the feature value calculates a value of the human motion parameter in the time interval by a feature value algorithm. For example, in the human activity state recognition device of claim 5 or 6, the microprocessor, the sensing module, the wireless transmission module and the battery are all covered by an outer casing to prevent liquid from affecting its function. The human body activity state recognition device of claim 7, wherein the monitoring body comprises a vibration motor module and a storage device, the vibration motor module receives vibration controlled by the microprocessor, and an action is stored in the storage device. The threshold is maintained, and according to the value of the percentage of the human body movement, after a value of the percentage of the human body movement is greater than the threshold of the motion maintenance, a warning signal is output to the vibration motor, and the vibration motor module is controlled to start vibrating. The human activity state recognizing device of claim 8, wherein the microprocessor performs the neural network operation, wherein: the receiving device sets a number of neurons of a neural network, an input layer hidden layer or an output layer; The receiving device sets the weighting value and the partial weight value by uniformly distributing random numbers; the receiving device reads the neural network training sample; and the receiving device performs the correction calculation of the weighting value and the partial weight value; The receiving device updates the weighting value and the biasing value; the receiving device determines whether the network converges; and the receiving device stores the weighting value and the biasing value. The human activity state recognition device of claim 9, wherein the receiving device receives a global satellite signal, and the global satellite signal calculates a displacement distance of the user, wherein the global satellite signal includes the longitude of the user, Information such as latitude, altitude and time.